cephi/ampere_research
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

updates + as-caida

Browse Source
This commit is contained in:
cephi 2024-12-15 15:04:09 -05:00
parent ecd4b403a4
commit dd6e050d6c
508 changed files with 5060 additions and 14102 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
analysis/data_stat.py
View File

@ -19,7 +19,6 @@ class Stat(Enum):
MATRIX_ROWS = 'matrix rows'
MATRIX_SIZE = 'matrix size'
MATRIX_NNZ = 'matrix nnz'
MATRIX_DENSITY_GROUP = 'matrix density group'
MATRIX_DENSITY = 'matrix density'
#POWER_BEFORE = 'power before'

11
pytorch/batch.py
View File

@ -132,15 +132,20 @@ for i, parameter in parameter_list:
synthetic_size = args.synthetic_size
synthetic_density = args.synthetic_density
output_filename_list = [
args.cpu.name.lower(),
output_filename_list = [args.cpu.name.lower()]
if args.cores is not None:
output_filename_list += [str(args.cores)]
else:
output_filename_list += ['max']
output_filename_list += [
args.format.name.lower(),
str(args.min_time_s),
str(args.baseline_time_s),
str(args.baseline_delay_s)]
if args.matrix_type == MatrixType.SUITESPARSE:
output_filename_list += [os.path.splitext(os.path.basename(parameter))[0]]
elif args.matrix_type == MatrixType.SYNTHETIC:
output_filename_list += [str(parameter[0]), str(parameter[1])]
output_filename_list += ['synthetic', str(parameter[0]), str(parameter[1])]
output_filename = '_'.join(output_filename_list)

1
pytorch/output_1core_after_test/altra_10_10_10_100000_0.0001.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 4372, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 999958, "MATRIX_DENSITY": 9.99958e-05, "TIME_S": 10.330792903900146, "TIME_S_1KI": 2.362944397049439, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 381.8866044712067, "W": 35.91605080170059, "J_1KI": 87.34826268783318, "W_1KI": 8.215016194350547, "W_D": 17.618050801700594, "J_D": 187.32843527841572, "W_D_1KI": 4.029746294990987, "J_D_1KI": 0.9217169018735102}

68
pytorch/output_1core_after_test/altra_10_10_10_100000_0.0001.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 100000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 999945, "MATRIX_DENSITY": 9.99945e-05, "TIME_S": 2.401212692260742}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 6, 18, ..., 999923, 999933,
999945]),
col_indices=tensor([ 2985, 7299, 36484, ..., 77100, 85631, 92891]),
values=tensor([ 0.2415, 0.2506, -1.0512, ..., 0.5862, -1.2492,
-0.0903]), size=(100000, 100000), nnz=999945,
layout=torch.sparse_csr)
tensor([0.5691, 0.2840, 0.2992, ..., 0.3981, 0.5874, 0.9189])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 999945
Density: 9.99945e-05
Time: 2.401212692260742 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 4372 -ss 100000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 999958, "MATRIX_DENSITY": 9.99958e-05, "TIME_S": 10.330792903900146}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 10, 15, ..., 999940, 999947,
999958]),
col_indices=tensor([ 6364, 15058, 52155, ..., 41882, 75278, 93727]),
values=tensor([ 1.1379, 2.3847, 1.1576, ..., 0.9163, 0.7641,
-1.0168]), size=(100000, 100000), nnz=999958,
layout=torch.sparse_csr)
tensor([0.2064, 0.7933, 0.3767, ..., 0.4884, 0.5023, 0.3792])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 999958
Density: 9.99958e-05
Time: 10.330792903900146 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 10, 15, ..., 999940, 999947,
999958]),
col_indices=tensor([ 6364, 15058, 52155, ..., 41882, 75278, 93727]),
values=tensor([ 1.1379, 2.3847, 1.1576, ..., 0.9163, 0.7641,
-1.0168]), size=(100000, 100000), nnz=999958,
layout=torch.sparse_csr)
tensor([0.2064, 0.7933, 0.3767, ..., 0.4884, 0.5023, 0.3792])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 999958
Density: 9.99958e-05
Time: 10.330792903900146 seconds
[20.52, 20.36, 20.28, 20.32, 20.32, 20.28, 20.2, 20.44, 20.4, 20.44]
[20.44, 20.28, 20.8, 21.68, 24.08, 26.36, 29.2, 30.64, 32.6, 32.12, 32.12, 32.44, 32.6, 32.44]
10.632755994796753
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 4372, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 999958, 'MATRIX_DENSITY': 9.99958e-05, 'TIME_S': 10.330792903900146, 'TIME_S_1KI': 2.362944397049439, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 381.8866044712067, 'W': 35.91605080170059}
[20.52, 20.36, 20.28, 20.32, 20.32, 20.28, 20.2, 20.44, 20.4, 20.44, 20.28, 20.24, 19.96, 19.96, 19.96, 20.0, 20.48, 20.92, 20.88, 20.68]
365.96
18.298
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 4372, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 999958, 'MATRIX_DENSITY': 9.99958e-05, 'TIME_S': 10.330792903900146, 'TIME_S_1KI': 2.362944397049439, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 381.8866044712067, 'W': 35.91605080170059, 'J_1KI': 87.34826268783318, 'W_1KI': 8.215016194350547, 'W_D': 17.618050801700594, 'J_D': 187.32843527841572, 'W_D_1KI': 4.029746294990987, 'J_D_1KI': 0.9217169018735102}

1
pytorch/output_1core_after_test/altra_10_10_10_100000_1e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 13545, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 99998, "MATRIX_DENSITY": 9.9998e-06, "TIME_S": 10.47447156906128, "TIME_S_1KI": 0.7733090859402938, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 345.2238781929017, "W": 32.996623304417554, "J_1KI": 25.487181852558265, "W_1KI": 2.4360740719392804, "W_D": 14.60462330441755, "J_D": 152.79941375160223, "W_D_1KI": 1.078229848978778, "J_D_1KI": 0.07960353259348675}

66
pytorch/output_1core_after_test/altra_10_10_10_100000_1e-05.output
View File

@ -1,66 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 100000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.7751424312591553}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 2, ..., 99996, 99999,
100000]),
col_indices=tensor([44500, 49971, 56483, ..., 66134, 68074, 1637]),
values=tensor([ 1.5203, 1.7392, 0.4724, ..., 0.1484, -0.5457,
0.0441]), size=(100000, 100000), nnz=100000,
layout=torch.sparse_csr)
tensor([0.4796, 0.4726, 0.4035, ..., 0.0030, 0.1184, 0.1782])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 0.7751424312591553 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 13545 -ss 100000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 99998, "MATRIX_DENSITY": 9.9998e-06, "TIME_S": 10.47447156906128}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 99996, 99997, 99998]),
col_indices=tensor([91921, 45205, 73439, ..., 30117, 55221, 9400]),
values=tensor([ 1.2826, -0.8828, -0.6837, ..., -2.0824, -1.6052,
1.5294]), size=(100000, 100000), nnz=99998,
layout=torch.sparse_csr)
tensor([0.0960, 0.3139, 0.1449, ..., 0.1558, 0.0708, 0.3546])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 99998
Density: 9.9998e-06
Time: 10.47447156906128 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 99996, 99997, 99998]),
col_indices=tensor([91921, 45205, 73439, ..., 30117, 55221, 9400]),
values=tensor([ 1.2826, -0.8828, -0.6837, ..., -2.0824, -1.6052,
1.5294]), size=(100000, 100000), nnz=99998,
layout=torch.sparse_csr)
tensor([0.0960, 0.3139, 0.1449, ..., 0.1558, 0.0708, 0.3546])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 99998
Density: 9.9998e-06
Time: 10.47447156906128 seconds
[20.76, 20.72, 20.56, 20.52, 20.12, 19.92, 19.92, 20.0, 20.12, 19.92]
[20.24, 20.32, 20.68, 22.96, 25.36, 27.6, 30.2, 31.48, 31.68, 31.76, 31.64, 31.32, 31.28]
10.462400197982788
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 13545, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 99998, 'MATRIX_DENSITY': 9.9998e-06, 'TIME_S': 10.47447156906128, 'TIME_S_1KI': 0.7733090859402938, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 345.2238781929017, 'W': 32.996623304417554}
[20.76, 20.72, 20.56, 20.52, 20.12, 19.92, 19.92, 20.0, 20.12, 19.92, 20.4, 20.52, 20.52, 20.84, 21.04, 20.76, 20.72, 20.56, 20.28, 20.36]
367.84000000000003
18.392000000000003
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 13545, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 99998, 'MATRIX_DENSITY': 9.9998e-06, 'TIME_S': 10.47447156906128, 'TIME_S_1KI': 0.7733090859402938, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 345.2238781929017, 'W': 32.996623304417554, 'J_1KI': 25.487181852558265, 'W_1KI': 2.4360740719392804, 'W_D': 14.60462330441755, 'J_D': 152.79941375160223, 'W_D_1KI': 1.078229848978778, 'J_D_1KI': 0.07960353259348675}

1
pytorch/output_1core_after_test/altra_10_10_10_100000_2e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 10495, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 200000, "MATRIX_DENSITY": 2e-05, "TIME_S": 10.752148389816284, "TIME_S_1KI": 1.0245019904541481, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 398.74920181274416, "W": 36.55315226044744, "J_1KI": 37.9942069378508, "W_1KI": 3.482911125340394, "W_D": 18.25915226044744, "J_D": 199.18452826595308, "W_D_1KI": 1.739795355926388, "J_D_1KI": 0.16577373567664488}

68
pytorch/output_1core_after_test/altra_10_10_10_100000_2e-05.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 100000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 199996, "MATRIX_DENSITY": 1.99996e-05, "TIME_S": 1.0004651546478271}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 5, ..., 199988, 199992,
199996]),
col_indices=tensor([39669, 76958, 86447, ..., 67341, 83508, 90452]),
values=tensor([-1.3977, 1.0356, -0.5900, ..., 0.8207, -1.1645,
0.3989]), size=(100000, 100000), nnz=199996,
layout=torch.sparse_csr)
tensor([0.9430, 0.8048, 0.8924, ..., 0.6826, 0.2927, 0.5723])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 199996
Density: 1.99996e-05
Time: 1.0004651546478271 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 10495 -ss 100000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 200000, "MATRIX_DENSITY": 2e-05, "TIME_S": 10.752148389816284}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 4, ..., 199995, 199997,
200000]),
col_indices=tensor([33783, 37586, 62221, ..., 65115, 69602, 99771]),
values=tensor([ 0.3562, 0.3055, 0.5875, ..., -0.0843, 2.8119,
0.1610]), size=(100000, 100000), nnz=200000,
layout=torch.sparse_csr)
tensor([0.8561, 0.5376, 0.4377, ..., 0.1840, 0.7093, 0.8920])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 200000
Density: 2e-05
Time: 10.752148389816284 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 4, ..., 199995, 199997,
200000]),
col_indices=tensor([33783, 37586, 62221, ..., 65115, 69602, 99771]),
values=tensor([ 0.3562, 0.3055, 0.5875, ..., -0.0843, 2.8119,
0.1610]), size=(100000, 100000), nnz=200000,
layout=torch.sparse_csr)
tensor([0.8561, 0.5376, 0.4377, ..., 0.1840, 0.7093, 0.8920])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 200000
Density: 2e-05
Time: 10.752148389816284 seconds
[20.0, 20.12, 20.0, 19.88, 20.2, 20.36, 20.6, 20.76, 20.64, 20.44]
[20.4, 20.48, 23.72, 25.72, 25.72, 27.88, 30.08, 32.52, 30.44, 31.72, 31.72, 32.04, 31.88, 31.68]
10.908750057220459
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 10495, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 200000, 'MATRIX_DENSITY': 2e-05, 'TIME_S': 10.752148389816284, 'TIME_S_1KI': 1.0245019904541481, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 398.74920181274416, 'W': 36.55315226044744}
[20.0, 20.12, 20.0, 19.88, 20.2, 20.36, 20.6, 20.76, 20.64, 20.44, 21.0, 20.64, 20.4, 20.16, 20.2, 20.2, 20.2, 20.32, 20.32, 20.32]
365.88
18.294
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 10495, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 200000, 'MATRIX_DENSITY': 2e-05, 'TIME_S': 10.752148389816284, 'TIME_S_1KI': 1.0245019904541481, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 398.74920181274416, 'W': 36.55315226044744, 'J_1KI': 37.9942069378508, 'W_1KI': 3.482911125340394, 'W_D': 18.25915226044744, 'J_D': 199.18452826595308, 'W_D_1KI': 1.739795355926388, 'J_D_1KI': 0.16577373567664488}

1
pytorch/output_1core_after_test/altra_10_10_10_100000_5e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 6654, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 499994, "MATRIX_DENSITY": 4.99994e-05, "TIME_S": 10.683140993118286, "TIME_S_1KI": 1.6055216400839023, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 366.1741095733642, "W": 37.080439049513444, "J_1KI": 55.03067471796877, "W_1KI": 5.572653899836706, "W_D": 18.798439049513444, "J_D": 185.63700583839412, "W_D_1KI": 2.825133611288465, "J_D_1KI": 0.4245767374945093}

68
pytorch/output_1core_after_test/altra_10_10_10_100000_5e-05.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 100000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 499988, "MATRIX_DENSITY": 4.99988e-05, "TIME_S": 1.5778212547302246}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 10, ..., 499983, 499985,
499988]),
col_indices=tensor([ 1371, 28809, 70668, ..., 42030, 54936, 56770]),
values=tensor([ 0.4333, 0.4225, 0.5901, ..., -0.2567, 0.8071,
-0.4001]), size=(100000, 100000), nnz=499988,
layout=torch.sparse_csr)
tensor([0.8725, 0.4955, 0.3045, ..., 0.0592, 0.4078, 0.2144])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 499988
Density: 4.99988e-05
Time: 1.5778212547302246 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 6654 -ss 100000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 499994, "MATRIX_DENSITY": 4.99994e-05, "TIME_S": 10.683140993118286}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 5, 10, ..., 499983, 499991,
499994]),
col_indices=tensor([11812, 14754, 24587, ..., 4989, 19562, 26481]),
values=tensor([-0.5266, -0.2099, 0.6678, ..., -1.2539, -0.8739,
-0.3506]), size=(100000, 100000), nnz=499994,
layout=torch.sparse_csr)
tensor([0.7326, 0.9445, 0.7161, ..., 0.6054, 0.1400, 0.0492])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 499994
Density: 4.99994e-05
Time: 10.683140993118286 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 5, 10, ..., 499983, 499991,
499994]),
col_indices=tensor([11812, 14754, 24587, ..., 4989, 19562, 26481]),
values=tensor([-0.5266, -0.2099, 0.6678, ..., -1.2539, -0.8739,
-0.3506]), size=(100000, 100000), nnz=499994,
layout=torch.sparse_csr)
tensor([0.7326, 0.9445, 0.7161, ..., 0.6054, 0.1400, 0.0492])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 499994
Density: 4.99994e-05
Time: 10.683140993118286 seconds
[20.12, 20.4, 20.36, 20.4, 20.52, 20.28, 20.08, 20.16, 20.52, 20.56]
[20.64, 20.68, 20.68, 23.88, 25.56, 28.76, 31.2, 33.8, 31.48, 32.0, 31.92, 32.0, 31.92]
9.875128746032715
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 6654, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 499994, 'MATRIX_DENSITY': 4.99994e-05, 'TIME_S': 10.683140993118286, 'TIME_S_1KI': 1.6055216400839023, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 366.1741095733642, 'W': 37.080439049513444}
[20.12, 20.4, 20.36, 20.4, 20.52, 20.28, 20.08, 20.16, 20.52, 20.56, 20.16, 20.12, 20.08, 20.08, 20.36, 20.4, 20.56, 20.56, 20.28, 20.12]
365.64
18.282
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 6654, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 499994, 'MATRIX_DENSITY': 4.99994e-05, 'TIME_S': 10.683140993118286, 'TIME_S_1KI': 1.6055216400839023, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 366.1741095733642, 'W': 37.080439049513444, 'J_1KI': 55.03067471796877, 'W_1KI': 5.572653899836706, 'W_D': 18.798439049513444, 'J_D': 185.63700583839412, 'W_D_1KI': 2.825133611288465, 'J_D_1KI': 0.4245767374945093}

1
pytorch/output_1core_after_test/altra_10_10_10_100000_8e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 5005, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 799969, "MATRIX_DENSITY": 7.99969e-05, "TIME_S": 10.528297424316406, "TIME_S_1KI": 2.103555928934347, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 353.08805488586427, "W": 33.32030349996928, "J_1KI": 70.5470639132596, "W_1KI": 6.657403296697158, "W_D": 14.945303499969278, "J_D": 158.3721511566639, "W_D_1KI": 2.986074625368487, "J_D_1KI": 0.5966183067669305}

68
pytorch/output_1core_after_test/altra_10_10_10_100000_8e-05.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 100000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 799967, "MATRIX_DENSITY": 7.99967e-05, "TIME_S": 2.0975828170776367}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 7, 13, ..., 799955, 799960,
799967]),
col_indices=tensor([ 6629, 8372, 29001, ..., 55409, 75475, 87705]),
values=tensor([ 0.3983, 0.9368, 0.8306, ..., -2.2845, 0.6609,
-0.9219]), size=(100000, 100000), nnz=799967,
layout=torch.sparse_csr)
tensor([0.4006, 0.2457, 0.6854, ..., 0.9449, 0.7766, 0.5729])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 799967
Density: 7.99967e-05
Time: 2.0975828170776367 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 5005 -ss 100000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 799969, "MATRIX_DENSITY": 7.99969e-05, "TIME_S": 10.528297424316406}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 7, 14, ..., 799947, 799960,
799969]),
col_indices=tensor([13025, 14589, 27413, ..., 85258, 89285, 92694]),
values=tensor([-0.0092, 0.8106, 0.5188, ..., -1.1562, 0.5281,
0.2289]), size=(100000, 100000), nnz=799969,
layout=torch.sparse_csr)
tensor([0.8836, 0.3169, 0.9227, ..., 0.6017, 0.3480, 0.8748])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 799969
Density: 7.99969e-05
Time: 10.528297424316406 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 7, 14, ..., 799947, 799960,
799969]),
col_indices=tensor([13025, 14589, 27413, ..., 85258, 89285, 92694]),
values=tensor([-0.0092, 0.8106, 0.5188, ..., -1.1562, 0.5281,
0.2289]), size=(100000, 100000), nnz=799969,
layout=torch.sparse_csr)
tensor([0.8836, 0.3169, 0.9227, ..., 0.6017, 0.3480, 0.8748])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 799969
Density: 7.99969e-05
Time: 10.528297424316406 seconds
[20.08, 20.04, 19.96, 20.04, 20.04, 20.24, 20.48, 20.56, 20.48, 20.36]
[20.0, 20.12, 20.84, 22.48, 25.32, 27.88, 30.36, 31.4, 32.84, 32.12, 32.12, 32.16, 32.32]
10.596783876419067
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 5005, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 799969, 'MATRIX_DENSITY': 7.99969e-05, 'TIME_S': 10.528297424316406, 'TIME_S_1KI': 2.103555928934347, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 353.08805488586427, 'W': 33.32030349996928}
[20.08, 20.04, 19.96, 20.04, 20.04, 20.24, 20.48, 20.56, 20.48, 20.36, 20.48, 20.6, 20.52, 20.28, 20.28, 20.44, 20.68, 20.96, 20.92, 21.04]
367.5
18.375
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 5005, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 799969, 'MATRIX_DENSITY': 7.99969e-05, 'TIME_S': 10.528297424316406, 'TIME_S_1KI': 2.103555928934347, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 353.08805488586427, 'W': 33.32030349996928, 'J_1KI': 70.5470639132596, 'W_1KI': 6.657403296697158, 'W_D': 14.945303499969278, 'J_D': 158.3721511566639, 'W_D_1KI': 2.986074625368487, 'J_D_1KI': 0.5966183067669305}

1
pytorch/output_1core_after_test/altra_10_10_10_10000_0.0001.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 176497, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.582123756408691, "TIME_S_1KI": 0.05995639447927552, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 323.9779005908966, "W": 29.969251252049236, "J_1KI": 1.83560004187548, "W_1KI": 0.16980034364351368, "W_D": 11.822251252049234, "J_D": 127.80259702467916, "W_D_1KI": 0.06698273201272109, "J_D_1KI": 0.0003795120144405916}

85
pytorch/output_1core_after_test/altra_10_10_10_10000_0.0001.output
View File

@ -1,85 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 10000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 9999, "MATRIX_DENSITY": 9.999e-05, "TIME_S": 0.06440186500549316}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 1, ..., 9999, 9999, 9999]),
col_indices=tensor([5106, 3897, 3155, ..., 1583, 3431, 5555]),
values=tensor([ 1.3508, -1.1736, 0.4296, ..., -0.8458, 0.0925,
0.1832]), size=(10000, 10000), nnz=9999,
layout=torch.sparse_csr)
tensor([0.5935, 0.4331, 0.3309, ..., 0.4577, 0.4204, 0.6600])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 9999
Density: 9.999e-05
Time: 0.06440186500549316 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 163038 -ss 10000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 9999, "MATRIX_DENSITY": 9.999e-05, "TIME_S": 9.699290990829468}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 9998, 9998, 9999]),
col_indices=tensor([ 552, 5534, 9404, ..., 8672, 9099, 1672]),
values=tensor([-0.4570, 0.0714, -1.0309, ..., 0.9768, -0.9088,
0.5389]), size=(10000, 10000), nnz=9999,
layout=torch.sparse_csr)
tensor([0.9350, 0.7973, 0.4526, ..., 0.7485, 0.8481, 0.0598])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 9999
Density: 9.999e-05
Time: 9.699290990829468 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 176497 -ss 10000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.582123756408691}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 4, ..., 9996, 9997, 10000]),
col_indices=tensor([7714, 1870, 5845, ..., 759, 3572, 7308]),
values=tensor([-1.0266, 0.5680, -0.8233, ..., -0.9435, -0.5643,
1.5314]), size=(10000, 10000), nnz=10000,
layout=torch.sparse_csr)
tensor([0.6158, 0.7644, 0.3713, ..., 0.4226, 0.3057, 0.7915])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 10.582123756408691 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 4, ..., 9996, 9997, 10000]),
col_indices=tensor([7714, 1870, 5845, ..., 759, 3572, 7308]),
values=tensor([-1.0266, 0.5680, -0.8233, ..., -0.9435, -0.5643,
1.5314]), size=(10000, 10000), nnz=10000,
layout=torch.sparse_csr)
tensor([0.6158, 0.7644, 0.3713, ..., 0.4226, 0.3057, 0.7915])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 10.582123756408691 seconds
[20.32, 20.24, 20.16, 20.2, 19.92, 19.96, 20.48, 20.36, 20.36, 20.4]
[20.48, 20.52, 20.28, 24.08, 24.84, 26.56, 26.92, 24.56, 24.12, 22.96, 23.08, 22.88, 22.8, 23.0]
10.810343503952026
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 176497, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.582123756408691, 'TIME_S_1KI': 0.05995639447927552, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 323.9779005908966, 'W': 29.969251252049236}
[20.32, 20.24, 20.16, 20.2, 19.92, 19.96, 20.48, 20.36, 20.36, 20.4, 20.24, 20.16, 19.92, 19.76, 19.76, 20.0, 20.12, 20.48, 20.44, 20.28]
362.94000000000005
18.147000000000002
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 176497, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.582123756408691, 'TIME_S_1KI': 0.05995639447927552, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 323.9779005908966, 'W': 29.969251252049236, 'J_1KI': 1.83560004187548, 'W_1KI': 0.16980034364351368, 'W_D': 11.822251252049234, 'J_D': 127.80259702467916, 'W_D_1KI': 0.06698273201272109, 'J_D_1KI': 0.0003795120144405916}

1
pytorch/output_1core_after_test/altra_10_10_10_10000_1e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 424922, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.511717319488525, "TIME_S_1KI": 0.024737992665685764, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 302.01861512184144, "W": 28.37936763657448, "J_1KI": 0.7107624814009194, "W_1KI": 0.0667872400971813, "W_D": 10.020367636574477, "J_D": 106.6386536643505, "W_D_1KI": 0.023581663544308077, "J_D_1KI": 5.549645239434079e-05}

1521
pytorch/output_1core_after_test/altra_10_10_10_10000_1e-05.output
View File

File diff suppressed because it is too large Load Diff

1
pytorch/output_1core_after_test/altra_10_10_10_10000_2e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 362139, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 2000, "MATRIX_DENSITY": 2e-05, "TIME_S": 10.370163679122925, "TIME_S_1KI": 0.028635865452555302, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 306.58433523178104, "W": 28.67356603561499, "J_1KI": 0.8465929801313337, "W_1KI": 0.07917834322073841, "W_D": 10.23556603561499, "J_D": 109.4410163302422, "W_D_1KI": 0.0282641914723766, "J_D_1KI": 7.804790832353488e-05}

85
pytorch/output_1core_after_test/altra_10_10_10_10000_2e-05.output
View File

@ -1,85 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 10000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 2000, "MATRIX_DENSITY": 2e-05, "TIME_S": 0.032781124114990234}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 2000, 2000, 2000]),
col_indices=tensor([6871, 3733, 5965, ..., 5141, 3011, 301]),
values=tensor([-0.4304, -1.2708, -0.2632, ..., 0.7868, 3.1604,
1.2152]), size=(10000, 10000), nnz=2000,
layout=torch.sparse_csr)
tensor([0.4678, 0.6370, 0.2747, ..., 0.6866, 0.0324, 0.0834])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 2000
Density: 2e-05
Time: 0.032781124114990234 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 320306 -ss 10000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 2000, "MATRIX_DENSITY": 2e-05, "TIME_S": 9.287068843841553}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 1, ..., 1999, 1999, 2000]),
col_indices=tensor([3694, 6091, 5443, ..., 5395, 436, 2041]),
values=tensor([-0.9064, 1.9159, -0.4201, ..., -1.3373, 0.3655,
-0.2885]), size=(10000, 10000), nnz=2000,
layout=torch.sparse_csr)
tensor([0.4015, 0.1742, 0.4171, ..., 0.9425, 0.5446, 0.5222])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 2000
Density: 2e-05
Time: 9.287068843841553 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 362139 -ss 10000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 2000, "MATRIX_DENSITY": 2e-05, "TIME_S": 10.370163679122925}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 2000, 2000, 2000]),
col_indices=tensor([7968, 3420, 6634, ..., 1670, 1798, 8896]),
values=tensor([ 0.3730, 1.3738, -1.4562, ..., -0.4679, -0.5220,
-3.1368]), size=(10000, 10000), nnz=2000,
layout=torch.sparse_csr)
tensor([0.3918, 0.4219, 0.1314, ..., 0.5461, 0.2473, 0.0750])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 2000
Density: 2e-05
Time: 10.370163679122925 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 2000, 2000, 2000]),
col_indices=tensor([7968, 3420, 6634, ..., 1670, 1798, 8896]),
values=tensor([ 0.3730, 1.3738, -1.4562, ..., -0.4679, -0.5220,
-3.1368]), size=(10000, 10000), nnz=2000,
layout=torch.sparse_csr)
tensor([0.3918, 0.4219, 0.1314, ..., 0.5461, 0.2473, 0.0750])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 2000
Density: 2e-05
Time: 10.370163679122925 seconds
[20.08, 20.28, 20.48, 20.6, 20.52, 20.52, 20.52, 20.48, 20.56, 20.56]
[20.68, 20.6, 20.64, 24.08, 25.96, 27.16, 27.88, 28.12, 24.28, 23.44, 23.32, 23.2, 23.12]
10.69222903251648
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 362139, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 2000, 'MATRIX_DENSITY': 2e-05, 'TIME_S': 10.370163679122925, 'TIME_S_1KI': 0.028635865452555302, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 306.58433523178104, 'W': 28.67356603561499}
[20.08, 20.28, 20.48, 20.6, 20.52, 20.52, 20.52, 20.48, 20.56, 20.56, 20.12, 20.04, 20.32, 20.64, 20.56, 20.68, 20.68, 20.6, 20.64, 20.52]
368.76
18.438
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 362139, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 2000, 'MATRIX_DENSITY': 2e-05, 'TIME_S': 10.370163679122925, 'TIME_S_1KI': 0.028635865452555302, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 306.58433523178104, 'W': 28.67356603561499, 'J_1KI': 0.8465929801313337, 'W_1KI': 0.07917834322073841, 'W_D': 10.23556603561499, 'J_D': 109.4410163302422, 'W_D_1KI': 0.0282641914723766, 'J_D_1KI': 7.804790832353488e-05}

1
pytorch/output_1core_after_test/altra_10_10_10_10000_5e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 236282, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000, "MATRIX_DENSITY": 5e-05, "TIME_S": 10.53740906715393, "TIME_S_1KI": 0.04459674908437346, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 324.57128691673273, "W": 29.735476961651, "J_1KI": 1.3736606551355275, "W_1KI": 0.1258474067497778, "W_D": 11.572476961650999, "J_D": 126.31691582083694, "W_D_1KI": 0.0489773954920434, "J_D_1KI": 0.00020728365043483382}

85
pytorch/output_1core_after_test/altra_10_10_10_10000_5e-05.output
View File

@ -1,85 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 10000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000, "MATRIX_DENSITY": 5e-05, "TIME_S": 0.04942727088928223}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 2, ..., 5000, 5000, 5000]),
col_indices=tensor([1168, 6226, 690, ..., 5217, 476, 6738]),
values=tensor([ 0.0787, 0.3273, -0.2779, ..., 1.4194, 0.8103,
1.1550]), size=(10000, 10000), nnz=5000,
layout=torch.sparse_csr)
tensor([0.0064, 0.2511, 0.1525, ..., 0.6286, 0.9165, 0.4351])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000
Density: 5e-05
Time: 0.04942727088928223 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 212433 -ss 10000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000, "MATRIX_DENSITY": 5e-05, "TIME_S": 9.440152168273926}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 4999, 4999, 5000]),
col_indices=tensor([9942, 1214, 2847, ..., 8254, 7960, 457]),
values=tensor([-0.9693, 1.4816, 0.2851, ..., 2.1213, -0.2351,
0.2580]), size=(10000, 10000), nnz=5000,
layout=torch.sparse_csr)
tensor([0.7203, 0.5491, 0.5318, ..., 0.5029, 0.8608, 0.3592])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000
Density: 5e-05
Time: 9.440152168273926 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 236282 -ss 10000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000, "MATRIX_DENSITY": 5e-05, "TIME_S": 10.53740906715393}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 3, ..., 5000, 5000, 5000]),
col_indices=tensor([5447, 7579, 9073, ..., 5210, 7678, 9855]),
values=tensor([-2.4915, -1.5336, 2.5123, ..., -0.4713, 0.8329,
-0.6699]), size=(10000, 10000), nnz=5000,
layout=torch.sparse_csr)
tensor([0.8488, 0.5259, 0.1601, ..., 0.9858, 0.5655, 0.9639])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000
Density: 5e-05
Time: 10.53740906715393 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 3, ..., 5000, 5000, 5000]),
col_indices=tensor([5447, 7579, 9073, ..., 5210, 7678, 9855]),
values=tensor([-2.4915, -1.5336, 2.5123, ..., -0.4713, 0.8329,
-0.6699]), size=(10000, 10000), nnz=5000,
layout=torch.sparse_csr)
tensor([0.8488, 0.5259, 0.1601, ..., 0.9858, 0.5655, 0.9639])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000
Density: 5e-05
Time: 10.53740906715393 seconds
[20.44, 20.28, 20.44, 20.48, 20.6, 20.4, 20.24, 20.2, 19.8, 19.92]
[19.92, 20.2, 21.36, 22.96, 22.96, 24.68, 25.4, 25.96, 24.48, 23.6, 23.48, 23.36, 23.28, 23.24]
10.915287733078003
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 236282, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000, 'MATRIX_DENSITY': 5e-05, 'TIME_S': 10.53740906715393, 'TIME_S_1KI': 0.04459674908437346, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 324.57128691673273, 'W': 29.735476961651}
[20.44, 20.28, 20.44, 20.48, 20.6, 20.4, 20.24, 20.2, 19.8, 19.92, 20.2, 20.0, 19.96, 20.04, 20.28, 20.16, 20.08, 20.0, 20.0, 20.04]
363.26
18.163
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 236282, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000, 'MATRIX_DENSITY': 5e-05, 'TIME_S': 10.53740906715393, 'TIME_S_1KI': 0.04459674908437346, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 324.57128691673273, 'W': 29.735476961651, 'J_1KI': 1.3736606551355275, 'W_1KI': 0.1258474067497778, 'W_D': 11.572476961650999, 'J_D': 126.31691582083694, 'W_D_1KI': 0.0489773954920434, 'J_D_1KI': 0.00020728365043483382}

1
pytorch/output_1core_after_test/altra_10_10_10_10000_8e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 185363, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 8000, "MATRIX_DENSITY": 8e-05, "TIME_S": 10.317452430725098, "TIME_S_1KI": 0.055660797628033096, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 282.4831217098236, "W": 27.119152879595266, "J_1KI": 1.5239455647018207, "W_1KI": 0.1463029454615822, "W_D": 8.652152879595263, "J_D": 90.12402289223665, "W_D_1KI": 0.046676806480232105, "J_D_1KI": 0.0002518129641850429}

84
pytorch/output_1core_after_test/altra_10_10_10_10000_8e-05.output
View File

@ -1,84 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 10000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 8000, "MATRIX_DENSITY": 8e-05, "TIME_S": 0.060555219650268555}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 2, ..., 7999, 8000, 8000]),
col_indices=tensor([9977, 5306, 1222, ..., 6555, 7712, 2915]),
values=tensor([0.7927, 2.3954, 0.9167, ..., 1.0032, 0.5486, 0.7967]),
size=(10000, 10000), nnz=8000, layout=torch.sparse_csr)
tensor([0.2944, 0.3792, 0.7257, ..., 0.8753, 0.2073, 0.0871])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 8000
Density: 8e-05
Time: 0.060555219650268555 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 173395 -ss 10000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 8000, "MATRIX_DENSITY": 8e-05, "TIME_S": 9.82205057144165}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 3, ..., 7999, 7999, 8000]),
col_indices=tensor([5642, 5987, 9672, ..., 9618, 963, 3909]),
values=tensor([ 2.0260, 0.0167, -1.0249, ..., -0.9431, 1.2350,
-0.3906]), size=(10000, 10000), nnz=8000,
layout=torch.sparse_csr)
tensor([0.6617, 0.5997, 0.7114, ..., 0.4730, 0.1362, 0.1168])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 8000
Density: 8e-05
Time: 9.82205057144165 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 185363 -ss 10000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 8000, "MATRIX_DENSITY": 8e-05, "TIME_S": 10.317452430725098}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 1, ..., 7999, 7999, 8000]),
col_indices=tensor([8903, 3321, 7408, ..., 5922, 9897, 4802]),
values=tensor([ 0.5381, -0.2046, 1.4195, ..., -1.2433, 1.3727,
0.7226]), size=(10000, 10000), nnz=8000,
layout=torch.sparse_csr)
tensor([0.3619, 0.5493, 0.6101, ..., 0.1612, 0.4763, 0.2515])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 8000
Density: 8e-05
Time: 10.317452430725098 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 1, ..., 7999, 7999, 8000]),
col_indices=tensor([8903, 3321, 7408, ..., 5922, 9897, 4802]),
values=tensor([ 0.5381, -0.2046, 1.4195, ..., -1.2433, 1.3727,
0.7226]), size=(10000, 10000), nnz=8000,
layout=torch.sparse_csr)
tensor([0.3619, 0.5493, 0.6101, ..., 0.1612, 0.4763, 0.2515])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 8000
Density: 8e-05
Time: 10.317452430725098 seconds
[20.48, 20.32, 20.48, 20.4, 20.4, 20.48, 20.36, 20.36, 20.2, 20.12]
[19.8, 19.92, 20.64, 20.64, 22.16, 23.8, 24.56, 24.84, 24.4, 24.04, 23.36, 23.0, 23.16]
10.416369676589966
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 185363, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 8000, 'MATRIX_DENSITY': 8e-05, 'TIME_S': 10.317452430725098, 'TIME_S_1KI': 0.055660797628033096, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 282.4831217098236, 'W': 27.119152879595266}
[20.48, 20.32, 20.48, 20.4, 20.4, 20.48, 20.36, 20.36, 20.2, 20.12, 20.36, 20.48, 20.72, 20.76, 20.72, 20.64, 20.68, 20.68, 20.68, 21.0]
369.34000000000003
18.467000000000002
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 185363, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 8000, 'MATRIX_DENSITY': 8e-05, 'TIME_S': 10.317452430725098, 'TIME_S_1KI': 0.055660797628033096, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 282.4831217098236, 'W': 27.119152879595266, 'J_1KI': 1.5239455647018207, 'W_1KI': 0.1463029454615822, 'W_D': 8.652152879595263, 'J_D': 90.12402289223665, 'W_D_1KI': 0.046676806480232105, 'J_D_1KI': 0.0002518129641850429}

1
pytorch/output_1core_after_test/altra_10_10_10_150000_0.0001.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 1834, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 2249895, "MATRIX_DENSITY": 9.999533333333333e-05, "TIME_S": 10.790888786315918, "TIME_S_1KI": 5.88379977443616, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 367.17931696891776, "W": 32.98158336912441, "J_1KI": 200.2068249557894, "W_1KI": 17.983415141289207, "W_D": 14.585583369124407, "J_D": 162.37924295902246, "W_D_1KI": 7.952880790144169, "J_D_1KI": 4.336358118944476}

71
pytorch/output_1core_after_test/altra_10_10_10_150000_0.0001.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 150000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 2249891, "MATRIX_DENSITY": 9.999515555555556e-05, "TIME_S": 5.724584579467773}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 16, 33, ..., 2249859,
2249877, 2249891]),
col_indices=tensor([ 15294, 19172, 20091, ..., 131171, 142636,
143029]),
values=tensor([ 1.3380, -0.3137, -0.5749, ..., 0.3744, 0.3646,
-2.0781]), size=(150000, 150000), nnz=2249891,
layout=torch.sparse_csr)
tensor([0.3273, 0.5078, 0.8205, ..., 0.0347, 0.2426, 0.7008])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 2249891
Density: 9.999515555555556e-05
Time: 5.724584579467773 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1834 -ss 150000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 2249895, "MATRIX_DENSITY": 9.999533333333333e-05, "TIME_S": 10.790888786315918}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 20, 31, ..., 2249870,
2249883, 2249895]),
col_indices=tensor([ 7356, 13304, 13563, ..., 98372, 126446,
139883]),
values=tensor([-1.7231, -0.1071, 2.0159, ..., -1.1470, -0.3672,
1.5446]), size=(150000, 150000), nnz=2249895,
layout=torch.sparse_csr)
tensor([0.4200, 0.0608, 0.0953, ..., 0.0975, 0.4627, 0.0936])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 2249895
Density: 9.999533333333333e-05
Time: 10.790888786315918 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 20, 31, ..., 2249870,
2249883, 2249895]),
col_indices=tensor([ 7356, 13304, 13563, ..., 98372, 126446,
139883]),
values=tensor([-1.7231, -0.1071, 2.0159, ..., -1.1470, -0.3672,
1.5446]), size=(150000, 150000), nnz=2249895,
layout=torch.sparse_csr)
tensor([0.4200, 0.0608, 0.0953, ..., 0.0975, 0.4627, 0.0936])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 2249895
Density: 9.999533333333333e-05
Time: 10.790888786315918 seconds
[19.96, 19.88, 20.12, 20.16, 20.32, 20.32, 20.32, 20.52, 20.6, 20.48]
[20.52, 20.44, 21.04, 22.2, 24.32, 26.88, 29.28, 31.0, 32.16, 32.32, 32.32, 32.24, 32.24, 32.36]
11.132858991622925
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1834, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 2249895, 'MATRIX_DENSITY': 9.999533333333333e-05, 'TIME_S': 10.790888786315918, 'TIME_S_1KI': 5.88379977443616, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 367.17931696891776, 'W': 32.98158336912441}
[19.96, 19.88, 20.12, 20.16, 20.32, 20.32, 20.32, 20.52, 20.6, 20.48, 20.56, 20.68, 20.56, 20.76, 20.92, 20.72, 20.64, 20.48, 20.32, 20.2]
367.92
18.396
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1834, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 2249895, 'MATRIX_DENSITY': 9.999533333333333e-05, 'TIME_S': 10.790888786315918, 'TIME_S_1KI': 5.88379977443616, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 367.17931696891776, 'W': 32.98158336912441, 'J_1KI': 200.2068249557894, 'W_1KI': 17.983415141289207, 'W_D': 14.585583369124407, 'J_D': 162.37924295902246, 'W_D_1KI': 7.952880790144169, 'J_D_1KI': 4.336358118944476}

1
pytorch/output_1core_after_test/altra_10_10_10_150000_1e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 7234, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 225000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.380627632141113, "TIME_S_1KI": 1.4349775548992416, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 386.428590927124, "W": 35.67289032159632, "J_1KI": 53.41838414806801, "W_1KI": 4.931281493170627, "W_D": 17.146890321596317, "J_D": 185.74465388202657, "W_D_1KI": 2.3703193698640193, "J_D_1KI": 0.32766372267957133}

69
pytorch/output_1core_after_test/altra_10_10_10_150000_1e-05.output
View File

@ -1,69 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 150000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 224999, "MATRIX_DENSITY": 9.999955555555555e-06, "TIME_S": 1.4513022899627686}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 7, ..., 224997, 224998,
224999]),
col_indices=tensor([ 1060, 41338, 58835, ..., 108277, 73571,
97514]),
values=tensor([-0.6349, 1.6645, 0.3729, ..., 0.0544, 1.2271,
0.2155]), size=(150000, 150000), nnz=224999,
layout=torch.sparse_csr)
tensor([0.9990, 0.4988, 0.7512, ..., 0.3421, 0.7043, 0.0135])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 224999
Density: 9.999955555555555e-06
Time: 1.4513022899627686 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 7234 -ss 150000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 225000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.380627632141113}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 2, ..., 225000, 225000,
225000]),
col_indices=tensor([97170, 89495, 1274, ..., 10485, 28306, 74671]),
values=tensor([-1.0399, 1.1242, -1.2641, ..., 2.0193, 0.6877,
-0.5401]), size=(150000, 150000), nnz=225000,
layout=torch.sparse_csr)
tensor([0.5775, 0.5693, 0.4176, ..., 0.7548, 0.5748, 0.0697])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 225000
Density: 1e-05
Time: 10.380627632141113 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 2, ..., 225000, 225000,
225000]),
col_indices=tensor([97170, 89495, 1274, ..., 10485, 28306, 74671]),
values=tensor([-1.0399, 1.1242, -1.2641, ..., 2.0193, 0.6877,
-0.5401]), size=(150000, 150000), nnz=225000,
layout=torch.sparse_csr)
tensor([0.5775, 0.5693, 0.4176, ..., 0.7548, 0.5748, 0.0697])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 225000
Density: 1e-05
Time: 10.380627632141113 seconds
[20.28, 20.4, 20.4, 20.44, 20.32, 20.76, 20.76, 20.8, 21.04, 20.88]
[20.4, 20.28, 22.2, 22.8, 25.44, 27.92, 27.92, 30.48, 30.52, 31.6, 31.28, 31.64, 31.72, 31.84]
10.832556247711182
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 7234, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 225000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.380627632141113, 'TIME_S_1KI': 1.4349775548992416, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 386.428590927124, 'W': 35.67289032159632}
[20.28, 20.4, 20.4, 20.44, 20.32, 20.76, 20.76, 20.8, 21.04, 20.88, 20.44, 20.36, 20.4, 20.24, 20.4, 20.72, 20.64, 20.6, 20.96, 20.96]
370.52000000000004
18.526000000000003
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 7234, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 225000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.380627632141113, 'TIME_S_1KI': 1.4349775548992416, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 386.428590927124, 'W': 35.67289032159632, 'J_1KI': 53.41838414806801, 'W_1KI': 4.931281493170627, 'W_D': 17.146890321596317, 'J_D': 185.74465388202657, 'W_D_1KI': 2.3703193698640193, 'J_D_1KI': 0.32766372267957133}

1
pytorch/output_1core_after_test/altra_10_10_10_150000_2e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 5267, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 449995, "MATRIX_DENSITY": 1.9999777777777777e-05, "TIME_S": 10.143731594085693, "TIME_S_1KI": 1.9259030936179407, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 345.50767498016353, "W": 33.64124287914862, "J_1KI": 65.59857128918996, "W_1KI": 6.387173510375664, "W_D": 15.246242879148621, "J_D": 156.5844029092788, "W_D_1KI": 2.894673035722161, "J_D_1KI": 0.5495866785118969}

71
pytorch/output_1core_after_test/altra_10_10_10_150000_2e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 150000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 449996, "MATRIX_DENSITY": 1.9999822222222222e-05, "TIME_S": 1.9931979179382324}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 9, ..., 449993, 449994,
449996]),
col_indices=tensor([ 22593, 92994, 310, ..., 102409, 47111,
69289]),
values=tensor([ 0.6471, 0.9609, 0.5622, ..., 2.1388, -1.1845,
0.1991]), size=(150000, 150000), nnz=449996,
layout=torch.sparse_csr)
tensor([0.7282, 0.9879, 0.2896, ..., 0.4436, 0.3832, 0.9789])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 449996
Density: 1.9999822222222222e-05
Time: 1.9931979179382324 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 5267 -ss 150000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 449995, "MATRIX_DENSITY": 1.9999777777777777e-05, "TIME_S": 10.143731594085693}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 5, ..., 449988, 449991,
449995]),
col_indices=tensor([ 72009, 57024, 70057, ..., 83430, 119068,
138373]),
values=tensor([-0.9929, 0.3427, 2.6993, ..., 1.4662, 0.4304,
0.3433]), size=(150000, 150000), nnz=449995,
layout=torch.sparse_csr)
tensor([0.8078, 0.3699, 0.1921, ..., 0.5384, 0.5304, 0.7616])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 449995
Density: 1.9999777777777777e-05
Time: 10.143731594085693 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 5, ..., 449988, 449991,
449995]),
col_indices=tensor([ 72009, 57024, 70057, ..., 83430, 119068,
138373]),
values=tensor([-0.9929, 0.3427, 2.6993, ..., 1.4662, 0.4304,
0.3433]), size=(150000, 150000), nnz=449995,
layout=torch.sparse_csr)
tensor([0.8078, 0.3699, 0.1921, ..., 0.5384, 0.5304, 0.7616])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 449995
Density: 1.9999777777777777e-05
Time: 10.143731594085693 seconds
[20.48, 20.48, 20.72, 20.64, 20.64, 20.48, 20.48, 20.32, 20.16, 20.4]
[20.36, 20.52, 23.76, 24.84, 27.28, 27.28, 29.96, 32.12, 31.04, 31.96, 31.24, 31.2, 31.32]
10.270359992980957
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 5267, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 449995, 'MATRIX_DENSITY': 1.9999777777777777e-05, 'TIME_S': 10.143731594085693, 'TIME_S_1KI': 1.9259030936179407, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 345.50767498016353, 'W': 33.64124287914862}
[20.48, 20.48, 20.72, 20.64, 20.64, 20.48, 20.48, 20.32, 20.16, 20.4, 20.24, 20.2, 20.16, 20.28, 20.24, 20.52, 20.6, 20.64, 20.52, 20.52]
367.9
18.395
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 5267, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 449995, 'MATRIX_DENSITY': 1.9999777777777777e-05, 'TIME_S': 10.143731594085693, 'TIME_S_1KI': 1.9259030936179407, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 345.50767498016353, 'W': 33.64124287914862, 'J_1KI': 65.59857128918996, 'W_1KI': 6.387173510375664, 'W_D': 15.246242879148621, 'J_D': 156.5844029092788, 'W_D_1KI': 2.894673035722161, 'J_D_1KI': 0.5495866785118969}

1
pytorch/output_1core_after_test/altra_10_10_10_150000_5e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 2982, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1124963, "MATRIX_DENSITY": 4.9998355555555557e-05, "TIME_S": 10.834063529968262, "TIME_S_1KI": 3.633153430572858, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 388.3960797214508, "W": 36.8830912456038, "J_1KI": 130.24684095286747, "W_1KI": 12.368575199733, "W_D": 18.6480912456038, "J_D": 196.3730611908436, "W_D_1KI": 6.253551725554594, "J_D_1KI": 2.0970998408969126}

71
pytorch/output_1core_after_test/altra_10_10_10_150000_5e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 150000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1124975, "MATRIX_DENSITY": 4.999888888888889e-05, "TIME_S": 3.5205483436584473}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 7, 18, ..., 1124957,
1124964, 1124975]),
col_indices=tensor([ 45673, 46869, 68642, ..., 93007, 132415,
145624]),
values=tensor([ 1.0589, -0.8292, -0.9400, ..., -0.5244, 0.2483,
-1.2673]), size=(150000, 150000), nnz=1124975,
layout=torch.sparse_csr)
tensor([0.2890, 0.6092, 0.8181, ..., 0.3578, 0.3655, 0.2203])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1124975
Density: 4.999888888888889e-05
Time: 3.5205483436584473 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 2982 -ss 150000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1124963, "MATRIX_DENSITY": 4.9998355555555557e-05, "TIME_S": 10.834063529968262}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 6, 16, ..., 1124949,
1124959, 1124963]),
col_indices=tensor([ 30949, 52207, 66032, ..., 93409, 116462,
142125]),
values=tensor([-1.0709, 0.6351, 0.9891, ..., -0.8011, -0.8370,
-1.9774]), size=(150000, 150000), nnz=1124963,
layout=torch.sparse_csr)
tensor([0.8443, 0.8073, 0.2696, ..., 0.1079, 0.2448, 0.9883])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1124963
Density: 4.9998355555555557e-05
Time: 10.834063529968262 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 6, 16, ..., 1124949,
1124959, 1124963]),
col_indices=tensor([ 30949, 52207, 66032, ..., 93409, 116462,
142125]),
values=tensor([-1.0709, 0.6351, 0.9891, ..., -0.8011, -0.8370,
-1.9774]), size=(150000, 150000), nnz=1124963,
layout=torch.sparse_csr)
tensor([0.8443, 0.8073, 0.2696, ..., 0.1079, 0.2448, 0.9883])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1124963
Density: 4.9998355555555557e-05
Time: 10.834063529968262 seconds
[20.44, 20.4, 20.2, 20.2, 20.28, 20.4, 20.28, 20.24, 20.24, 20.16]
[20.12, 20.08, 20.52, 25.28, 26.48, 29.56, 32.08, 31.44, 32.04, 32.2, 32.04, 31.84, 31.76, 32.04]
10.53046441078186
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 2982, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 1124963, 'MATRIX_DENSITY': 4.9998355555555557e-05, 'TIME_S': 10.834063529968262, 'TIME_S_1KI': 3.633153430572858, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 388.3960797214508, 'W': 36.8830912456038}
[20.44, 20.4, 20.2, 20.2, 20.28, 20.4, 20.28, 20.24, 20.24, 20.16, 20.24, 20.24, 20.16, 20.16, 20.12, 20.12, 20.12, 20.32, 20.52, 20.56]
364.7
18.235
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 2982, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 1124963, 'MATRIX_DENSITY': 4.9998355555555557e-05, 'TIME_S': 10.834063529968262, 'TIME_S_1KI': 3.633153430572858, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 388.3960797214508, 'W': 36.8830912456038, 'J_1KI': 130.24684095286747, 'W_1KI': 12.368575199733, 'W_D': 18.6480912456038, 'J_D': 196.3730611908436, 'W_D_1KI': 6.253551725554594, 'J_D_1KI': 2.0970998408969126}

1
pytorch/output_1core_after_test/altra_10_10_10_150000_8e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 2120, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1799917, "MATRIX_DENSITY": 7.999631111111111e-05, "TIME_S": 10.117506265640259, "TIME_S_1KI": 4.772408615868047, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 423.4199336338043, "W": 36.590743995468394, "J_1KI": 199.72638378953033, "W_1KI": 17.259784903522828, "W_D": 18.381743995468394, "J_D": 212.70944432282448, "W_D_1KI": 8.670633960126601, "J_D_1KI": 4.089921679305}

71
pytorch/output_1core_after_test/altra_10_10_10_150000_8e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 150000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1799926, "MATRIX_DENSITY": 7.999671111111111e-05, "TIME_S": 4.95142388343811}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 7, 15, ..., 1799908,
1799919, 1799926]),
col_indices=tensor([ 1214, 9526, 13372, ..., 119996, 126785,
136891]),
values=tensor([-0.1068, 0.2457, -0.1318, ..., 1.2429, -0.8245,
-0.4292]), size=(150000, 150000), nnz=1799926,
layout=torch.sparse_csr)
tensor([0.8310, 0.0944, 0.9117, ..., 0.1166, 0.0113, 0.2839])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1799926
Density: 7.999671111111111e-05
Time: 4.95142388343811 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 2120 -ss 150000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1799917, "MATRIX_DENSITY": 7.999631111111111e-05, "TIME_S": 10.117506265640259}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 8, 18, ..., 1799892,
1799905, 1799917]),
col_indices=tensor([ 1343, 1624, 10718, ..., 128180, 139489,
145861]),
values=tensor([ 1.0168, -0.7101, 0.5768, ..., -0.0198, -0.1886,
-0.2993]), size=(150000, 150000), nnz=1799917,
layout=torch.sparse_csr)
tensor([0.6697, 0.3813, 0.8738, ..., 0.3394, 0.0147, 0.2298])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1799917
Density: 7.999631111111111e-05
Time: 10.117506265640259 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 8, 18, ..., 1799892,
1799905, 1799917]),
col_indices=tensor([ 1343, 1624, 10718, ..., 128180, 139489,
145861]),
values=tensor([ 1.0168, -0.7101, 0.5768, ..., -0.0198, -0.1886,
-0.2993]), size=(150000, 150000), nnz=1799917,
layout=torch.sparse_csr)
tensor([0.6697, 0.3813, 0.8738, ..., 0.3394, 0.0147, 0.2298])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1799917
Density: 7.999631111111111e-05
Time: 10.117506265640259 seconds
[20.28, 20.24, 20.28, 20.28, 20.24, 20.08, 20.04, 19.92, 20.12, 20.08]
[20.32, 20.6, 23.68, 23.68, 25.6, 28.08, 30.28, 33.04, 31.4, 31.84, 32.6, 32.6, 32.4, 32.52, 32.6]
11.571777105331421
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 2120, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 1799917, 'MATRIX_DENSITY': 7.999631111111111e-05, 'TIME_S': 10.117506265640259, 'TIME_S_1KI': 4.772408615868047, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 423.4199336338043, 'W': 36.590743995468394}
[20.28, 20.24, 20.28, 20.28, 20.24, 20.08, 20.04, 19.92, 20.12, 20.08, 20.36, 20.24, 20.2, 20.2, 20.52, 20.36, 20.24, 20.24, 20.4, 20.44]
364.18
18.209
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 2120, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 1799917, 'MATRIX_DENSITY': 7.999631111111111e-05, 'TIME_S': 10.117506265640259, 'TIME_S_1KI': 4.772408615868047, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 423.4199336338043, 'W': 36.590743995468394, 'J_1KI': 199.72638378953033, 'W_1KI': 17.259784903522828, 'W_D': 18.381743995468394, 'J_D': 212.70944432282448, 'W_D_1KI': 8.670633960126601, 'J_D_1KI': 4.089921679305}

1
pytorch/output_1core_after_test/altra_10_10_10_200000_0.0001.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 3999801, "MATRIX_DENSITY": 9.9995025e-05, "TIME_S": 13.38718843460083, "TIME_S_1KI": 13.38718843460083, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 463.430492515564, "W": 35.3250089736942, "J_1KI": 463.430492515564, "W_1KI": 35.3250089736942, "W_D": 16.8970089736942, "J_D": 221.67267378616333, "W_D_1KI": 16.8970089736942, "J_D_1KI": 16.8970089736942}

49
pytorch/output_1core_after_test/altra_10_10_10_200000_0.0001.output
View File

@ -1,49 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 200000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 3999801, "MATRIX_DENSITY": 9.9995025e-05, "TIME_S": 13.38718843460083}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 30, 49, ..., 3999760,
3999779, 3999801]),
col_indices=tensor([ 725, 16500, 17380, ..., 191062, 191507,
194960]),
values=tensor([ 1.0478, 2.6646, -0.7212, ..., -0.5863, -0.3201,
2.0476]), size=(200000, 200000), nnz=3999801,
layout=torch.sparse_csr)
tensor([0.6554, 0.2512, 0.1221, ..., 0.3923, 0.3935, 0.0593])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 3999801
Density: 9.9995025e-05
Time: 13.38718843460083 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 30, 49, ..., 3999760,
3999779, 3999801]),
col_indices=tensor([ 725, 16500, 17380, ..., 191062, 191507,
194960]),
values=tensor([ 1.0478, 2.6646, -0.7212, ..., -0.5863, -0.3201,
2.0476]), size=(200000, 200000), nnz=3999801,
layout=torch.sparse_csr)
tensor([0.6554, 0.2512, 0.1221, ..., 0.3923, 0.3935, 0.0593])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 3999801
Density: 9.9995025e-05
Time: 13.38718843460083 seconds
[20.04, 20.24, 20.72, 20.6, 20.56, 20.76, 20.76, 20.6, 20.48, 20.36]
[20.4, 20.4, 20.64, 22.08, 24.48, 25.76, 28.6, 30.32, 31.84, 31.76, 32.68, 32.84, 32.8, 32.96, 32.96, 32.88, 32.68]
13.119048118591309
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 3999801, 'MATRIX_DENSITY': 9.9995025e-05, 'TIME_S': 13.38718843460083, 'TIME_S_1KI': 13.38718843460083, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 463.430492515564, 'W': 35.3250089736942}
[20.04, 20.24, 20.72, 20.6, 20.56, 20.76, 20.76, 20.6, 20.48, 20.36, 20.2, 20.08, 20.04, 20.12, 20.12, 20.56, 20.64, 20.72, 20.84, 20.84]
368.56000000000006
18.428000000000004
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 3999801, 'MATRIX_DENSITY': 9.9995025e-05, 'TIME_S': 13.38718843460083, 'TIME_S_1KI': 13.38718843460083, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 463.430492515564, 'W': 35.3250089736942, 'J_1KI': 463.430492515564, 'W_1KI': 35.3250089736942, 'W_D': 16.8970089736942, 'J_D': 221.67267378616333, 'W_D_1KI': 16.8970089736942, 'J_D_1KI': 16.8970089736942}

1
pytorch/output_1core_after_test/altra_10_10_10_200000_1e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 4517, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 399999, "MATRIX_DENSITY": 9.999975e-06, "TIME_S": 10.48474931716919, "TIME_S_1KI": 2.321175407830239, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 350.1860060596466, "W": 33.203515607185345, "J_1KI": 77.52623556777654, "W_1KI": 7.350789375068706, "W_D": 14.75751560718534, "J_D": 155.6424178385734, "W_D_1KI": 3.267105514099035, "J_D_1KI": 0.7232910148547785}

71
pytorch/output_1core_after_test/altra_10_10_10_200000_1e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 200000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 399997, "MATRIX_DENSITY": 9.999925e-06, "TIME_S": 2.32439923286438}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 4, ..., 399995, 399996,
399997]),
col_indices=tensor([ 53721, 100176, 137115, ..., 76474, 111928,
67722]),
values=tensor([ 0.8787, 0.6153, 0.7457, ..., 1.5157, 1.9555,
-1.5636]), size=(200000, 200000), nnz=399997,
layout=torch.sparse_csr)
tensor([0.4176, 0.1900, 0.9801, ..., 0.0553, 0.1816, 0.9381])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 399997
Density: 9.999925e-06
Time: 2.32439923286438 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 4517 -ss 200000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 399999, "MATRIX_DENSITY": 9.999975e-06, "TIME_S": 10.48474931716919}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 3, ..., 399994, 399998,
399999]),
col_indices=tensor([ 23160, 67764, 94980, ..., 158664, 163872,
193419]),
values=tensor([ 0.4031, -0.2737, -0.3940, ..., -0.8147, 0.5871,
0.1087]), size=(200000, 200000), nnz=399999,
layout=torch.sparse_csr)
tensor([0.9178, 0.7444, 0.9877, ..., 0.0829, 0.8958, 0.1485])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 399999
Density: 9.999975e-06
Time: 10.48474931716919 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 3, ..., 399994, 399998,
399999]),
col_indices=tensor([ 23160, 67764, 94980, ..., 158664, 163872,
193419]),
values=tensor([ 0.4031, -0.2737, -0.3940, ..., -0.8147, 0.5871,
0.1087]), size=(200000, 200000), nnz=399999,
layout=torch.sparse_csr)
tensor([0.9178, 0.7444, 0.9877, ..., 0.0829, 0.8958, 0.1485])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 399999
Density: 9.999975e-06
Time: 10.48474931716919 seconds
[20.44, 20.44, 20.44, 20.28, 20.32, 20.44, 20.28, 20.44, 20.6, 20.32]
[20.24, 20.2, 21.16, 23.04, 25.4, 27.88, 30.44, 30.44, 31.44, 32.28, 31.92, 32.08, 32.28]
10.546654462814331
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 4517, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 399999, 'MATRIX_DENSITY': 9.999975e-06, 'TIME_S': 10.48474931716919, 'TIME_S_1KI': 2.321175407830239, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 350.1860060596466, 'W': 33.203515607185345}
[20.44, 20.44, 20.44, 20.28, 20.32, 20.44, 20.28, 20.44, 20.6, 20.32, 20.36, 20.4, 20.52, 20.8, 20.8, 20.84, 20.72, 20.56, 20.32, 20.32]
368.9200000000001
18.446000000000005
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 4517, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 399999, 'MATRIX_DENSITY': 9.999975e-06, 'TIME_S': 10.48474931716919, 'TIME_S_1KI': 2.321175407830239, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 350.1860060596466, 'W': 33.203515607185345, 'J_1KI': 77.52623556777654, 'W_1KI': 7.350789375068706, 'W_D': 14.75751560718534, 'J_D': 155.6424178385734, 'W_D_1KI': 3.267105514099035, 'J_D_1KI': 0.7232910148547785}

1
pytorch/output_1core_after_test/altra_10_10_10_200000_2e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 3140, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 799993, "MATRIX_DENSITY": 1.9999825e-05, "TIME_S": 10.441259145736694, "TIME_S_1KI": 3.325241766158183, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 358.6360774898529, "W": 34.0829065787975, "J_1KI": 114.21531130250091, "W_1KI": 10.854428846750796, "W_D": 15.735906578797497, "J_D": 165.58047354674338, "W_D_1KI": 5.0114352161775475, "J_D_1KI": 1.5959984764896646}

71
pytorch/output_1core_after_test/altra_10_10_10_200000_2e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 200000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 799992, "MATRIX_DENSITY": 1.99998e-05, "TIME_S": 3.34374737739563}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 6, ..., 799986, 799987,
799992]),
col_indices=tensor([ 86419, 94662, 114023, ..., 79708, 99766,
133740]),
values=tensor([ 0.4049, -1.9715, 0.0309, ..., 0.2150, 1.8425,
-0.1233]), size=(200000, 200000), nnz=799992,
layout=torch.sparse_csr)
tensor([0.9939, 0.9385, 0.2115, ..., 0.3578, 0.8825, 0.1238])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 799992
Density: 1.99998e-05
Time: 3.34374737739563 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 3140 -ss 200000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 799993, "MATRIX_DENSITY": 1.9999825e-05, "TIME_S": 10.441259145736694}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 8, ..., 799988, 799989,
799993]),
col_indices=tensor([ 56748, 164626, 184621, ..., 38108, 111530,
145433]),
values=tensor([-0.3915, -2.3778, -0.8361, ..., 0.8031, 1.0234,
-0.4414]), size=(200000, 200000), nnz=799993,
layout=torch.sparse_csr)
tensor([0.2982, 0.7175, 0.9318, ..., 0.5875, 0.3003, 0.7472])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 799993
Density: 1.9999825e-05
Time: 10.441259145736694 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 8, ..., 799988, 799989,
799993]),
col_indices=tensor([ 56748, 164626, 184621, ..., 38108, 111530,
145433]),
values=tensor([-0.3915, -2.3778, -0.8361, ..., 0.8031, 1.0234,
-0.4414]), size=(200000, 200000), nnz=799993,
layout=torch.sparse_csr)
tensor([0.2982, 0.7175, 0.9318, ..., 0.5875, 0.3003, 0.7472])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 799993
Density: 1.9999825e-05
Time: 10.441259145736694 seconds
[20.04, 20.24, 20.24, 20.2, 20.2, 20.44, 20.36, 20.48, 20.8, 20.8]
[20.72, 20.68, 23.56, 24.64, 26.8, 29.72, 32.0, 31.16, 31.16, 32.16, 31.92, 31.88, 31.88]
10.522461652755737
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 3140, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 799993, 'MATRIX_DENSITY': 1.9999825e-05, 'TIME_S': 10.441259145736694, 'TIME_S_1KI': 3.325241766158183, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 358.6360774898529, 'W': 34.0829065787975}
[20.04, 20.24, 20.24, 20.2, 20.2, 20.44, 20.36, 20.48, 20.8, 20.8, 20.32, 20.2, 20.28, 20.32, 20.4, 20.4, 20.48, 20.36, 20.6, 20.72]
366.94
18.347
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 3140, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 799993, 'MATRIX_DENSITY': 1.9999825e-05, 'TIME_S': 10.441259145736694, 'TIME_S_1KI': 3.325241766158183, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 358.6360774898529, 'W': 34.0829065787975, 'J_1KI': 114.21531130250091, 'W_1KI': 10.854428846750796, 'W_D': 15.735906578797497, 'J_D': 165.58047354674338, 'W_D_1KI': 5.0114352161775475, 'J_D_1KI': 1.5959984764896646}

1
pytorch/output_1core_after_test/altra_10_10_10_200000_5e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 1622, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 1999942, "MATRIX_DENSITY": 4.999855e-05, "TIME_S": 10.43860411643982, "TIME_S_1KI": 6.435637556374735, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 392.18215005874634, "W": 37.165866811389904, "J_1KI": 241.78924171316052, "W_1KI": 22.913604692595502, "W_D": 18.725866811389906, "J_D": 197.59933879852295, "W_D_1KI": 11.54492405141178, "J_D_1KI": 7.1177090329295805}

71
pytorch/output_1core_after_test/altra_10_10_10_200000_5e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 200000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 1999950, "MATRIX_DENSITY": 4.999875e-05, "TIME_S": 6.472595930099487}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 11, 24, ..., 1999927,
1999941, 1999950]),
col_indices=tensor([ 15347, 15852, 27357, ..., 175265, 186435,
196056]),
values=tensor([-0.1963, -0.5915, -0.4592, ..., -2.0562, -0.0108,
-1.5764]), size=(200000, 200000), nnz=1999950,
layout=torch.sparse_csr)
tensor([0.6431, 0.3328, 0.3894, ..., 0.6282, 0.8103, 0.7215])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 1999950
Density: 4.999875e-05
Time: 6.472595930099487 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1622 -ss 200000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 1999942, "MATRIX_DENSITY": 4.999855e-05, "TIME_S": 10.43860411643982}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 12, 27, ..., 1999920,
1999931, 1999942]),
col_indices=tensor([ 7401, 8093, 12306, ..., 126794, 152727,
181903]),
values=tensor([-1.8596, 1.4317, -0.3064, ..., -0.2326, 0.0517,
-1.4306]), size=(200000, 200000), nnz=1999942,
layout=torch.sparse_csr)
tensor([0.3318, 0.9133, 0.7802, ..., 0.9835, 0.0623, 0.0964])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 1999942
Density: 4.999855e-05
Time: 10.43860411643982 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 12, 27, ..., 1999920,
1999931, 1999942]),
col_indices=tensor([ 7401, 8093, 12306, ..., 126794, 152727,
181903]),
values=tensor([-1.8596, 1.4317, -0.3064, ..., -0.2326, 0.0517,
-1.4306]), size=(200000, 200000), nnz=1999942,
layout=torch.sparse_csr)
tensor([0.3318, 0.9133, 0.7802, ..., 0.9835, 0.0623, 0.0964])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 1999942
Density: 4.999855e-05
Time: 10.43860411643982 seconds
[20.28, 20.4, 20.4, 20.32, 20.56, 20.44, 20.64, 20.68, 20.64, 20.48]
[20.56, 20.28, 23.4, 24.56, 26.36, 29.12, 31.56, 31.56, 30.96, 32.48, 32.44, 32.36, 32.52, 32.6]
10.552213191986084
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1622, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 1999942, 'MATRIX_DENSITY': 4.999855e-05, 'TIME_S': 10.43860411643982, 'TIME_S_1KI': 6.435637556374735, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 392.18215005874634, 'W': 37.165866811389904}
[20.28, 20.4, 20.4, 20.32, 20.56, 20.44, 20.64, 20.68, 20.64, 20.48, 20.52, 20.32, 20.28, 20.08, 20.32, 20.48, 20.52, 20.68, 20.96, 20.88]
368.79999999999995
18.439999999999998
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1622, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 1999942, 'MATRIX_DENSITY': 4.999855e-05, 'TIME_S': 10.43860411643982, 'TIME_S_1KI': 6.435637556374735, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 392.18215005874634, 'W': 37.165866811389904, 'J_1KI': 241.78924171316052, 'W_1KI': 22.913604692595502, 'W_D': 18.725866811389906, 'J_D': 197.59933879852295, 'W_D_1KI': 11.54492405141178, 'J_D_1KI': 7.1177090329295805}

1
pytorch/output_1core_after_test/altra_10_10_10_200000_8e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 3199886, "MATRIX_DENSITY": 7.999715e-05, "TIME_S": 11.041945934295654, "TIME_S_1KI": 11.041945934295654, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 420.73246814727787, "W": 37.627043073018406, "J_1KI": 420.73246814727787, "W_1KI": 37.627043073018406, "W_D": 17.962043073018407, "J_D": 200.8452989625931, "W_D_1KI": 17.962043073018407, "J_D_1KI": 17.962043073018407}

49
pytorch/output_1core_after_test/altra_10_10_10_200000_8e-05.output
View File

@ -1,49 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 200000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 3199886, "MATRIX_DENSITY": 7.999715e-05, "TIME_S": 11.041945934295654}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 13, 29, ..., 3199850,
3199873, 3199886]),
col_indices=tensor([ 1474, 32917, 59625, ..., 164776, 165534,
165742]),
values=tensor([ 0.5733, -1.4295, 0.3956, ..., 0.0085, 0.5997,
0.3568]), size=(200000, 200000), nnz=3199886,
layout=torch.sparse_csr)
tensor([0.1783, 0.3519, 0.3509, ..., 0.9970, 0.4624, 0.8799])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 3199886
Density: 7.999715e-05
Time: 11.041945934295654 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 13, 29, ..., 3199850,
3199873, 3199886]),
col_indices=tensor([ 1474, 32917, 59625, ..., 164776, 165534,
165742]),
values=tensor([ 0.5733, -1.4295, 0.3956, ..., 0.0085, 0.5997,
0.3568]), size=(200000, 200000), nnz=3199886,
layout=torch.sparse_csr)
tensor([0.1783, 0.3519, 0.3509, ..., 0.9970, 0.4624, 0.8799])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 3199886
Density: 7.999715e-05
Time: 11.041945934295654 seconds
[21.72, 22.52, 23.56, 24.04, 24.04, 23.64, 23.04, 23.16, 23.04, 23.04]
[23.24, 23.4, 23.0, 24.04, 25.0, 26.92, 29.2, 31.12, 32.12, 33.16, 34.04, 34.56, 34.68, 34.56, 34.2]
11.18165111541748
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 3199886, 'MATRIX_DENSITY': 7.999715e-05, 'TIME_S': 11.041945934295654, 'TIME_S_1KI': 11.041945934295654, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 420.73246814727787, 'W': 37.627043073018406}
[21.72, 22.52, 23.56, 24.04, 24.04, 23.64, 23.04, 23.16, 23.04, 23.04, 20.24, 20.32, 20.48, 20.44, 20.44, 20.44, 20.48, 20.52, 20.4, 20.48]
393.29999999999995
19.665
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 3199886, 'MATRIX_DENSITY': 7.999715e-05, 'TIME_S': 11.041945934295654, 'TIME_S_1KI': 11.041945934295654, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 420.73246814727787, 'W': 37.627043073018406, 'J_1KI': 420.73246814727787, 'W_1KI': 37.627043073018406, 'W_D': 17.962043073018407, 'J_D': 200.8452989625931, 'W_D_1KI': 17.962043073018407, 'J_D_1KI': 17.962043073018407}

1
pytorch/output_1core_after_test/altra_10_10_10_20000_0.0001.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 58715, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 39997, "MATRIX_DENSITY": 9.99925e-05, "TIME_S": 10.23157262802124, "TIME_S_1KI": 0.17425824113124824, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 312.74463246345516, "W": 30.541103907235787, "J_1KI": 5.326486118767865, "W_1KI": 0.5201584587794564, "W_D": 12.259103907235787, "J_D": 125.53472059965131, "W_D_1KI": 0.2087899839433839, "J_D_1KI": 0.003555990529564573}

65
pytorch/output_1core_after_test/altra_10_10_10_20000_0.0001.output
View File

@ -1,65 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 20000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 39999, "MATRIX_DENSITY": 9.99975e-05, "TIME_S": 0.17882966995239258}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 4, ..., 39995, 39997, 39999]),
col_indices=tensor([ 1398, 8266, 9733, ..., 5901, 6485, 19808]),
values=tensor([ 1.4442, 0.3161, 0.6925, ..., -1.6441, -1.7494,
-0.0189]), size=(20000, 20000), nnz=39999,
layout=torch.sparse_csr)
tensor([0.3364, 0.8734, 0.2560, ..., 0.4245, 0.7879, 0.4227])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 39999
Density: 9.99975e-05
Time: 0.17882966995239258 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 58715 -ss 20000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 39997, "MATRIX_DENSITY": 9.99925e-05, "TIME_S": 10.23157262802124}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 2, ..., 39994, 39995, 39997]),
col_indices=tensor([ 385, 13575, 993, ..., 7905, 3269, 19471]),
values=tensor([-0.9779, 0.5850, -0.0057, ..., 1.2177, 0.6236,
-0.3848]), size=(20000, 20000), nnz=39997,
layout=torch.sparse_csr)
tensor([0.4733, 0.3692, 0.8512, ..., 0.6396, 0.1954, 0.4828])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 39997
Density: 9.99925e-05
Time: 10.23157262802124 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 2, ..., 39994, 39995, 39997]),
col_indices=tensor([ 385, 13575, 993, ..., 7905, 3269, 19471]),
values=tensor([-0.9779, 0.5850, -0.0057, ..., 1.2177, 0.6236,
-0.3848]), size=(20000, 20000), nnz=39997,
layout=torch.sparse_csr)
tensor([0.4733, 0.3692, 0.8512, ..., 0.6396, 0.1954, 0.4828])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 39997
Density: 9.99925e-05
Time: 10.23157262802124 seconds
[19.96, 19.8, 19.8, 19.88, 20.12, 20.6, 20.88, 20.88, 21.08, 21.08]
[20.88, 20.84, 20.88, 24.6, 26.68, 28.84, 29.92, 30.28, 26.48, 25.4, 25.04, 24.96, 24.84]
10.240122079849243
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 58715, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [20000, 20000], 'MATRIX_ROWS': 20000, 'MATRIX_SIZE': 400000000, 'MATRIX_NNZ': 39997, 'MATRIX_DENSITY': 9.99925e-05, 'TIME_S': 10.23157262802124, 'TIME_S_1KI': 0.17425824113124824, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 312.74463246345516, 'W': 30.541103907235787}
[19.96, 19.8, 19.8, 19.88, 20.12, 20.6, 20.88, 20.88, 21.08, 21.08, 20.16, 20.24, 20.16, 20.12, 20.16, 20.16, 20.24, 20.32, 20.4, 20.4]
365.64
18.282
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 58715, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [20000, 20000], 'MATRIX_ROWS': 20000, 'MATRIX_SIZE': 400000000, 'MATRIX_NNZ': 39997, 'MATRIX_DENSITY': 9.99925e-05, 'TIME_S': 10.23157262802124, 'TIME_S_1KI': 0.17425824113124824, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 312.74463246345516, 'W': 30.541103907235787, 'J_1KI': 5.326486118767865, 'W_1KI': 0.5201584587794564, 'W_D': 12.259103907235787, 'J_D': 125.53472059965131, 'W_D_1KI': 0.2087899839433839, 'J_D_1KI': 0.003555990529564573}

1
pytorch/output_1core_after_test/altra_10_10_10_20000_1e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 175318, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 4000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.609092712402344, "TIME_S_1KI": 0.06051342538930597, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 296.8691830253601, "W": 28.18179818289856, "J_1KI": 1.6933183302647765, "W_1KI": 0.1607467469563796, "W_D": 9.98079818289856, "J_D": 105.1384792151451, "W_D_1KI": 0.05692968310668933, "J_D_1KI": 0.00032472240789131366}

85
pytorch/output_1core_after_test/altra_10_10_10_20000_1e-05.output
View File

@ -1,85 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 20000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 4000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.06297636032104492}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 2, ..., 4000, 4000, 4000]),
col_indices=tensor([12357, 15223, 9231, ..., 10258, 1732, 65]),
values=tensor([ 0.8374, 0.1792, -0.7911, ..., -0.9962, 0.9719,
1.4813]), size=(20000, 20000), nnz=4000,
layout=torch.sparse_csr)
tensor([0.0473, 0.1860, 0.9022, ..., 0.7616, 0.8259, 0.0597])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 4000
Density: 1e-05
Time: 0.06297636032104492 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 166729 -ss 20000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 4000, "MATRIX_DENSITY": 1e-05, "TIME_S": 9.985570907592773}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 4000, 4000, 4000]),
col_indices=tensor([14783, 10920, 11726, ..., 6504, 7631, 16250]),
values=tensor([-0.3836, -0.4175, -1.2173, ..., 1.7428, 0.1050,
0.7111]), size=(20000, 20000), nnz=4000,
layout=torch.sparse_csr)
tensor([0.9840, 0.6859, 0.7359, ..., 0.2102, 0.3574, 0.8617])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 4000
Density: 1e-05
Time: 9.985570907592773 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 175318 -ss 20000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 4000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.609092712402344}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 3999, 3999, 4000]),
col_indices=tensor([11238, 15714, 3500, ..., 13500, 6351, 10546]),
values=tensor([-1.2322, -0.5076, 1.3304, ..., -0.0344, 1.2521,
0.5111]), size=(20000, 20000), nnz=4000,
layout=torch.sparse_csr)
tensor([0.2801, 0.3002, 0.3051, ..., 0.7279, 0.9688, 0.7873])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 4000
Density: 1e-05
Time: 10.609092712402344 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 3999, 3999, 4000]),
col_indices=tensor([11238, 15714, 3500, ..., 13500, 6351, 10546]),
values=tensor([-1.2322, -0.5076, 1.3304, ..., -0.0344, 1.2521,
0.5111]), size=(20000, 20000), nnz=4000,
layout=torch.sparse_csr)
tensor([0.2801, 0.3002, 0.3051, ..., 0.7279, 0.9688, 0.7873])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 4000
Density: 1e-05
Time: 10.609092712402344 seconds
[20.48, 20.4, 20.48, 20.44, 20.44, 20.32, 20.24, 20.32, 20.6, 20.52]
[20.56, 20.4, 20.64, 24.4, 25.8, 26.48, 27.36, 24.32, 23.96, 23.12, 23.08, 23.08, 23.16]
10.534075260162354
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 175318, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [20000, 20000], 'MATRIX_ROWS': 20000, 'MATRIX_SIZE': 400000000, 'MATRIX_NNZ': 4000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.609092712402344, 'TIME_S_1KI': 0.06051342538930597, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 296.8691830253601, 'W': 28.18179818289856}
[20.48, 20.4, 20.48, 20.44, 20.44, 20.32, 20.24, 20.32, 20.6, 20.52, 20.28, 20.2, 20.16, 19.92, 19.96, 19.8, 19.96, 19.8, 20.16, 20.36]
364.02
18.201
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 175318, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [20000, 20000], 'MATRIX_ROWS': 20000, 'MATRIX_SIZE': 400000000, 'MATRIX_NNZ': 4000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.609092712402344, 'TIME_S_1KI': 0.06051342538930597, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 296.8691830253601, 'W': 28.18179818289856, 'J_1KI': 1.6933183302647765, 'W_1KI': 0.1607467469563796, 'W_D': 9.98079818289856, 'J_D': 105.1384792151451, 'W_D_1KI': 0.05692968310668933, 'J_D_1KI': 0.00032472240789131366}

1
pytorch/output_1core_after_test/altra_10_10_10_20000_2e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 118942, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 8000, "MATRIX_DENSITY": 2e-05, "TIME_S": 10.07962417602539, "TIME_S_1KI": 0.08474402798023735, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 297.63386821746826, "W": 28.954115811135768, "J_1KI": 2.5023445731320164, "W_1KI": 0.24343054439252548, "W_D": 10.466115811135769, "J_D": 107.58645003700256, "W_D_1KI": 0.08799344059403548, "J_D_1KI": 0.0007398012526612591}

66
pytorch/output_1core_after_test/altra_10_10_10_20000_2e-05.output
View File

@ -1,66 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 20000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 8000, "MATRIX_DENSITY": 2e-05, "TIME_S": 0.08827829360961914}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 3, ..., 8000, 8000, 8000]),
col_indices=tensor([11437, 4018, 19190, ..., 10689, 12356, 1797]),
values=tensor([ 0.9012, 2.0083, 1.2437, ..., -1.5308, 0.0468,
-1.8336]), size=(20000, 20000), nnz=8000,
layout=torch.sparse_csr)
tensor([7.3997e-01, 9.3806e-01, 1.1245e-01, ..., 3.6502e-04, 1.7307e-01,
5.9848e-01])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 8000
Density: 2e-05
Time: 0.08827829360961914 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 118942 -ss 20000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 8000, "MATRIX_DENSITY": 2e-05, "TIME_S": 10.07962417602539}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 8000, 8000, 8000]),
col_indices=tensor([ 5851, 16585, 17651, ..., 8618, 3900, 5823]),
values=tensor([ 0.3386, 1.2811, 0.0393, ..., -0.3587, 0.6718,
0.5199]), size=(20000, 20000), nnz=8000,
layout=torch.sparse_csr)
tensor([0.4256, 0.9835, 0.0324, ..., 0.1762, 0.1511, 0.0829])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 8000
Density: 2e-05
Time: 10.07962417602539 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 8000, 8000, 8000]),
col_indices=tensor([ 5851, 16585, 17651, ..., 8618, 3900, 5823]),
values=tensor([ 0.3386, 1.2811, 0.0393, ..., -0.3587, 0.6718,
0.5199]), size=(20000, 20000), nnz=8000,
layout=torch.sparse_csr)
tensor([0.4256, 0.9835, 0.0324, ..., 0.1762, 0.1511, 0.0829])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 8000
Density: 2e-05
Time: 10.07962417602539 seconds
[19.88, 20.04, 20.08, 20.4, 20.56, 20.6, 20.6, 20.64, 20.64, 20.4]
[20.68, 20.76, 20.84, 24.64, 26.64, 27.2, 27.92, 25.6, 24.76, 23.44, 23.72, 23.44, 23.52]
10.27950119972229
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 118942, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [20000, 20000], 'MATRIX_ROWS': 20000, 'MATRIX_SIZE': 400000000, 'MATRIX_NNZ': 8000, 'MATRIX_DENSITY': 2e-05, 'TIME_S': 10.07962417602539, 'TIME_S_1KI': 0.08474402798023735, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 297.63386821746826, 'W': 28.954115811135768}
[19.88, 20.04, 20.08, 20.4, 20.56, 20.6, 20.6, 20.64, 20.64, 20.4, 20.28, 20.52, 20.88, 20.84, 20.84, 20.68, 20.44, 20.68, 20.68, 20.72]
369.76
18.488
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 118942, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [20000, 20000], 'MATRIX_ROWS': 20000, 'MATRIX_SIZE': 400000000, 'MATRIX_NNZ': 8000, 'MATRIX_DENSITY': 2e-05, 'TIME_S': 10.07962417602539, 'TIME_S_1KI': 0.08474402798023735, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 297.63386821746826, 'W': 28.954115811135768, 'J_1KI': 2.5023445731320164, 'W_1KI': 0.24343054439252548, 'W_D': 10.466115811135769, 'J_D': 107.58645003700256, 'W_D_1KI': 0.08799344059403548, 'J_D_1KI': 0.0007398012526612591}

1
pytorch/output_1core_after_test/altra_10_10_10_20000_5e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 76193, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 20000, "MATRIX_DENSITY": 5e-05, "TIME_S": 10.029513835906982, "TIME_S_1KI": 0.1316330087528642, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 282.77299644470213, "W": 27.674543734558664, "J_1KI": 3.711272642430435, "W_1KI": 0.3632163549743239, "W_D": 9.323543734558665, "J_D": 95.26611981725691, "W_D_1KI": 0.12236745809403313, "J_D_1KI": 0.0016060196880820171}

65
pytorch/output_1core_after_test/altra_10_10_10_20000_5e-05.output
View File

@ -1,65 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 20000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 19998, "MATRIX_DENSITY": 4.9995e-05, "TIME_S": 0.13780617713928223}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 2, ..., 19996, 19997, 19998]),
col_indices=tensor([ 8564, 15371, 12390, ..., 9506, 1971, 15999]),
values=tensor([ 0.3043, -0.2329, -1.5565, ..., -0.7374, 0.0559,
1.5079]), size=(20000, 20000), nnz=19998,
layout=torch.sparse_csr)
tensor([0.5830, 0.7136, 0.4497, ..., 0.0929, 0.6777, 0.4633])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 19998
Density: 4.9995e-05
Time: 0.13780617713928223 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 76193 -ss 20000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 20000, "MATRIX_DENSITY": 5e-05, "TIME_S": 10.029513835906982}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 3, ..., 19996, 19998, 20000]),
col_indices=tensor([ 5096, 14038, 11024, ..., 11411, 2364, 2810]),
values=tensor([-1.0860, -0.1176, -0.2257, ..., 1.5728, -0.1349,
0.0418]), size=(20000, 20000), nnz=20000,
layout=torch.sparse_csr)
tensor([0.2807, 0.2512, 0.1212, ..., 0.2572, 0.0420, 0.0768])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 20000
Density: 5e-05
Time: 10.029513835906982 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 3, ..., 19996, 19998, 20000]),
col_indices=tensor([ 5096, 14038, 11024, ..., 11411, 2364, 2810]),
values=tensor([-1.0860, -0.1176, -0.2257, ..., 1.5728, -0.1349,
0.0418]), size=(20000, 20000), nnz=20000,
layout=torch.sparse_csr)
tensor([0.2807, 0.2512, 0.1212, ..., 0.2572, 0.0420, 0.0768])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 20000
Density: 5e-05
Time: 10.029513835906982 seconds
[20.72, 20.72, 20.64, 20.36, 20.4, 20.4, 20.24, 20.36, 20.48, 20.6]
[20.4, 20.32, 21.36, 22.4, 22.4, 23.88, 24.8, 25.24, 24.56, 24.32, 23.48, 23.16, 23.2]
10.217801570892334
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 76193, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [20000, 20000], 'MATRIX_ROWS': 20000, 'MATRIX_SIZE': 400000000, 'MATRIX_NNZ': 20000, 'MATRIX_DENSITY': 5e-05, 'TIME_S': 10.029513835906982, 'TIME_S_1KI': 0.1316330087528642, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 282.77299644470213, 'W': 27.674543734558664}
[20.72, 20.72, 20.64, 20.36, 20.4, 20.4, 20.24, 20.36, 20.48, 20.6, 20.28, 20.16, 20.16, 20.08, 20.2, 20.32, 20.4, 20.52, 20.52, 20.52]
367.02
18.351
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 76193, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [20000, 20000], 'MATRIX_ROWS': 20000, 'MATRIX_SIZE': 400000000, 'MATRIX_NNZ': 20000, 'MATRIX_DENSITY': 5e-05, 'TIME_S': 10.029513835906982, 'TIME_S_1KI': 0.1316330087528642, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 282.77299644470213, 'W': 27.674543734558664, 'J_1KI': 3.711272642430435, 'W_1KI': 0.3632163549743239, 'W_D': 9.323543734558665, 'J_D': 95.26611981725691, 'W_D_1KI': 0.12236745809403313, 'J_D_1KI': 0.0016060196880820171}

1
pytorch/output_1core_after_test/altra_10_10_10_20000_8e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 63424, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 31997, "MATRIX_DENSITY": 7.99925e-05, "TIME_S": 10.19437575340271, "TIME_S_1KI": 0.1607337246689378, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 285.14172214508056, "W": 27.849590636275384, "J_1KI": 4.495801623125009, "W_1KI": 0.43910176961836817, "W_D": 9.509590636275387, "J_D": 97.36520318508151, "W_D_1KI": 0.14993678475459427, "J_D_1KI": 0.0023640386092740016}

65
pytorch/output_1core_after_test/altra_10_10_10_20000_8e-05.output
View File

@ -1,65 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 20000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 31998, "MATRIX_DENSITY": 7.9995e-05, "TIME_S": 0.16555237770080566}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 1, ..., 31992, 31996, 31998]),
col_indices=tensor([13174, 16154, 19104, ..., 17316, 14628, 14714]),
values=tensor([ 0.2961, -0.6988, 1.4292, ..., 0.9249, -0.4549,
0.1182]), size=(20000, 20000), nnz=31998,
layout=torch.sparse_csr)
tensor([0.8123, 0.8879, 0.3353, ..., 0.0309, 0.7117, 0.9836])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 31998
Density: 7.9995e-05
Time: 0.16555237770080566 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 63424 -ss 20000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [20000, 20000], "MATRIX_ROWS": 20000, "MATRIX_SIZE": 400000000, "MATRIX_NNZ": 31997, "MATRIX_DENSITY": 7.99925e-05, "TIME_S": 10.19437575340271}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 1, ..., 31993, 31995, 31997]),
col_indices=tensor([ 1951, 4400, 13355, ..., 16423, 6899, 14719]),
values=tensor([-0.3339, 0.8334, 0.7225, ..., -0.9410, 1.7196,
-1.1716]), size=(20000, 20000), nnz=31997,
layout=torch.sparse_csr)
tensor([0.5883, 0.4106, 0.3171, ..., 0.8266, 0.4259, 0.6836])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 31997
Density: 7.99925e-05
Time: 10.19437575340271 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 1, ..., 31993, 31995, 31997]),
col_indices=tensor([ 1951, 4400, 13355, ..., 16423, 6899, 14719]),
values=tensor([-0.3339, 0.8334, 0.7225, ..., -0.9410, 1.7196,
-1.1716]), size=(20000, 20000), nnz=31997,
layout=torch.sparse_csr)
tensor([0.5883, 0.4106, 0.3171, ..., 0.8266, 0.4259, 0.6836])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([20000, 20000])
Rows: 20000
Size: 400000000
NNZ: 31997
Density: 7.99925e-05
Time: 10.19437575340271 seconds
[20.2, 20.24, 20.44, 20.56, 20.32, 20.24, 20.24, 20.68, 20.48, 20.48]
[20.76, 20.36, 20.52, 21.4, 23.0, 23.92, 24.8, 24.76, 25.0, 24.2, 24.12, 24.6, 24.48]
10.238632440567017
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 63424, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [20000, 20000], 'MATRIX_ROWS': 20000, 'MATRIX_SIZE': 400000000, 'MATRIX_NNZ': 31997, 'MATRIX_DENSITY': 7.99925e-05, 'TIME_S': 10.19437575340271, 'TIME_S_1KI': 0.1607337246689378, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 285.14172214508056, 'W': 27.849590636275384}
[20.2, 20.24, 20.44, 20.56, 20.32, 20.24, 20.24, 20.68, 20.48, 20.48, 20.2, 20.36, 20.36, 20.44, 20.44, 20.16, 20.28, 20.24, 20.52, 20.72]
366.79999999999995
18.339999999999996
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 63424, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [20000, 20000], 'MATRIX_ROWS': 20000, 'MATRIX_SIZE': 400000000, 'MATRIX_NNZ': 31997, 'MATRIX_DENSITY': 7.99925e-05, 'TIME_S': 10.19437575340271, 'TIME_S_1KI': 0.1607337246689378, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 285.14172214508056, 'W': 27.849590636275384, 'J_1KI': 4.495801623125009, 'W_1KI': 0.43910176961836817, 'W_D': 9.509590636275387, 'J_D': 97.36520318508151, 'W_D_1KI': 0.14993678475459427, 'J_D_1KI': 0.0023640386092740016}

1
pytorch/output_1core_after_test/altra_10_10_10_50000_0.0001.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 16114, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 249986, "MATRIX_DENSITY": 9.99944e-05, "TIME_S": 10.773088455200195, "TIME_S_1KI": 0.6685545770882584, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 356.3190417861938, "W": 34.1705954897688, "J_1KI": 22.112389337606665, "W_1KI": 2.120553276018915, "W_D": 15.8825954897688, "J_D": 165.6181616058349, "W_D_1KI": 0.9856395364136031, "J_D_1KI": 0.06116665858344317}

68
pytorch/output_1core_after_test/altra_10_10_10_50000_0.0001.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 50000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 249990, "MATRIX_DENSITY": 9.9996e-05, "TIME_S": 0.6515696048736572}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 6, 12, ..., 249983, 249988,
249990]),
col_indices=tensor([ 6662, 8889, 16052, ..., 41480, 19736, 47943]),
values=tensor([ 0.7313, 1.1544, -0.5654, ..., 0.5067, 2.7032,
1.2092]), size=(50000, 50000), nnz=249990,
layout=torch.sparse_csr)
tensor([0.0703, 0.6351, 0.6923, ..., 0.0380, 0.6908, 0.4954])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 249990
Density: 9.9996e-05
Time: 0.6515696048736572 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 16114 -ss 50000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 249986, "MATRIX_DENSITY": 9.99944e-05, "TIME_S": 10.773088455200195}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 6, ..., 249975, 249982,
249986]),
col_indices=tensor([ 9416, 19517, 30063, ..., 32154, 36782, 41226]),
values=tensor([ 0.8423, -0.3244, 1.0233, ..., 0.0855, 0.0139,
0.1714]), size=(50000, 50000), nnz=249986,
layout=torch.sparse_csr)
tensor([0.3530, 0.4524, 0.0808, ..., 0.2188, 0.9274, 0.8850])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 249986
Density: 9.99944e-05
Time: 10.773088455200195 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 6, ..., 249975, 249982,
249986]),
col_indices=tensor([ 9416, 19517, 30063, ..., 32154, 36782, 41226]),
values=tensor([ 0.8423, -0.3244, 1.0233, ..., 0.0855, 0.0139,
0.1714]), size=(50000, 50000), nnz=249986,
layout=torch.sparse_csr)
tensor([0.3530, 0.4524, 0.0808, ..., 0.2188, 0.9274, 0.8850])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 249986
Density: 9.99944e-05
Time: 10.773088455200195 seconds
[20.04, 20.08, 20.2, 20.16, 20.16, 20.24, 20.44, 20.8, 20.76, 20.8]
[20.8, 20.72, 20.76, 24.72, 26.88, 29.84, 32.2, 31.64, 32.56, 32.28, 32.28, 32.28, 32.08]
10.427650928497314
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 16114, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 249986, 'MATRIX_DENSITY': 9.99944e-05, 'TIME_S': 10.773088455200195, 'TIME_S_1KI': 0.6685545770882584, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 356.3190417861938, 'W': 34.1705954897688}
[20.04, 20.08, 20.2, 20.16, 20.16, 20.24, 20.44, 20.8, 20.76, 20.8, 20.12, 20.08, 20.28, 20.56, 20.48, 20.52, 20.28, 20.16, 20.04, 20.08]
365.76
18.288
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 16114, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 249986, 'MATRIX_DENSITY': 9.99944e-05, 'TIME_S': 10.773088455200195, 'TIME_S_1KI': 0.6685545770882584, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 356.3190417861938, 'W': 34.1705954897688, 'J_1KI': 22.112389337606665, 'W_1KI': 2.120553276018915, 'W_D': 15.8825954897688, 'J_D': 165.6181616058349, 'W_D_1KI': 0.9856395364136031, 'J_D_1KI': 0.06116665858344317}

1
pytorch/output_1core_after_test/altra_10_10_10_50000_1e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 39558, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.424606084823608, "TIME_S_1KI": 0.26352712687253166, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 324.11351709365846, "W": 30.726844173746105, "J_1KI": 8.193374717975086, "W_1KI": 0.7767542386810785, "W_D": 12.161844173746108, "J_D": 128.28580986738208, "W_D_1KI": 0.30744335339870843, "J_D_1KI": 0.007771964037583004}

65
pytorch/output_1core_after_test/altra_10_10_10_50000_1e-05.output
View File

@ -1,65 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 50000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.26543164253234863}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 2, ..., 24999, 24999, 25000]),
col_indices=tensor([22098, 23271, 16509, ..., 49035, 19856, 29710]),
values=tensor([-1.0630, -0.7063, -0.4487, ..., 0.5192, -0.7952,
-0.0211]), size=(50000, 50000), nnz=25000,
layout=torch.sparse_csr)
tensor([0.5959, 0.4899, 0.5718, ..., 0.0559, 0.3906, 0.5621])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 0.26543164253234863 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 39558 -ss 50000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.424606084823608}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 24999, 25000, 25000]),
col_indices=tensor([48058, 11500, 28092, ..., 28559, 6217, 24317]),
values=tensor([-0.1678, 1.6669, -1.7696, ..., 1.2303, 1.8947,
-1.1526]), size=(50000, 50000), nnz=25000,
layout=torch.sparse_csr)
tensor([0.2915, 0.7675, 0.3440, ..., 0.8675, 0.1613, 0.7154])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 10.424606084823608 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 24999, 25000, 25000]),
col_indices=tensor([48058, 11500, 28092, ..., 28559, 6217, 24317]),
values=tensor([-0.1678, 1.6669, -1.7696, ..., 1.2303, 1.8947,
-1.1526]), size=(50000, 50000), nnz=25000,
layout=torch.sparse_csr)
tensor([0.2915, 0.7675, 0.3440, ..., 0.8675, 0.1613, 0.7154])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 10.424606084823608 seconds
[20.36, 20.32, 20.28, 20.52, 20.52, 20.64, 20.8, 20.64, 20.64, 20.64]
[20.64, 20.64, 20.28, 23.88, 25.6, 27.36, 28.68, 29.84, 27.6, 27.0, 27.08, 27.28, 27.24]
10.548220157623291
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 39558, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 25000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.424606084823608, 'TIME_S_1KI': 0.26352712687253166, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 324.11351709365846, 'W': 30.726844173746105}
[20.36, 20.32, 20.28, 20.52, 20.52, 20.64, 20.8, 20.64, 20.64, 20.64, 20.64, 20.84, 21.04, 20.96, 20.92, 20.76, 20.76, 20.56, 20.16, 20.24]
371.29999999999995
18.564999999999998
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 39558, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 25000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.424606084823608, 'TIME_S_1KI': 0.26352712687253166, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 324.11351709365846, 'W': 30.726844173746105, 'J_1KI': 8.193374717975086, 'W_1KI': 0.7767542386810785, 'W_D': 12.161844173746108, 'J_D': 128.28580986738208, 'W_D_1KI': 0.30744335339870843, 'J_D_1KI': 0.007771964037583004}

1
pytorch/output_1core_after_test/altra_10_10_10_50000_2e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 29180, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 50000, "MATRIX_DENSITY": 2e-05, "TIME_S": 10.342068433761597, "TIME_S_1KI": 0.35442318141746393, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 314.51990249633786, "W": 30.360391076955718, "J_1KI": 10.778612148606507, "W_1KI": 1.040452058840155, "W_D": 11.935391076955717, "J_D": 123.64524647474285, "W_D_1KI": 0.4090264248442672, "J_D_1KI": 0.014017355203710322}

65
pytorch/output_1core_after_test/altra_10_10_10_50000_2e-05.output
View File

@ -1,65 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 50000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 49998, "MATRIX_DENSITY": 1.99992e-05, "TIME_S": 0.3598310947418213}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 1, ..., 49996, 49996, 49998]),
col_indices=tensor([30529, 7062, 29506, ..., 35953, 37426, 43003]),
values=tensor([ 0.6020, 0.0624, 0.2604, ..., 1.2885, -2.2140,
1.3375]), size=(50000, 50000), nnz=49998,
layout=torch.sparse_csr)
tensor([0.3283, 0.8413, 0.6070, ..., 0.6287, 0.3886, 0.1587])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 49998
Density: 1.99992e-05
Time: 0.3598310947418213 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 29180 -ss 50000 -sd 2e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 50000, "MATRIX_DENSITY": 2e-05, "TIME_S": 10.342068433761597}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 4, ..., 49998, 49999, 50000]),
col_indices=tensor([24033, 20967, 34679, ..., 22694, 884, 7980]),
values=tensor([-0.8208, 1.0839, 0.5317, ..., -1.4749, 0.0532,
-0.4907]), size=(50000, 50000), nnz=50000,
layout=torch.sparse_csr)
tensor([0.2365, 0.0619, 0.0494, ..., 0.8664, 0.4569, 0.5629])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 50000
Density: 2e-05
Time: 10.342068433761597 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 4, ..., 49998, 49999, 50000]),
col_indices=tensor([24033, 20967, 34679, ..., 22694, 884, 7980]),
values=tensor([-0.8208, 1.0839, 0.5317, ..., -1.4749, 0.0532,
-0.4907]), size=(50000, 50000), nnz=50000,
layout=torch.sparse_csr)
tensor([0.2365, 0.0619, 0.0494, ..., 0.8664, 0.4569, 0.5629])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 50000
Density: 2e-05
Time: 10.342068433761597 seconds
[20.28, 20.36, 20.08, 20.16, 20.16, 20.08, 20.24, 20.08, 20.2, 20.24]
[20.24, 20.44, 20.36, 21.44, 22.44, 25.12, 26.56, 27.96, 28.88, 28.88, 29.08, 28.76, 28.48]
10.359547138214111
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 29180, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 50000, 'MATRIX_DENSITY': 2e-05, 'TIME_S': 10.342068433761597, 'TIME_S_1KI': 0.35442318141746393, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 314.51990249633786, 'W': 30.360391076955718}
[20.28, 20.36, 20.08, 20.16, 20.16, 20.08, 20.24, 20.08, 20.2, 20.24, 20.4, 20.36, 20.44, 20.84, 20.92, 21.04, 21.04, 21.0, 20.68, 20.72]
368.5
18.425
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 29180, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 50000, 'MATRIX_DENSITY': 2e-05, 'TIME_S': 10.342068433761597, 'TIME_S_1KI': 0.35442318141746393, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 314.51990249633786, 'W': 30.360391076955718, 'J_1KI': 10.778612148606507, 'W_1KI': 1.040452058840155, 'W_D': 11.935391076955717, 'J_D': 123.64524647474285, 'W_D_1KI': 0.4090264248442672, 'J_D_1KI': 0.014017355203710322}

1
pytorch/output_1core_after_test/altra_10_10_10_50000_5e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 21246, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 124997, "MATRIX_DENSITY": 4.99988e-05, "TIME_S": 10.39963436126709, "TIME_S_1KI": 0.4894866968496229, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 348.6815179252624, "W": 32.90522736665826, "J_1KI": 16.41163126825108, "W_1KI": 1.5487728215503274, "W_D": 14.517227366658261, "J_D": 153.83236279964444, "W_D_1KI": 0.6832922605035424, "J_D_1KI": 0.03216098373828214}

68
pytorch/output_1core_after_test/altra_10_10_10_50000_5e-05.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 50000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 124997, "MATRIX_DENSITY": 4.99988e-05, "TIME_S": 0.49421024322509766}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 3, ..., 124993, 124996,
124997]),
col_indices=tensor([ 1273, 22428, 29987, ..., 14261, 20854, 19550]),
values=tensor([ 1.3964, 1.1880, -2.4586, ..., 0.2900, -1.6227,
1.1179]), size=(50000, 50000), nnz=124997,
layout=torch.sparse_csr)
tensor([0.8998, 0.8412, 0.9493, ..., 0.6279, 0.1643, 0.8336])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 124997
Density: 4.99988e-05
Time: 0.49421024322509766 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 21246 -ss 50000 -sd 5e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 124997, "MATRIX_DENSITY": 4.99988e-05, "TIME_S": 10.39963436126709}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 5, ..., 124992, 124995,
124997]),
col_indices=tensor([33843, 5335, 28219, ..., 48135, 16744, 44054]),
values=tensor([ 0.6356, -0.4709, 1.5854, ..., -0.0049, -0.5240,
-0.7657]), size=(50000, 50000), nnz=124997,
layout=torch.sparse_csr)
tensor([0.3855, 0.1623, 0.0527, ..., 0.6589, 0.5383, 0.0901])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 124997
Density: 4.99988e-05
Time: 10.39963436126709 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 5, ..., 124992, 124995,
124997]),
col_indices=tensor([33843, 5335, 28219, ..., 48135, 16744, 44054]),
values=tensor([ 0.6356, -0.4709, 1.5854, ..., -0.0049, -0.5240,
-0.7657]), size=(50000, 50000), nnz=124997,
layout=torch.sparse_csr)
tensor([0.3855, 0.1623, 0.0527, ..., 0.6589, 0.5383, 0.0901])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 124997
Density: 4.99988e-05
Time: 10.39963436126709 seconds
[20.44, 20.6, 20.44, 20.56, 20.72, 20.48, 20.56, 20.8, 20.88, 20.88]
[20.8, 20.76, 20.96, 21.92, 24.04, 27.24, 29.92, 31.4, 31.72, 31.72, 31.76, 31.84, 31.92]
10.596538782119751
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 21246, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 124997, 'MATRIX_DENSITY': 4.99988e-05, 'TIME_S': 10.39963436126709, 'TIME_S_1KI': 0.4894866968496229, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 348.6815179252624, 'W': 32.90522736665826}
[20.44, 20.6, 20.44, 20.56, 20.72, 20.48, 20.56, 20.8, 20.88, 20.88, 20.28, 20.32, 20.28, 20.4, 20.16, 20.16, 20.24, 20.08, 20.12, 20.32]
367.76
18.387999999999998
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 21246, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 124997, 'MATRIX_DENSITY': 4.99988e-05, 'TIME_S': 10.39963436126709, 'TIME_S_1KI': 0.4894866968496229, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 348.6815179252624, 'W': 32.90522736665826, 'J_1KI': 16.41163126825108, 'W_1KI': 1.5487728215503274, 'W_D': 14.517227366658261, 'J_D': 153.83236279964444, 'W_D_1KI': 0.6832922605035424, 'J_D_1KI': 0.03216098373828214}

1
pytorch/output_1core_after_test/altra_10_10_10_50000_8e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 17928, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 199993, "MATRIX_DENSITY": 7.99972e-05, "TIME_S": 10.449234962463379, "TIME_S_1KI": 0.5828444311949675, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 345.35830185890194, "W": 33.32021277823694, "J_1KI": 19.263626832825857, "W_1KI": 1.858557160767344, "W_D": 14.958212778236941, "J_D": 155.03931497430793, "W_D_1KI": 0.8343492178847022, "J_D_1KI": 0.0465388898864738}

68
pytorch/output_1core_after_test/altra_10_10_10_50000_8e-05.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 50000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 199993, "MATRIX_DENSITY": 7.99972e-05, "TIME_S": 0.5856485366821289}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 8, 13, ..., 199985, 199988,
199993]),
col_indices=tensor([ 4116, 20821, 23313, ..., 36221, 39671, 48300]),
values=tensor([-1.1656, 0.6488, 0.3884, ..., 0.8608, -1.0532,
-1.6884]), size=(50000, 50000), nnz=199993,
layout=torch.sparse_csr)
tensor([0.0024, 0.6993, 0.1691, ..., 0.8154, 0.5901, 0.4003])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 199993
Density: 7.99972e-05
Time: 0.5856485366821289 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 17928 -ss 50000 -sd 8e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 199993, "MATRIX_DENSITY": 7.99972e-05, "TIME_S": 10.449234962463379}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 5, ..., 199987, 199991,
199993]),
col_indices=tensor([17990, 18143, 26452, ..., 25515, 3657, 45119]),
values=tensor([-0.8402, -1.2988, 1.1344, ..., -1.1042, 0.4643,
1.1586]), size=(50000, 50000), nnz=199993,
layout=torch.sparse_csr)
tensor([0.2314, 0.4382, 0.4620, ..., 0.3725, 0.7017, 0.5878])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 199993
Density: 7.99972e-05
Time: 10.449234962463379 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at /space/jenkins/workspace/Releases/pytorch-dls/pytorch-dls/aten/src/ATen/SparseCsrTensorImpl.cpp:55.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 5, ..., 199987, 199991,
199993]),
col_indices=tensor([17990, 18143, 26452, ..., 25515, 3657, 45119]),
values=tensor([-0.8402, -1.2988, 1.1344, ..., -1.1042, 0.4643,
1.1586]), size=(50000, 50000), nnz=199993,
layout=torch.sparse_csr)
tensor([0.2314, 0.4382, 0.4620, ..., 0.3725, 0.7017, 0.5878])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 199993
Density: 7.99972e-05
Time: 10.449234962463379 seconds
[20.36, 20.36, 20.16, 20.32, 20.4, 20.52, 20.64, 20.64, 20.6, 20.56]
[20.44, 20.44, 20.36, 23.72, 25.8, 28.2, 31.08, 30.32, 31.88, 31.92, 31.76, 31.88, 32.12]
10.364828824996948
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 17928, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 199993, 'MATRIX_DENSITY': 7.99972e-05, 'TIME_S': 10.449234962463379, 'TIME_S_1KI': 0.5828444311949675, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 345.35830185890194, 'W': 33.32021277823694}
[20.36, 20.36, 20.16, 20.32, 20.4, 20.52, 20.64, 20.64, 20.6, 20.56, 20.76, 20.64, 20.32, 20.32, 20.08, 20.4, 20.32, 20.28, 20.36, 20.08]
367.24
18.362000000000002
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 17928, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 199993, 'MATRIX_DENSITY': 7.99972e-05, 'TIME_S': 10.449234962463379, 'TIME_S_1KI': 0.5828444311949675, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 345.35830185890194, 'W': 33.32021277823694, 'J_1KI': 19.263626832825857, 'W_1KI': 1.858557160767344, 'W_D': 14.958212778236941, 'J_D': 155.03931497430793, 'W_D_1KI': 0.8343492178847022, 'J_D_1KI': 0.0465388898864738}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_100000_0.0001.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 7321, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 999958, "MATRIX_DENSITY": 9.99958e-05, "TIME_S": 10.493394613265991, "TIME_S_1KI": 1.4333280444291752, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 693.8586197495459, "W": 66.35, "J_1KI": 94.77648132079578, "W_1KI": 9.06296953968037, "W_D": 31.555249999999994, "J_D": 329.9906889352202, "W_D_1KI": 4.310237672449118, "J_D_1KI": 0.5887498528137027}

68
pytorch/output_1core_after_test/epyc_7313p_10_10_10_100000_0.0001.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '100000', '-sd', '0.0001', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 999955, "MATRIX_DENSITY": 9.99955e-05, "TIME_S": 1.434152603149414}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 11, 23, ..., 999937, 999947,
999955]),
col_indices=tensor([17714, 27606, 40423, ..., 56745, 68426, 94681]),
values=tensor([-0.0848, -2.5543, -0.9845, ..., 1.0991, -1.2721,
0.0094]), size=(100000, 100000), nnz=999955,
layout=torch.sparse_csr)
tensor([0.0317, 0.5212, 0.6740, ..., 0.1470, 0.6060, 0.4229])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 999955
Density: 9.99955e-05
Time: 1.434152603149414 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '7321', '-ss', '100000', '-sd', '0.0001', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 999958, "MATRIX_DENSITY": 9.99958e-05, "TIME_S": 10.493394613265991}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 17, ..., 999929, 999946,
999958]),
col_indices=tensor([ 3686, 11174, 36004, ..., 72478, 81947, 88062]),
values=tensor([ 1.2821, -0.7142, -0.0602, ..., 0.1059, 0.3571,
1.8677]), size=(100000, 100000), nnz=999958,
layout=torch.sparse_csr)
tensor([0.8470, 0.5279, 0.2762, ..., 0.6136, 0.0054, 0.0656])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 999958
Density: 9.99958e-05
Time: 10.493394613265991 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 17, ..., 999929, 999946,
999958]),
col_indices=tensor([ 3686, 11174, 36004, ..., 72478, 81947, 88062]),
values=tensor([ 1.2821, -0.7142, -0.0602, ..., 0.1059, 0.3571,
1.8677]), size=(100000, 100000), nnz=999958,
layout=torch.sparse_csr)
tensor([0.8470, 0.5279, 0.2762, ..., 0.6136, 0.0054, 0.0656])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 999958
Density: 9.99958e-05
Time: 10.493394613265991 seconds
[39.01, 38.31, 38.71, 38.24, 38.61, 38.69, 38.97, 38.52, 39.54, 38.43]
[66.35]
10.457552671432495
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 7321, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 999958, 'MATRIX_DENSITY': 9.99958e-05, 'TIME_S': 10.493394613265991, 'TIME_S_1KI': 1.4333280444291752, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 693.8586197495459, 'W': 66.35}
[39.01, 38.31, 38.71, 38.24, 38.61, 38.69, 38.97, 38.52, 39.54, 38.43, 39.14, 38.72, 38.58, 38.9, 38.67, 38.68, 38.47, 38.41, 38.35, 38.47]
695.895
34.79475
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 7321, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 999958, 'MATRIX_DENSITY': 9.99958e-05, 'TIME_S': 10.493394613265991, 'TIME_S_1KI': 1.4333280444291752, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 693.8586197495459, 'W': 66.35, 'J_1KI': 94.77648132079578, 'W_1KI': 9.06296953968037, 'W_D': 31.555249999999994, 'J_D': 329.9906889352202, 'W_D_1KI': 4.310237672449118, 'J_D_1KI': 0.5887498528137027}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_100000_1e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 15459, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.295618772506714, "TIME_S_1KI": 0.6659951337412972, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 663.4292171406746, "W": 64.51, "J_1KI": 42.91540313996213, "W_1KI": 4.172973672294457, "W_D": 29.505250000000004, "J_D": 303.43582249325516, "W_D_1KI": 1.9086131056342588, "J_D_1KI": 0.12346290870264952}

67
pytorch/output_1core_after_test/epyc_7313p_10_10_10_100000_1e-05.output
View File

@ -1,67 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '100000', '-sd', '1e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 99999, "MATRIX_DENSITY": 9.9999e-06, "TIME_S": 0.6791770458221436}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 1, ..., 99996, 99996, 99999]),
col_indices=tensor([48006, 64298, 50858, ..., 16925, 31708, 60124]),
values=tensor([ 0.5949, -1.1126, -0.4425, ..., 1.9222, -0.2766,
-0.1611]), size=(100000, 100000), nnz=99999,
layout=torch.sparse_csr)
tensor([0.6661, 0.7299, 0.6911, ..., 0.4623, 0.9962, 0.3767])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 99999
Density: 9.9999e-06
Time: 0.6791770458221436 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '15459', '-ss', '100000', '-sd', '1e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.295618772506714}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 1, ..., 99998, 100000,
100000]),
col_indices=tensor([31661, 76136, 71092, ..., 68291, 34176, 79322]),
values=tensor([-1.4568, -0.5642, -0.1260, ..., -2.0915, -0.5754,
-0.9900]), size=(100000, 100000), nnz=100000,
layout=torch.sparse_csr)
tensor([0.6003, 0.7344, 0.3335, ..., 0.5656, 0.2704, 0.5992])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 10.295618772506714 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 1, ..., 99998, 100000,
100000]),
col_indices=tensor([31661, 76136, 71092, ..., 68291, 34176, 79322]),
values=tensor([-1.4568, -0.5642, -0.1260, ..., -2.0915, -0.5754,
-0.9900]), size=(100000, 100000), nnz=100000,
layout=torch.sparse_csr)
tensor([0.6003, 0.7344, 0.3335, ..., 0.5656, 0.2704, 0.5992])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 10.295618772506714 seconds
[39.16, 38.35, 38.36, 38.49, 38.33, 38.35, 38.75, 38.82, 38.39, 39.78]
[64.51]
10.284129858016968
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 15459, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.295618772506714, 'TIME_S_1KI': 0.6659951337412972, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 663.4292171406746, 'W': 64.51}
[39.16, 38.35, 38.36, 38.49, 38.33, 38.35, 38.75, 38.82, 38.39, 39.78, 39.65, 38.37, 38.32, 38.26, 38.42, 38.71, 38.38, 44.4, 38.96, 38.28]
700.095
35.00475
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 15459, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.295618772506714, 'TIME_S_1KI': 0.6659951337412972, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 663.4292171406746, 'W': 64.51, 'J_1KI': 42.91540313996213, 'W_1KI': 4.172973672294457, 'W_D': 29.505250000000004, 'J_D': 303.43582249325516, 'W_D_1KI': 1.9086131056342588, 'J_D_1KI': 0.12346290870264952}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_100000_2e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 12799, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 199999, "MATRIX_DENSITY": 1.99999e-05, "TIME_S": 10.312466621398926, "TIME_S_1KI": 0.8057244020156986, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 665.1102061700822, "W": 64.37, "J_1KI": 51.965794684747415, "W_1KI": 5.029299163997187, "W_D": 29.493500000000004, "J_D": 304.74487906908996, "W_D_1KI": 2.3043597156027817, "J_D_1KI": 0.1800421685758873}

68
pytorch/output_1core_after_test/epyc_7313p_10_10_10_100000_2e-05.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '100000', '-sd', '2e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 199996, "MATRIX_DENSITY": 1.99996e-05, "TIME_S": 0.8203706741333008}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 4, ..., 199991, 199992,
199996]),
col_indices=tensor([51426, 90007, 40378, ..., 18735, 37776, 48454]),
values=tensor([ 0.7391, 0.9740, -1.1861, ..., -0.5652, -0.6436,
1.0422]), size=(100000, 100000), nnz=199996,
layout=torch.sparse_csr)
tensor([0.7835, 0.3777, 0.7585, ..., 0.8549, 0.3936, 0.4815])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 199996
Density: 1.99996e-05
Time: 0.8203706741333008 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '12799', '-ss', '100000', '-sd', '2e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 199999, "MATRIX_DENSITY": 1.99999e-05, "TIME_S": 10.312466621398926}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 5, ..., 199994, 199998,
199999]),
col_indices=tensor([27560, 28667, 53651, ..., 54900, 68740, 23475]),
values=tensor([-1.6088, -0.0747, 0.2674, ..., 0.3290, 0.5072,
1.0750]), size=(100000, 100000), nnz=199999,
layout=torch.sparse_csr)
tensor([0.6315, 0.0366, 0.8205, ..., 0.3363, 0.5692, 0.3406])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 199999
Density: 1.99999e-05
Time: 10.312466621398926 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 5, ..., 199994, 199998,
199999]),
col_indices=tensor([27560, 28667, 53651, ..., 54900, 68740, 23475]),
values=tensor([-1.6088, -0.0747, 0.2674, ..., 0.3290, 0.5072,
1.0750]), size=(100000, 100000), nnz=199999,
layout=torch.sparse_csr)
tensor([0.6315, 0.0366, 0.8205, ..., 0.3363, 0.5692, 0.3406])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 199999
Density: 1.99999e-05
Time: 10.312466621398926 seconds
[39.81, 39.25, 38.39, 38.68, 38.87, 38.32, 38.84, 38.77, 38.55, 38.33]
[64.37]
10.332611560821533
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 12799, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 199999, 'MATRIX_DENSITY': 1.99999e-05, 'TIME_S': 10.312466621398926, 'TIME_S_1KI': 0.8057244020156986, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 665.1102061700822, 'W': 64.37}
[39.81, 39.25, 38.39, 38.68, 38.87, 38.32, 38.84, 38.77, 38.55, 38.33, 40.43, 38.95, 38.61, 38.52, 38.38, 39.43, 38.59, 38.26, 38.46, 38.75]
697.53
34.8765
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 12799, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 199999, 'MATRIX_DENSITY': 1.99999e-05, 'TIME_S': 10.312466621398926, 'TIME_S_1KI': 0.8057244020156986, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 665.1102061700822, 'W': 64.37, 'J_1KI': 51.965794684747415, 'W_1KI': 5.029299163997187, 'W_D': 29.493500000000004, 'J_D': 304.74487906908996, 'W_D_1KI': 2.3043597156027817, 'J_D_1KI': 0.1800421685758873}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_100000_5e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 9599, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 499988, "MATRIX_DENSITY": 4.99988e-05, "TIME_S": 10.298398733139038, "TIME_S_1KI": 1.0728616244545304, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 673.3587252640724, "W": 65.39, "J_1KI": 70.14884105261719, "W_1KI": 6.812167934159809, "W_D": 30.4705, "J_D": 313.7723969744444, "W_D_1KI": 3.1743410771955416, "J_D_1KI": 0.33069497626789685}

68
pytorch/output_1core_after_test/epyc_7313p_10_10_10_100000_5e-05.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '100000', '-sd', '5e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 499987, "MATRIX_DENSITY": 4.99987e-05, "TIME_S": 1.0938327312469482}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 6, 10, ..., 499975, 499979,
499987]),
col_indices=tensor([ 625, 1232, 18696, ..., 77518, 94690, 99471]),
values=tensor([-1.1636, 2.1655, -1.0596, ..., -1.6108, 0.9892,
-1.1686]), size=(100000, 100000), nnz=499987,
layout=torch.sparse_csr)
tensor([0.2570, 0.8095, 0.4051, ..., 0.4677, 0.3527, 0.8430])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 499987
Density: 4.99987e-05
Time: 1.0938327312469482 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '9599', '-ss', '100000', '-sd', '5e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 499988, "MATRIX_DENSITY": 4.99988e-05, "TIME_S": 10.298398733139038}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 4, ..., 499972, 499977,
499988]),
col_indices=tensor([37027, 6807, 46560, ..., 75712, 82456, 83079]),
values=tensor([ 1.4255, -0.1200, -0.1371, ..., 0.2939, 0.4596,
1.2418]), size=(100000, 100000), nnz=499988,
layout=torch.sparse_csr)
tensor([0.5521, 0.1482, 0.5901, ..., 0.2982, 0.5753, 0.2296])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 499988
Density: 4.99988e-05
Time: 10.298398733139038 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 4, ..., 499972, 499977,
499988]),
col_indices=tensor([37027, 6807, 46560, ..., 75712, 82456, 83079]),
values=tensor([ 1.4255, -0.1200, -0.1371, ..., 0.2939, 0.4596,
1.2418]), size=(100000, 100000), nnz=499988,
layout=torch.sparse_csr)
tensor([0.5521, 0.1482, 0.5901, ..., 0.2982, 0.5753, 0.2296])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 499988
Density: 4.99988e-05
Time: 10.298398733139038 seconds
[40.42, 38.28, 38.67, 38.64, 38.43, 39.44, 38.38, 38.76, 39.56, 38.41]
[65.39]
10.297579526901245
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 9599, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 499988, 'MATRIX_DENSITY': 4.99988e-05, 'TIME_S': 10.298398733139038, 'TIME_S_1KI': 1.0728616244545304, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 673.3587252640724, 'W': 65.39}
[40.42, 38.28, 38.67, 38.64, 38.43, 39.44, 38.38, 38.76, 39.56, 38.41, 39.54, 38.71, 39.2, 38.37, 38.42, 39.64, 38.35, 38.47, 38.46, 38.85]
698.39
34.9195
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 9599, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 499988, 'MATRIX_DENSITY': 4.99988e-05, 'TIME_S': 10.298398733139038, 'TIME_S_1KI': 1.0728616244545304, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 673.3587252640724, 'W': 65.39, 'J_1KI': 70.14884105261719, 'W_1KI': 6.812167934159809, 'W_D': 30.4705, 'J_D': 313.7723969744444, 'W_D_1KI': 3.1743410771955416, 'J_D_1KI': 0.33069497626789685}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_100000_8e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 7573, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 799974, "MATRIX_DENSITY": 7.99974e-05, "TIME_S": 10.329928636550903, "TIME_S_1KI": 1.364047093166632, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 680.7610593819618, "W": 65.53, "J_1KI": 89.89318095628704, "W_1KI": 8.653109731942427, "W_D": 30.261250000000004, "J_D": 314.3702213981748, "W_D_1KI": 3.9959395219860037, "J_D_1KI": 0.5276560837166253}

68
pytorch/output_1core_after_test/epyc_7313p_10_10_10_100000_8e-05.output
View File

@ -1,68 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '100000', '-sd', '8e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 799979, "MATRIX_DENSITY": 7.99979e-05, "TIME_S": 1.3864548206329346}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 10, 22, ..., 799960, 799969,
799979]),
col_indices=tensor([ 3991, 12470, 47738, ..., 59230, 62610, 86559]),
values=tensor([-1.5517, -1.1019, -2.2061, ..., 0.0714, 0.2519,
-0.0928]), size=(100000, 100000), nnz=799979,
layout=torch.sparse_csr)
tensor([0.8557, 0.9882, 0.2106, ..., 0.6867, 0.1131, 0.9591])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 799979
Density: 7.99979e-05
Time: 1.3864548206329346 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '7573', '-ss', '100000', '-sd', '8e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 799974, "MATRIX_DENSITY": 7.99974e-05, "TIME_S": 10.329928636550903}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 6, 18, ..., 799958, 799967,
799974]),
col_indices=tensor([22158, 27819, 31162, ..., 83457, 91150, 93673]),
values=tensor([ 1.7487, -0.8213, -0.1355, ..., -0.8810, -2.4345,
0.2948]), size=(100000, 100000), nnz=799974,
layout=torch.sparse_csr)
tensor([0.3306, 0.6421, 0.3776, ..., 0.4090, 0.4110, 0.5706])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 799974
Density: 7.99974e-05
Time: 10.329928636550903 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 6, 18, ..., 799958, 799967,
799974]),
col_indices=tensor([22158, 27819, 31162, ..., 83457, 91150, 93673]),
values=tensor([ 1.7487, -0.8213, -0.1355, ..., -0.8810, -2.4345,
0.2948]), size=(100000, 100000), nnz=799974,
layout=torch.sparse_csr)
tensor([0.3306, 0.6421, 0.3776, ..., 0.4090, 0.4110, 0.5706])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 799974
Density: 7.99974e-05
Time: 10.329928636550903 seconds
[39.09, 38.34, 38.83, 38.27, 38.53, 38.34, 38.66, 38.78, 39.1, 44.17]
[65.53]
10.388540506362915
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 7573, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 799974, 'MATRIX_DENSITY': 7.99974e-05, 'TIME_S': 10.329928636550903, 'TIME_S_1KI': 1.364047093166632, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 680.7610593819618, 'W': 65.53}
[39.09, 38.34, 38.83, 38.27, 38.53, 38.34, 38.66, 38.78, 39.1, 44.17, 39.27, 38.54, 38.41, 38.25, 38.45, 40.7, 44.33, 38.48, 38.64, 38.92]
705.375
35.26875
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 7573, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 799974, 'MATRIX_DENSITY': 7.99974e-05, 'TIME_S': 10.329928636550903, 'TIME_S_1KI': 1.364047093166632, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 680.7610593819618, 'W': 65.53, 'J_1KI': 89.89318095628704, 'W_1KI': 8.653109731942427, 'W_D': 30.261250000000004, 'J_D': 314.3702213981748, 'W_D_1KI': 3.9959395219860037, 'J_D_1KI': 0.5276560837166253}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_10000_0.0001.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 364192, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 9999, "MATRIX_DENSITY": 9.999e-05, "TIME_S": 10.286681890487671, "TIME_S_1KI": 0.028245216508016844, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 658.1757550382615, "W": 65.18, "J_1KI": 1.8072218913053046, "W_1KI": 0.17897153149986822, "W_D": 29.735250000000008, "J_D": 300.26113255602127, "W_D_1KI": 0.08164718060803094, "J_D_1KI": 0.00022418718864783123}

85
pytorch/output_1core_after_test/epyc_7313p_10_10_10_10000_0.0001.output
View File

@ -1,85 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '0.0001', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 9998, "MATRIX_DENSITY": 9.998e-05, "TIME_S": 0.03789472579956055}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 4, ..., 9993, 9997, 9998]),
col_indices=tensor([4827, 4877, 7090, ..., 3097, 4386, 1589]),
values=tensor([ 0.2815, -0.0621, 0.7820, ..., 0.1907, 0.2517,
-0.5782]), size=(10000, 10000), nnz=9998,
layout=torch.sparse_csr)
tensor([0.7159, 0.5102, 0.1780, ..., 0.6649, 0.0132, 0.6435])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 9998
Density: 9.998e-05
Time: 0.03789472579956055 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '277083', '-ss', '10000', '-sd', '0.0001', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 9999, "MATRIX_DENSITY": 9.999e-05, "TIME_S": 7.98856258392334}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 2, ..., 9998, 9998, 9999]),
col_indices=tensor([4687, 6305, 8321, ..., 5297, 8865, 3125]),
values=tensor([-0.2973, -0.7293, -0.4701, ..., 1.0040, -0.5152,
0.5670]), size=(10000, 10000), nnz=9999,
layout=torch.sparse_csr)
tensor([0.6850, 0.3394, 0.0913, ..., 0.6251, 0.5060, 0.1073])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 9999
Density: 9.999e-05
Time: 7.98856258392334 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '364192', '-ss', '10000', '-sd', '0.0001', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 9999, "MATRIX_DENSITY": 9.999e-05, "TIME_S": 10.286681890487671}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 2, ..., 9996, 9998, 9999]),
col_indices=tensor([9034, 7140, 1786, ..., 4605, 7715, 5729]),
values=tensor([-1.2303, 1.0912, 0.2060, ..., 0.0412, -0.6363,
-0.6436]), size=(10000, 10000), nnz=9999,
layout=torch.sparse_csr)
tensor([0.2105, 0.8829, 0.5834, ..., 0.8176, 0.5853, 0.6953])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 9999
Density: 9.999e-05
Time: 10.286681890487671 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 2, ..., 9996, 9998, 9999]),
col_indices=tensor([9034, 7140, 1786, ..., 4605, 7715, 5729]),
values=tensor([-1.2303, 1.0912, 0.2060, ..., 0.0412, -0.6363,
-0.6436]), size=(10000, 10000), nnz=9999,
layout=torch.sparse_csr)
tensor([0.2105, 0.8829, 0.5834, ..., 0.8176, 0.5853, 0.6953])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 9999
Density: 9.999e-05
Time: 10.286681890487671 seconds
[39.11, 38.65, 38.48, 38.91, 44.04, 38.52, 38.51, 38.72, 38.44, 38.98]
[65.18]
10.097817659378052
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 364192, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 9999, 'MATRIX_DENSITY': 9.999e-05, 'TIME_S': 10.286681890487671, 'TIME_S_1KI': 0.028245216508016844, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 658.1757550382615, 'W': 65.18}
[39.11, 38.65, 38.48, 38.91, 44.04, 38.52, 38.51, 38.72, 38.44, 38.98, 40.98, 44.91, 38.6, 38.78, 38.6, 38.41, 38.68, 38.86, 38.79, 38.92]
708.895
35.44475
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 364192, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 9999, 'MATRIX_DENSITY': 9.999e-05, 'TIME_S': 10.286681890487671, 'TIME_S_1KI': 0.028245216508016844, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 658.1757550382615, 'W': 65.18, 'J_1KI': 1.8072218913053046, 'W_1KI': 0.17897153149986822, 'W_D': 29.735250000000008, 'J_D': 300.26113255602127, 'W_D_1KI': 0.08164718060803094, 'J_D_1KI': 0.00022418718864783123}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_10000_1e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 687353, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000, "MATRIX_DENSITY": 1e-05, "TIME_S": 11.203821897506714, "TIME_S_1KI": 0.016299953440963688, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 712.5261029052734, "W": 65.11, "J_1KI": 1.036623253125066, "W_1KI": 0.09472570862424402, "W_D": 30.116, "J_D": 329.5720490722656, "W_D_1KI": 0.04381445923710233, "J_D_1KI": 6.374375209987056e-05}

1900
pytorch/output_1core_after_test/epyc_7313p_10_10_10_10000_1e-05.output
View File

File diff suppressed because it is too large Load Diff

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_10000_2e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 602748, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 2000, "MATRIX_DENSITY": 2e-05, "TIME_S": 10.21125841140747, "TIME_S_1KI": 0.01694117344463602, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 683.3000339078902, "W": 65.71, "J_1KI": 1.1336413126346172, "W_1KI": 0.10901736712523309, "W_D": 30.58899999999999, "J_D": 318.0865125126838, "W_D_1KI": 0.05074923516958993, "J_D_1KI": 8.41964389257035e-05}

85
pytorch/output_1core_after_test/epyc_7313p_10_10_10_10000_2e-05.output
View File

@ -1,85 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '2e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 2000, "MATRIX_DENSITY": 2e-05, "TIME_S": 0.02713489532470703}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 1, ..., 2000, 2000, 2000]),
col_indices=tensor([2645, 76, 1809, ..., 1614, 2006, 9458]),
values=tensor([ 1.3874, -1.1677, 0.9784, ..., -0.1842, 0.3648,
-1.9952]), size=(10000, 10000), nnz=2000,
layout=torch.sparse_csr)
tensor([0.1031, 0.9681, 0.6651, ..., 0.3559, 0.0936, 0.1162])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 2000
Density: 2e-05
Time: 0.02713489532470703 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '386955', '-ss', '10000', '-sd', '2e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 2000, "MATRIX_DENSITY": 2e-05, "TIME_S": 6.74083685874939}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 1, ..., 2000, 2000, 2000]),
col_indices=tensor([3749, 8011, 2966, ..., 9889, 3092, 195]),
values=tensor([-0.5969, -1.4126, 0.8624, ..., -0.7538, 0.1983,
1.5978]), size=(10000, 10000), nnz=2000,
layout=torch.sparse_csr)
tensor([0.3440, 0.9117, 0.4915, ..., 0.1931, 0.2897, 0.9406])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 2000
Density: 2e-05
Time: 6.74083685874939 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '602748', '-ss', '10000', '-sd', '2e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 2000, "MATRIX_DENSITY": 2e-05, "TIME_S": 10.21125841140747}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 2000, 2000, 2000]),
col_indices=tensor([1802, 2893, 8322, ..., 2363, 3238, 4274]),
values=tensor([ 0.0300, 1.2086, -0.2481, ..., -0.0804, 2.1056,
0.1582]), size=(10000, 10000), nnz=2000,
layout=torch.sparse_csr)
tensor([0.1979, 0.3419, 0.8009, ..., 0.6035, 0.7629, 0.5268])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 2000
Density: 2e-05
Time: 10.21125841140747 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 2000, 2000, 2000]),
col_indices=tensor([1802, 2893, 8322, ..., 2363, 3238, 4274]),
values=tensor([ 0.0300, 1.2086, -0.2481, ..., -0.0804, 2.1056,
0.1582]), size=(10000, 10000), nnz=2000,
layout=torch.sparse_csr)
tensor([0.1979, 0.3419, 0.8009, ..., 0.6035, 0.7629, 0.5268])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 2000
Density: 2e-05
Time: 10.21125841140747 seconds
[41.3, 38.81, 38.44, 38.39, 38.33, 38.28, 38.37, 38.72, 38.46, 38.78]
[65.71]
10.398722171783447
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 602748, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 2000, 'MATRIX_DENSITY': 2e-05, 'TIME_S': 10.21125841140747, 'TIME_S_1KI': 0.01694117344463602, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 683.3000339078902, 'W': 65.71}
[41.3, 38.81, 38.44, 38.39, 38.33, 38.28, 38.37, 38.72, 38.46, 38.78, 39.88, 38.49, 38.53, 38.42, 38.35, 38.28, 39.06, 44.36, 40.0, 38.3]
702.4200000000001
35.121
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 602748, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 2000, 'MATRIX_DENSITY': 2e-05, 'TIME_S': 10.21125841140747, 'TIME_S_1KI': 0.01694117344463602, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 683.3000339078902, 'W': 65.71, 'J_1KI': 1.1336413126346172, 'W_1KI': 0.10901736712523309, 'W_D': 30.58899999999999, 'J_D': 318.0865125126838, 'W_D_1KI': 0.05074923516958993, 'J_D_1KI': 8.41964389257035e-05}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_10000_5e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 475418, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000, "MATRIX_DENSITY": 5e-05, "TIME_S": 10.451735258102417, "TIME_S_1KI": 0.021984306984805826, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 688.6502168655395, "W": 65.24, "J_1KI": 1.4485152368348264, "W_1KI": 0.1372266090051281, "W_D": 30.341749999999998, "J_D": 320.27671240925787, "W_D_1KI": 0.06382120575998383, "J_D_1KI": 0.00013424229995495299}

85
pytorch/output_1core_after_test/epyc_7313p_10_10_10_10000_5e-05.output
View File

@ -1,85 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '5e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000, "MATRIX_DENSITY": 5e-05, "TIME_S": 0.032665252685546875}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 4998, 4998, 5000]),
col_indices=tensor([2543, 4228, 5675, ..., 7099, 979, 1021]),
values=tensor([ 2.5612, -1.4114, -1.2194, ..., -1.6806, 0.1446,
-0.8334]), size=(10000, 10000), nnz=5000,
layout=torch.sparse_csr)
tensor([0.1235, 0.4410, 0.8098, ..., 0.0872, 0.6747, 0.3389])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000
Density: 5e-05
Time: 0.032665252685546875 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '321442', '-ss', '10000', '-sd', '5e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000, "MATRIX_DENSITY": 5e-05, "TIME_S": 7.099309921264648}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 4, ..., 4999, 4999, 5000]),
col_indices=tensor([ 689, 2907, 3020, ..., 8328, 764, 7546]),
values=tensor([ 1.2282, 0.2524, -0.1503, ..., -1.6702, 1.0701,
-0.5727]), size=(10000, 10000), nnz=5000,
layout=torch.sparse_csr)
tensor([0.2869, 0.4983, 0.2994, ..., 0.7250, 0.9680, 0.2854])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000
Density: 5e-05
Time: 7.099309921264648 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '475418', '-ss', '10000', '-sd', '5e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000, "MATRIX_DENSITY": 5e-05, "TIME_S": 10.451735258102417}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 4, ..., 5000, 5000, 5000]),
col_indices=tensor([1429, 1621, 2379, ..., 8177, 7655, 3539]),
values=tensor([ 1.6185, -0.3081, 0.3132, ..., 0.9048, 0.9246,
0.1203]), size=(10000, 10000), nnz=5000,
layout=torch.sparse_csr)
tensor([0.9646, 0.4747, 0.7415, ..., 0.6425, 0.1934, 0.4010])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000
Density: 5e-05
Time: 10.451735258102417 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3, 4, ..., 5000, 5000, 5000]),
col_indices=tensor([1429, 1621, 2379, ..., 8177, 7655, 3539]),
values=tensor([ 1.6185, -0.3081, 0.3132, ..., 0.9048, 0.9246,
0.1203]), size=(10000, 10000), nnz=5000,
layout=torch.sparse_csr)
tensor([0.9646, 0.4747, 0.7415, ..., 0.6425, 0.1934, 0.4010])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000
Density: 5e-05
Time: 10.451735258102417 seconds
[39.03, 39.16, 38.53, 39.16, 38.96, 38.83, 39.53, 38.34, 38.57, 38.38]
[65.24]
10.555644035339355
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 475418, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000, 'MATRIX_DENSITY': 5e-05, 'TIME_S': 10.451735258102417, 'TIME_S_1KI': 0.021984306984805826, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 688.6502168655395, 'W': 65.24}
[39.03, 39.16, 38.53, 39.16, 38.96, 38.83, 39.53, 38.34, 38.57, 38.38, 39.85, 38.31, 39.06, 39.04, 38.89, 38.51, 38.44, 38.39, 38.43, 38.37]
697.9649999999999
34.89825
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 475418, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000, 'MATRIX_DENSITY': 5e-05, 'TIME_S': 10.451735258102417, 'TIME_S_1KI': 0.021984306984805826, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 688.6502168655395, 'W': 65.24, 'J_1KI': 1.4485152368348264, 'W_1KI': 0.1372266090051281, 'W_D': 30.341749999999998, 'J_D': 320.27671240925787, 'W_D_1KI': 0.06382120575998383, 'J_D_1KI': 0.00013424229995495299}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_10000_8e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 399296, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 7999, "MATRIX_DENSITY": 7.999e-05, "TIME_S": 10.458914041519165, "TIME_S_1KI": 0.026193385462211404, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 673.009458374977, "W": 65.55, "J_1KI": 1.6854901085284526, "W_1KI": 0.16416392851418496, "W_D": 30.665499999999994, "J_D": 314.8462478382587, "W_D_1KI": 0.07679891609232248, "J_D_1KI": 0.00019233580124099032}

85
pytorch/output_1core_after_test/epyc_7313p_10_10_10_10000_8e-05.output
View File

@ -1,85 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '8e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 8000, "MATRIX_DENSITY": 8e-05, "TIME_S": 0.03641247749328613}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 3, ..., 7998, 7999, 8000]),
col_indices=tensor([6117, 8477, 6510, ..., 3404, 5465, 8467]),
values=tensor([-1.6081, -0.0719, -2.1307, ..., 0.1441, 0.1036,
0.0623]), size=(10000, 10000), nnz=8000,
layout=torch.sparse_csr)
tensor([0.8595, 0.7644, 0.3531, ..., 0.0315, 0.6130, 0.2786])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 8000
Density: 8e-05
Time: 0.03641247749328613 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '288362', '-ss', '10000', '-sd', '8e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 7999, "MATRIX_DENSITY": 7.999e-05, "TIME_S": 7.582841157913208}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 7998, 7998, 7999]),
col_indices=tensor([7743, 8729, 4527, ..., 5020, 3758, 9585]),
values=tensor([-0.7905, 0.7067, -0.3667, ..., -1.9197, 0.6727,
-0.2685]), size=(10000, 10000), nnz=7999,
layout=torch.sparse_csr)
tensor([0.2965, 0.4690, 0.6034, ..., 0.9291, 0.5376, 0.8914])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 7999
Density: 7.999e-05
Time: 7.582841157913208 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '399296', '-ss', '10000', '-sd', '8e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 7999, "MATRIX_DENSITY": 7.999e-05, "TIME_S": 10.458914041519165}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 7998, 7998, 7999]),
col_indices=tensor([4248, 4898, 9130, ..., 2629, 4508, 4391]),
values=tensor([-0.1804, -0.2691, 0.3496, ..., -0.6907, 1.8081,
-1.1816]), size=(10000, 10000), nnz=7999,
layout=torch.sparse_csr)
tensor([0.3531, 0.0473, 0.4264, ..., 0.6320, 0.2793, 0.8248])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 7999
Density: 7.999e-05
Time: 10.458914041519165 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 0, ..., 7998, 7998, 7999]),
col_indices=tensor([4248, 4898, 9130, ..., 2629, 4508, 4391]),
values=tensor([-0.1804, -0.2691, 0.3496, ..., -0.6907, 1.8081,
-1.1816]), size=(10000, 10000), nnz=7999,
layout=torch.sparse_csr)
tensor([0.3531, 0.0473, 0.4264, ..., 0.6320, 0.2793, 0.8248])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 7999
Density: 7.999e-05
Time: 10.458914041519165 seconds
[39.14, 38.74, 38.55, 38.47, 38.62, 38.75, 38.59, 38.85, 38.96, 38.38]
[65.55]
10.267116069793701
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 399296, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 7999, 'MATRIX_DENSITY': 7.999e-05, 'TIME_S': 10.458914041519165, 'TIME_S_1KI': 0.026193385462211404, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 673.009458374977, 'W': 65.55}
[39.14, 38.74, 38.55, 38.47, 38.62, 38.75, 38.59, 38.85, 38.96, 38.38, 39.03, 38.81, 38.52, 39.68, 38.42, 38.47, 39.2, 38.75, 38.86, 38.35]
697.69
34.8845
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 399296, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 7999, 'MATRIX_DENSITY': 7.999e-05, 'TIME_S': 10.458914041519165, 'TIME_S_1KI': 0.026193385462211404, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 673.009458374977, 'W': 65.55, 'J_1KI': 1.6854901085284526, 'W_1KI': 0.16416392851418496, 'W_D': 30.665499999999994, 'J_D': 314.8462478382587, 'W_D_1KI': 0.07679891609232248, 'J_D_1KI': 0.00019233580124099032}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_150000_0.0001.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 3623, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 2249858, "MATRIX_DENSITY": 9.999368888888889e-05, "TIME_S": 10.667315244674683, "TIME_S_1KI": 2.944332112800078, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 750.546510848999, "W": 70.88, "J_1KI": 207.16160939801242, "W_1KI": 19.563897322660775, "W_D": 35.933749999999996, "J_D": 380.5015615719556, "W_D_1KI": 9.918230747998896, "J_D_1KI": 2.737574040297791}

71
pytorch/output_1core_after_test/epyc_7313p_10_10_10_150000_0.0001.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '150000', '-sd', '0.0001', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 2249884, "MATRIX_DENSITY": 9.999484444444444e-05, "TIME_S": 2.8976516723632812}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 22, 43, ..., 2249846,
2249863, 2249884]),
col_indices=tensor([ 2507, 16314, 31317, ..., 120903, 121359,
147768]),
values=tensor([-0.6085, -0.7004, 0.1228, ..., 0.9020, -0.4601,
-1.0639]), size=(150000, 150000), nnz=2249884,
layout=torch.sparse_csr)
tensor([0.3195, 0.5583, 0.9597, ..., 0.5573, 0.0634, 0.8941])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 2249884
Density: 9.999484444444444e-05
Time: 2.8976516723632812 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '3623', '-ss', '150000', '-sd', '0.0001', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 2249858, "MATRIX_DENSITY": 9.999368888888889e-05, "TIME_S": 10.667315244674683}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 11, 24, ..., 2249832,
2249846, 2249858]),
col_indices=tensor([ 10890, 12729, 17252, ..., 102978, 126802,
132653]),
values=tensor([ 1.4097, 0.2679, 0.6261, ..., 1.5911, 1.7075,
-0.0145]), size=(150000, 150000), nnz=2249858,
layout=torch.sparse_csr)
tensor([0.0131, 0.1649, 0.4269, ..., 0.2547, 0.5949, 0.0782])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 2249858
Density: 9.999368888888889e-05
Time: 10.667315244674683 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 11, 24, ..., 2249832,
2249846, 2249858]),
col_indices=tensor([ 10890, 12729, 17252, ..., 102978, 126802,
132653]),
values=tensor([ 1.4097, 0.2679, 0.6261, ..., 1.5911, 1.7075,
-0.0145]), size=(150000, 150000), nnz=2249858,
layout=torch.sparse_csr)
tensor([0.0131, 0.1649, 0.4269, ..., 0.2547, 0.5949, 0.0782])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 2249858
Density: 9.999368888888889e-05
Time: 10.667315244674683 seconds
[39.59, 39.85, 38.38, 38.2, 38.69, 38.44, 38.21, 38.33, 38.22, 38.26]
[70.88]
10.588974475860596
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 3623, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 2249858, 'MATRIX_DENSITY': 9.999368888888889e-05, 'TIME_S': 10.667315244674683, 'TIME_S_1KI': 2.944332112800078, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 750.546510848999, 'W': 70.88}
[39.59, 39.85, 38.38, 38.2, 38.69, 38.44, 38.21, 38.33, 38.22, 38.26, 39.35, 38.9, 38.85, 38.27, 38.23, 38.22, 38.53, 38.31, 43.38, 38.63]
698.925
34.94625
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 3623, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 2249858, 'MATRIX_DENSITY': 9.999368888888889e-05, 'TIME_S': 10.667315244674683, 'TIME_S_1KI': 2.944332112800078, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 750.546510848999, 'W': 70.88, 'J_1KI': 207.16160939801242, 'W_1KI': 19.563897322660775, 'W_D': 35.933749999999996, 'J_D': 380.5015615719556, 'W_D_1KI': 9.918230747998896, 'J_D_1KI': 2.737574040297791}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_150000_1e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 9166, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 224999, "MATRIX_DENSITY": 9.999955555555555e-06, "TIME_S": 10.329043865203857, "TIME_S_1KI": 1.1268867406942895, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 669.3000064229966, "W": 64.62, "J_1KI": 73.01985669026801, "W_1KI": 7.049967270346935, "W_D": 29.8575, "J_D": 309.24829683959484, "W_D_1KI": 3.2574187213615535, "J_D_1KI": 0.35538061546602157}

71
pytorch/output_1core_after_test/epyc_7313p_10_10_10_150000_1e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '150000', '-sd', '1e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 224998, "MATRIX_DENSITY": 9.999911111111111e-06, "TIME_S": 1.1455013751983643}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 4, ..., 224998, 224998,
224998]),
col_indices=tensor([100518, 131563, 9790, ..., 76958, 129090,
127826]),
values=tensor([-0.9354, 0.0861, 0.0469, ..., -0.0733, -0.3369,
-0.3156]), size=(150000, 150000), nnz=224998,
layout=torch.sparse_csr)
tensor([0.7914, 0.1064, 0.9881, ..., 0.4061, 0.8175, 0.2421])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 224998
Density: 9.999911111111111e-06
Time: 1.1455013751983643 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '9166', '-ss', '150000', '-sd', '1e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 224999, "MATRIX_DENSITY": 9.999955555555555e-06, "TIME_S": 10.329043865203857}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 1, ..., 224996, 224996,
224999]),
col_indices=tensor([101672, 82567, 101421, ..., 14061, 17263,
44668]),
values=tensor([-1.0159, 1.4417, -1.5888, ..., 0.7553, -0.8014,
-0.0962]), size=(150000, 150000), nnz=224999,
layout=torch.sparse_csr)
tensor([0.8156, 0.5997, 0.6168, ..., 0.9317, 0.7110, 0.6190])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 224999
Density: 9.999955555555555e-06
Time: 10.329043865203857 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 0, 1, ..., 224996, 224996,
224999]),
col_indices=tensor([101672, 82567, 101421, ..., 14061, 17263,
44668]),
values=tensor([-1.0159, 1.4417, -1.5888, ..., 0.7553, -0.8014,
-0.0962]), size=(150000, 150000), nnz=224999,
layout=torch.sparse_csr)
tensor([0.8156, 0.5997, 0.6168, ..., 0.9317, 0.7110, 0.6190])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 224999
Density: 9.999955555555555e-06
Time: 10.329043865203857 seconds
[39.27, 38.75, 38.65, 38.37, 38.39, 38.52, 38.44, 38.75, 38.32, 38.42]
[64.62]
10.357474565505981
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 9166, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 224999, 'MATRIX_DENSITY': 9.999955555555555e-06, 'TIME_S': 10.329043865203857, 'TIME_S_1KI': 1.1268867406942895, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 669.3000064229966, 'W': 64.62}
[39.27, 38.75, 38.65, 38.37, 38.39, 38.52, 38.44, 38.75, 38.32, 38.42, 39.1, 38.8, 39.12, 38.37, 38.31, 39.57, 38.4, 38.47, 38.33, 38.59]
695.25
34.7625
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 9166, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 224999, 'MATRIX_DENSITY': 9.999955555555555e-06, 'TIME_S': 10.329043865203857, 'TIME_S_1KI': 1.1268867406942895, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 669.3000064229966, 'W': 64.62, 'J_1KI': 73.01985669026801, 'W_1KI': 7.049967270346935, 'W_D': 29.8575, 'J_D': 309.24829683959484, 'W_D_1KI': 3.2574187213615535, 'J_D_1KI': 0.35538061546602157}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_150000_2e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 6901, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 449998, "MATRIX_DENSITY": 1.999991111111111e-05, "TIME_S": 10.454267024993896, "TIME_S_1KI": 1.5148916135333859, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 675.2335901641846, "W": 64.84, "J_1KI": 97.845760058569, "W_1KI": 9.39573974786263, "W_D": 29.732999999999997, "J_D": 309.63479852485654, "W_D_1KI": 4.308506013621214, "J_D_1KI": 0.6243306786873227}

71
pytorch/output_1core_after_test/epyc_7313p_10_10_10_150000_2e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '150000', '-sd', '2e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 449999, "MATRIX_DENSITY": 1.9999955555555556e-05, "TIME_S": 1.5213685035705566}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 7, 12, ..., 449995, 449998,
449999]),
col_indices=tensor([ 8360, 33252, 44362, ..., 21408, 124412,
124334]),
values=tensor([-0.5427, 0.0515, 0.0818, ..., -1.2949, 1.3790,
0.6657]), size=(150000, 150000), nnz=449999,
layout=torch.sparse_csr)
tensor([0.2708, 0.0613, 0.1710, ..., 0.9721, 0.9560, 0.0829])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 449999
Density: 1.9999955555555556e-05
Time: 1.5213685035705566 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '6901', '-ss', '150000', '-sd', '2e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 449998, "MATRIX_DENSITY": 1.999991111111111e-05, "TIME_S": 10.454267024993896}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 5, 9, ..., 449993, 449994,
449998]),
col_indices=tensor([ 20846, 26091, 54441, ..., 30732, 48515,
104522]),
values=tensor([-1.2604, -0.1905, -0.1295, ..., 0.2361, 0.1736,
-0.7596]), size=(150000, 150000), nnz=449998,
layout=torch.sparse_csr)
tensor([0.1571, 0.6178, 0.3753, ..., 0.9438, 0.5462, 0.3709])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 449998
Density: 1.999991111111111e-05
Time: 10.454267024993896 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 5, 9, ..., 449993, 449994,
449998]),
col_indices=tensor([ 20846, 26091, 54441, ..., 30732, 48515,
104522]),
values=tensor([-1.2604, -0.1905, -0.1295, ..., 0.2361, 0.1736,
-0.7596]), size=(150000, 150000), nnz=449998,
layout=torch.sparse_csr)
tensor([0.1571, 0.6178, 0.3753, ..., 0.9438, 0.5462, 0.3709])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 449998
Density: 1.999991111111111e-05
Time: 10.454267024993896 seconds
[40.27, 38.44, 38.96, 38.34, 38.54, 38.31, 38.38, 38.3, 38.55, 45.52]
[64.84]
10.413843154907227
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 6901, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 449998, 'MATRIX_DENSITY': 1.999991111111111e-05, 'TIME_S': 10.454267024993896, 'TIME_S_1KI': 1.5148916135333859, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 675.2335901641846, 'W': 64.84}
[40.27, 38.44, 38.96, 38.34, 38.54, 38.31, 38.38, 38.3, 38.55, 45.52, 39.02, 38.38, 38.38, 38.52, 38.37, 38.29, 43.89, 38.47, 38.4, 38.43]
702.1400000000001
35.107000000000006
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 6901, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 449998, 'MATRIX_DENSITY': 1.999991111111111e-05, 'TIME_S': 10.454267024993896, 'TIME_S_1KI': 1.5148916135333859, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 675.2335901641846, 'W': 64.84, 'J_1KI': 97.845760058569, 'W_1KI': 9.39573974786263, 'W_D': 29.732999999999997, 'J_D': 309.63479852485654, 'W_D_1KI': 4.308506013621214, 'J_D_1KI': 0.6243306786873227}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_150000_5e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 5109, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1124975, "MATRIX_DENSITY": 4.999888888888889e-05, "TIME_S": 10.481255531311035, "TIME_S_1KI": 2.051527800217466, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 704.4864772510529, "W": 66.92, "J_1KI": 137.89126585458072, "W_1KI": 13.098453709140731, "W_D": 32.048, "J_D": 337.37870028305053, "W_D_1KI": 6.272851830103739, "J_D_1KI": 1.2278042337255313}

71
pytorch/output_1core_after_test/epyc_7313p_10_10_10_150000_5e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '150000', '-sd', '5e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1124977, "MATRIX_DENSITY": 4.9998977777777776e-05, "TIME_S": 2.0548505783081055}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 10, 18, ..., 1124965,
1124969, 1124977]),
col_indices=tensor([ 31453, 47537, 66534, ..., 102759, 106663,
136823]),
values=tensor([ 1.5868, -0.7934, 0.7941, ..., -1.6063, 2.0651,
0.5690]), size=(150000, 150000), nnz=1124977,
layout=torch.sparse_csr)
tensor([0.6210, 0.5889, 0.5144, ..., 0.1174, 0.3534, 0.8613])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1124977
Density: 4.9998977777777776e-05
Time: 2.0548505783081055 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '5109', '-ss', '150000', '-sd', '5e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1124975, "MATRIX_DENSITY": 4.999888888888889e-05, "TIME_S": 10.481255531311035}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 7, 17, ..., 1124954,
1124964, 1124975]),
col_indices=tensor([ 13599, 24811, 30718, ..., 135710, 137278,
148349]),
values=tensor([ 1.1951, -1.4782, 1.4832, ..., 1.3639, 0.0995,
0.3762]), size=(150000, 150000), nnz=1124975,
layout=torch.sparse_csr)
tensor([0.0815, 0.9569, 0.3513, ..., 0.2811, 0.4692, 0.1935])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1124975
Density: 4.999888888888889e-05
Time: 10.481255531311035 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 7, 17, ..., 1124954,
1124964, 1124975]),
col_indices=tensor([ 13599, 24811, 30718, ..., 135710, 137278,
148349]),
values=tensor([ 1.1951, -1.4782, 1.4832, ..., 1.3639, 0.0995,
0.3762]), size=(150000, 150000), nnz=1124975,
layout=torch.sparse_csr)
tensor([0.0815, 0.9569, 0.3513, ..., 0.2811, 0.4692, 0.1935])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1124975
Density: 4.999888888888889e-05
Time: 10.481255531311035 seconds
[39.89, 38.34, 38.79, 38.32, 38.57, 38.23, 38.56, 38.9, 38.74, 38.66]
[66.92]
10.52729344367981
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 5109, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 1124975, 'MATRIX_DENSITY': 4.999888888888889e-05, 'TIME_S': 10.481255531311035, 'TIME_S_1KI': 2.051527800217466, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 704.4864772510529, 'W': 66.92}
[39.89, 38.34, 38.79, 38.32, 38.57, 38.23, 38.56, 38.9, 38.74, 38.66, 39.47, 40.06, 38.45, 38.3, 38.41, 39.98, 38.88, 38.28, 38.36, 38.52]
697.44
34.872
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 5109, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 1124975, 'MATRIX_DENSITY': 4.999888888888889e-05, 'TIME_S': 10.481255531311035, 'TIME_S_1KI': 2.051527800217466, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 704.4864772510529, 'W': 66.92, 'J_1KI': 137.89126585458072, 'W_1KI': 13.098453709140731, 'W_D': 32.048, 'J_D': 337.37870028305053, 'W_D_1KI': 6.272851830103739, 'J_D_1KI': 1.2278042337255313}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_150000_8e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 4220, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1799931, "MATRIX_DENSITY": 7.999693333333333e-05, "TIME_S": 10.419165134429932, "TIME_S_1KI": 2.4689964773530644, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 692.4653435611725, "W": 67.59, "J_1KI": 164.0913136400883, "W_1KI": 16.016587677725116, "W_D": 32.66975000000001, "J_D": 334.70438907837877, "W_D_1KI": 7.741646919431282, "J_D_1KI": 1.8345134880168914}

71
pytorch/output_1core_after_test/epyc_7313p_10_10_10_150000_8e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '150000', '-sd', '8e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1799939, "MATRIX_DENSITY": 7.999728888888888e-05, "TIME_S": 2.488084554672241}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 12, 22, ..., 1799913,
1799924, 1799939]),
col_indices=tensor([ 13746, 15057, 18265, ..., 123846, 124411,
145916]),
values=tensor([-0.7466, 0.0637, -0.8689, ..., -0.5743, -0.3689,
0.2622]), size=(150000, 150000), nnz=1799939,
layout=torch.sparse_csr)
tensor([0.9366, 0.5730, 0.1137, ..., 0.8382, 0.9191, 0.9155])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1799939
Density: 7.999728888888888e-05
Time: 2.488084554672241 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '4220', '-ss', '150000', '-sd', '8e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 8e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [150000, 150000], "MATRIX_ROWS": 150000, "MATRIX_SIZE": 22500000000, "MATRIX_NNZ": 1799931, "MATRIX_DENSITY": 7.999693333333333e-05, "TIME_S": 10.419165134429932}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 11, 23, ..., 1799907,
1799918, 1799931]),
col_indices=tensor([ 5037, 12265, 35290, ..., 122025, 127242,
133587]),
values=tensor([-0.8165, -0.4506, -1.1214, ..., 0.3012, 0.9164,
0.9097]), size=(150000, 150000), nnz=1799931,
layout=torch.sparse_csr)
tensor([0.5758, 0.1969, 0.5929, ..., 0.3681, 0.1106, 0.1361])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1799931
Density: 7.999693333333333e-05
Time: 10.419165134429932 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 11, 23, ..., 1799907,
1799918, 1799931]),
col_indices=tensor([ 5037, 12265, 35290, ..., 122025, 127242,
133587]),
values=tensor([-0.8165, -0.4506, -1.1214, ..., 0.3012, 0.9164,
0.9097]), size=(150000, 150000), nnz=1799931,
layout=torch.sparse_csr)
tensor([0.5758, 0.1969, 0.5929, ..., 0.3681, 0.1106, 0.1361])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([150000, 150000])
Rows: 150000
Size: 22500000000
NNZ: 1799931
Density: 7.999693333333333e-05
Time: 10.419165134429932 seconds
[44.49, 38.32, 38.33, 39.2, 38.32, 39.11, 38.45, 38.58, 38.73, 38.29]
[67.59]
10.245085716247559
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 4220, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 1799931, 'MATRIX_DENSITY': 7.999693333333333e-05, 'TIME_S': 10.419165134429932, 'TIME_S_1KI': 2.4689964773530644, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 692.4653435611725, 'W': 67.59}
[44.49, 38.32, 38.33, 39.2, 38.32, 39.11, 38.45, 38.58, 38.73, 38.29, 39.26, 38.32, 38.8, 38.76, 38.46, 39.62, 38.37, 38.42, 38.47, 38.25]
698.405
34.920249999999996
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 4220, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 8e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [150000, 150000], 'MATRIX_ROWS': 150000, 'MATRIX_SIZE': 22500000000, 'MATRIX_NNZ': 1799931, 'MATRIX_DENSITY': 7.999693333333333e-05, 'TIME_S': 10.419165134429932, 'TIME_S_1KI': 2.4689964773530644, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 692.4653435611725, 'W': 67.59, 'J_1KI': 164.0913136400883, 'W_1KI': 16.016587677725116, 'W_D': 32.66975000000001, 'J_D': 334.70438907837877, 'W_D_1KI': 7.741646919431282, 'J_D_1KI': 1.8345134880168914}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_200000_0.0001.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 2126, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 3999815, "MATRIX_DENSITY": 9.9995375e-05, "TIME_S": 10.061469793319702, "TIME_S_1KI": 4.732582216989512, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 814.6817216920853, "W": 76.98, "J_1KI": 383.19930465290935, "W_1KI": 36.20884289746002, "W_D": 42.339000000000006, "J_D": 448.07494693064695, "W_D_1KI": 19.914863593603013, "J_D_1KI": 9.367292377047512}

71
pytorch/output_1core_after_test/epyc_7313p_10_10_10_200000_0.0001.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '200000', '-sd', '0.0001', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 3999788, "MATRIX_DENSITY": 9.99947e-05, "TIME_S": 4.937520503997803}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 17, 37, ..., 3999749,
3999767, 3999788]),
col_indices=tensor([ 1100, 11052, 12103, ..., 167542, 179467,
199307]),
values=tensor([ 0.5475, -1.1224, 0.0722, ..., -0.1144, 1.0163,
0.7412]), size=(200000, 200000), nnz=3999788,
layout=torch.sparse_csr)
tensor([0.8210, 0.9181, 0.4540, ..., 0.0365, 0.2540, 0.3511])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 3999788
Density: 9.99947e-05
Time: 4.937520503997803 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '2126', '-ss', '200000', '-sd', '0.0001', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 0.0001, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 3999815, "MATRIX_DENSITY": 9.9995375e-05, "TIME_S": 10.061469793319702}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 17, 34, ..., 3999773,
3999791, 3999815]),
col_indices=tensor([ 25603, 32992, 34253, ..., 185349, 188179,
193803]),
values=tensor([-0.5584, -0.3177, -0.8346, ..., -0.6017, 0.3720,
0.8986]), size=(200000, 200000), nnz=3999815,
layout=torch.sparse_csr)
tensor([0.9144, 0.6087, 0.6108, ..., 0.3591, 0.6548, 0.2005])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 3999815
Density: 9.9995375e-05
Time: 10.061469793319702 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 17, 34, ..., 3999773,
3999791, 3999815]),
col_indices=tensor([ 25603, 32992, 34253, ..., 185349, 188179,
193803]),
values=tensor([-0.5584, -0.3177, -0.8346, ..., -0.6017, 0.3720,
0.8986]), size=(200000, 200000), nnz=3999815,
layout=torch.sparse_csr)
tensor([0.9144, 0.6087, 0.6108, ..., 0.3591, 0.6548, 0.2005])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 3999815
Density: 9.9995375e-05
Time: 10.061469793319702 seconds
[38.95, 38.12, 38.27, 38.18, 38.23, 38.11, 38.27, 39.77, 38.33, 38.11]
[76.98]
10.583030939102173
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 2126, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 3999815, 'MATRIX_DENSITY': 9.9995375e-05, 'TIME_S': 10.061469793319702, 'TIME_S_1KI': 4.732582216989512, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 814.6817216920853, 'W': 76.98}
[38.95, 38.12, 38.27, 38.18, 38.23, 38.11, 38.27, 39.77, 38.33, 38.11, 39.73, 38.91, 38.21, 38.63, 38.46, 38.62, 38.71, 38.18, 38.28, 38.29]
692.8199999999999
34.641
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 2126, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 0.0001, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 3999815, 'MATRIX_DENSITY': 9.9995375e-05, 'TIME_S': 10.061469793319702, 'TIME_S_1KI': 4.732582216989512, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 814.6817216920853, 'W': 76.98, 'J_1KI': 383.19930465290935, 'W_1KI': 36.20884289746002, 'W_D': 42.339000000000006, 'J_D': 448.07494693064695, 'W_D_1KI': 19.914863593603013, 'J_D_1KI': 9.367292377047512}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_200000_1e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 6335, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 399999, "MATRIX_DENSITY": 9.999975e-06, "TIME_S": 10.42294692993164, "TIME_S_1KI": 1.6452954901233845, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 677.8716199588775, "W": 64.66, "J_1KI": 107.00420204560024, "W_1KI": 10.20678768745067, "W_D": 29.919749999999993, "J_D": 313.66763688933844, "W_D_1KI": 4.7229281767955795, "J_D_1KI": 0.7455293096757032}

71
pytorch/output_1core_after_test/epyc_7313p_10_10_10_200000_1e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '200000', '-sd', '1e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 399998, "MATRIX_DENSITY": 9.99995e-06, "TIME_S": 1.6573054790496826}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1, 3, ..., 399994, 399997,
399998]),
col_indices=tensor([ 66300, 2284, 53244, ..., 49679, 62137,
168627]),
values=tensor([ 0.5044, -0.0503, -2.0900, ..., 0.4461, -0.3815,
-1.5372]), size=(200000, 200000), nnz=399998,
layout=torch.sparse_csr)
tensor([0.1588, 0.0055, 0.7727, ..., 0.8522, 0.5649, 0.6738])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 399998
Density: 9.99995e-06
Time: 1.6573054790496826 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '6335', '-ss', '200000', '-sd', '1e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 1e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 399999, "MATRIX_DENSITY": 9.999975e-06, "TIME_S": 10.42294692993164}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 5, ..., 399991, 399995,
399999]),
col_indices=tensor([ 6449, 13437, 68699, ..., 173042, 178967,
192775]),
values=tensor([ 1.5716, 0.2369, 0.7778, ..., 0.5457, 0.4701,
-0.8057]), size=(200000, 200000), nnz=399999,
layout=torch.sparse_csr)
tensor([0.7762, 0.0703, 0.2592, ..., 0.1464, 0.7439, 0.6172])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 399999
Density: 9.999975e-06
Time: 10.42294692993164 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 4, 5, ..., 399991, 399995,
399999]),
col_indices=tensor([ 6449, 13437, 68699, ..., 173042, 178967,
192775]),
values=tensor([ 1.5716, 0.2369, 0.7778, ..., 0.5457, 0.4701,
-0.8057]), size=(200000, 200000), nnz=399999,
layout=torch.sparse_csr)
tensor([0.7762, 0.0703, 0.2592, ..., 0.1464, 0.7439, 0.6172])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 399999
Density: 9.999975e-06
Time: 10.42294692993164 seconds
[39.04, 38.53, 39.14, 38.5, 38.68, 38.37, 38.33, 38.63, 38.3, 40.39]
[64.66]
10.483631610870361
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 6335, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 399999, 'MATRIX_DENSITY': 9.999975e-06, 'TIME_S': 10.42294692993164, 'TIME_S_1KI': 1.6452954901233845, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 677.8716199588775, 'W': 64.66}
[39.04, 38.53, 39.14, 38.5, 38.68, 38.37, 38.33, 38.63, 38.3, 40.39, 39.02, 38.55, 38.31, 38.69, 38.65, 38.25, 38.41, 38.23, 38.67, 38.68]
694.8050000000001
34.74025
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 6335, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 1e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 399999, 'MATRIX_DENSITY': 9.999975e-06, 'TIME_S': 10.42294692993164, 'TIME_S_1KI': 1.6452954901233845, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 677.8716199588775, 'W': 64.66, 'J_1KI': 107.00420204560024, 'W_1KI': 10.20678768745067, 'W_D': 29.919749999999993, 'J_D': 313.66763688933844, 'W_D_1KI': 4.7229281767955795, 'J_D_1KI': 0.7455293096757032}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_200000_2e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 4694, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 799988, "MATRIX_DENSITY": 1.99997e-05, "TIME_S": 10.433407306671143, "TIME_S_1KI": 2.2227113989499663, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 685.4626195144654, "W": 65.56, "J_1KI": 146.02953121313706, "W_1KI": 13.966766084363018, "W_D": 30.37675, "J_D": 317.60412793374064, "W_D_1KI": 6.471399659139327, "J_D_1KI": 1.3786535277246117}

71
pytorch/output_1core_after_test/epyc_7313p_10_10_10_200000_2e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '200000', '-sd', '2e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 799991, "MATRIX_DENSITY": 1.9999775e-05, "TIME_S": 2.2366011142730713}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2, 5, ..., 799986, 799987,
799991]),
col_indices=tensor([150067, 181412, 47629, ..., 70392, 74082,
103785]),
values=tensor([ 1.4929, -0.1532, -0.9013, ..., -0.6923, 0.5828,
-0.1352]), size=(200000, 200000), nnz=799991,
layout=torch.sparse_csr)
tensor([0.5880, 0.2805, 0.9130, ..., 0.2024, 0.8848, 0.8005])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 799991
Density: 1.9999775e-05
Time: 2.2366011142730713 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '4694', '-ss', '200000', '-sd', '2e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 2e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 799988, "MATRIX_DENSITY": 1.99997e-05, "TIME_S": 10.433407306671143}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 8, 10, ..., 799981, 799982,
799988]),
col_indices=tensor([ 26638, 34068, 51464, ..., 90655, 104981,
178084]),
values=tensor([ 0.5771, -0.4861, -1.4112, ..., 1.0374, 0.9570,
0.9346]), size=(200000, 200000), nnz=799988,
layout=torch.sparse_csr)
tensor([0.0015, 0.2894, 0.6814, ..., 0.9382, 0.9968, 0.5782])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 799988
Density: 1.99997e-05
Time: 10.433407306671143 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 8, 10, ..., 799981, 799982,
799988]),
col_indices=tensor([ 26638, 34068, 51464, ..., 90655, 104981,
178084]),
values=tensor([ 0.5771, -0.4861, -1.4112, ..., 1.0374, 0.9570,
0.9346]), size=(200000, 200000), nnz=799988,
layout=torch.sparse_csr)
tensor([0.0015, 0.2894, 0.6814, ..., 0.9382, 0.9968, 0.5782])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 799988
Density: 1.99997e-05
Time: 10.433407306671143 seconds
[39.12, 38.28, 38.59, 38.33, 38.43, 38.29, 39.08, 38.65, 39.89, 43.78]
[65.56]
10.455500602722168
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 4694, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 799988, 'MATRIX_DENSITY': 1.99997e-05, 'TIME_S': 10.433407306671143, 'TIME_S_1KI': 2.2227113989499663, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 685.4626195144654, 'W': 65.56}
[39.12, 38.28, 38.59, 38.33, 38.43, 38.29, 39.08, 38.65, 39.89, 43.78, 39.0, 38.3, 38.84, 38.7, 38.69, 38.36, 43.77, 38.52, 38.75, 38.49]
703.665
35.18325
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 4694, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 2e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 799988, 'MATRIX_DENSITY': 1.99997e-05, 'TIME_S': 10.433407306671143, 'TIME_S_1KI': 2.2227113989499663, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 685.4626195144654, 'W': 65.56, 'J_1KI': 146.02953121313706, 'W_1KI': 13.966766084363018, 'W_D': 30.37675, 'J_D': 317.60412793374064, 'W_D_1KI': 6.471399659139327, 'J_D_1KI': 1.3786535277246117}

1
pytorch/output_1core_after_test/epyc_7313p_10_10_10_200000_5e-05.json
View File

@ -1 +0,0 @@
{"CPU": "Epyc 7313P", "CORES": 1, "ITERATIONS": 3293, "MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 1999955, "MATRIX_DENSITY": 4.9998875e-05, "TIME_S": 10.427314758300781, "TIME_S_1KI": 3.16650918867318, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 724.648955361843, "W": 69.33, "J_1KI": 220.05738091765656, "W_1KI": 21.053750379593076, "W_D": 34.4645, "J_D": 360.22881756913665, "W_D_1KI": 10.465988460370484, "J_D_1KI": 3.178253404303214}

71
pytorch/output_1core_after_test/epyc_7313p_10_10_10_200000_5e-05.output
View File

@ -1,71 +0,0 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '200000', '-sd', '5e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 1999939, "MATRIX_DENSITY": 4.9998475e-05, "TIME_S": 3.1881461143493652}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 15, 21, ..., 1999920,
1999927, 1999939]),
col_indices=tensor([ 21664, 28016, 30855, ..., 178410, 188321,
197716]),
values=tensor([-0.8535, -1.1008, -0.1929, ..., -0.4377, -0.2253,
-1.5714]), size=(200000, 200000), nnz=1999939,
layout=torch.sparse_csr)
tensor([0.6684, 0.0214, 0.6544, ..., 0.8819, 0.1706, 0.8563])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 1999939
Density: 4.9998475e-05
Time: 3.1881461143493652 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:1}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '3293', '-ss', '200000', '-sd', '5e-05', '-c', '1']
{"MATRIX_TYPE": "synthetic", "MATRIX_DENSITY_GROUP": 5e-05, "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [200000, 200000], "MATRIX_ROWS": 200000, "MATRIX_SIZE": 40000000000, "MATRIX_NNZ": 1999955, "MATRIX_DENSITY": 4.9998875e-05, "TIME_S": 10.427314758300781}
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 10, 18, ..., 1999938,
1999949, 1999955]),
col_indices=tensor([ 20189, 20497, 53226, ..., 105399, 143618,
172009]),
values=tensor([-0.5680, 0.4071, -0.7459, ..., 0.6726, 0.9697,
0.1668]), size=(200000, 200000), nnz=1999955,
layout=torch.sparse_csr)
tensor([0.0111, 0.0503, 0.6606, ..., 0.3364, 0.9745, 0.1391])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 1999955
Density: 4.9998875e-05
Time: 10.427314758300781 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:62: UserWarning: Sparse CSR tensor support is in beta state. If you miss a functionality in the sparse tensor support, please submit a feature request to https://github.com/pytorch/pytorch/issues. (Triggered internally at ../aten/src/ATen/SparseCsrTensorImpl.cpp:53.)
matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 10, 18, ..., 1999938,
1999949, 1999955]),
col_indices=tensor([ 20189, 20497, 53226, ..., 105399, 143618,
172009]),
values=tensor([-0.5680, 0.4071, -0.7459, ..., 0.6726, 0.9697,
0.1668]), size=(200000, 200000), nnz=1999955,
layout=torch.sparse_csr)
tensor([0.0111, 0.0503, 0.6606, ..., 0.3364, 0.9745, 0.1391])
Matrix Type: synthetic
Matrix: csr
Shape: torch.Size([200000, 200000])
Rows: 200000
Size: 40000000000
NNZ: 1999955
Density: 4.9998875e-05
Time: 10.427314758300781 seconds
[39.96, 38.8, 38.3, 38.76, 38.9, 38.39, 38.83, 38.3, 38.63, 38.25]
[69.33]
10.452170133590698
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 3293, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 1999955, 'MATRIX_DENSITY': 4.9998875e-05, 'TIME_S': 10.427314758300781, 'TIME_S_1KI': 3.16650918867318, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 724.648955361843, 'W': 69.33}
[39.96, 38.8, 38.3, 38.76, 38.9, 38.39, 38.83, 38.3, 38.63, 38.25, 39.14, 38.68, 38.82, 39.03, 38.72, 38.25, 39.51, 38.98, 38.47, 38.53]
697.31
34.8655
{'CPU': 'Epyc 7313P', 'CORES': 1, 'ITERATIONS': 3293, 'MATRIX_TYPE': 'synthetic', 'MATRIX_DENSITY_GROUP': 5e-05, 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [200000, 200000], 'MATRIX_ROWS': 200000, 'MATRIX_SIZE': 40000000000, 'MATRIX_NNZ': 1999955, 'MATRIX_DENSITY': 4.9998875e-05, 'TIME_S': 10.427314758300781, 'TIME_S_1KI': 3.16650918867318, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 724.648955361843, 'W': 69.33, 'J_1KI': 220.05738091765656, 'W_1KI': 21.053750379593076, 'W_D': 34.4645, 'J_D': 360.22881756913665, 'W_D_1KI': 10.465988460370484, 'J_D_1KI': 3.178253404303214}

Some files were not shown because too many files have changed in this diff Show More