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

Synthetic data

Browse Source
This commit is contained in:
cephi 2024-12-15 22:56:17 -05:00
parent d5caf9c543
commit cf56df3114
369 changed files with 24997 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
pytorch/batch.py
View File

@ -117,7 +117,7 @@ elif args.matrix_type == MatrixType.SYNTHETIC:
parameter_list = enumerate([(size, density)
for size in args.synthetic_size
for density in args.synthetic_density
if size ** 2 * density < 100000000])
if size ** 2 * density < 10000000])
#for i, matrix in enumerate(glob.glob(f'{args.matrix_dir.rstrip("/")}/*.mtx')):
for i, parameter in parameter_list:

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_100000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 1770, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.45119595527649, "TIME_S_1KI": 5.904630483207056, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 518.2880183124543, "W": 35.449367856062224, "J_1KI": 292.81808944206455, "W_1KI": 20.027891444102952, "W_D": 16.922367856062227, "J_D": 247.4137349045278, "W_D_1KI": 9.560659805684875, "J_D_1KI": 5.401502715076201}

65
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_100000_0.0001.output Normal file
View File

@ -0,0 +1,65 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 100000 -sd 0.0001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 5.932083368301392}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 10, ..., 999982,
999993, 1000000]),
col_indices=tensor([37897, 46445, 60989, ..., 76977, 92294, 96477]),
values=tensor([0.9469, 0.5853, 0.3833, ..., 0.6631, 0.6410, 0.8148]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.0925, 0.0591, 0.1895, ..., 0.1208, 0.2736, 0.9441])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 5.932083368301392 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1770 -ss 100000 -sd 0.0001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.45119595527649}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 15, ..., 999984,
999991, 1000000]),
col_indices=tensor([ 6148, 23043, 28153, ..., 62723, 86562, 96964]),
values=tensor([0.4836, 0.5090, 0.9509, ..., 0.7452, 0.4499, 0.9407]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.2481, 0.7173, 0.6398, ..., 0.9063, 0.5779, 0.5048])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 10.45119595527649 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 15, ..., 999984,
999991, 1000000]),
col_indices=tensor([ 6148, 23043, 28153, ..., 62723, 86562, 96964]),
values=tensor([0.4836, 0.5090, 0.9509, ..., 0.7452, 0.4499, 0.9407]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.2481, 0.7173, 0.6398, ..., 0.9063, 0.5779, 0.5048])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 10.45119595527649 seconds
[20.2, 20.56, 20.52, 20.56, 20.56, 20.6, 20.4, 20.64, 20.8, 20.88]
[20.88, 21.44, 21.2, 22.44, 23.6, 27.28, 34.76, 40.76, 46.84, 51.32, 53.0, 52.92, 53.08, 52.84]
14.62051510810852
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1770, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.45119595527649, 'TIME_S_1KI': 5.904630483207056, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 518.2880183124543, 'W': 35.449367856062224}
[20.2, 20.56, 20.52, 20.56, 20.56, 20.6, 20.4, 20.64, 20.8, 20.88, 20.56, 20.52, 20.64, 20.64, 20.52, 20.52, 20.6, 20.68, 20.6, 20.72]
370.53999999999996
18.526999999999997
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1770, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.45119595527649, 'TIME_S_1KI': 5.904630483207056, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 518.2880183124543, 'W': 35.449367856062224, 'J_1KI': 292.81808944206455, 'W_1KI': 20.027891444102952, 'W_D': 16.922367856062227, 'J_D': 247.4137349045278, 'W_D_1KI': 9.560659805684875, 'J_D_1KI': 5.401502715076201}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_100000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 11801, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.278456687927246, "TIME_S_1KI": 0.8709818394989616, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 437.5742044067384, "W": 32.19958635623455, "J_1KI": 37.0794173719802, "W_1KI": 2.728547271945983, "W_D": 13.391586356234548, "J_D": 181.9841000671388, "W_D_1KI": 1.1347840315426274, "J_D_1KI": 0.09615998911470446}

85
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_100000_1e-05.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 100000 -sd 1e-05 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 1.063995361328125}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 99997, 99999,
100000]),
col_indices=tensor([67343, 31299, 81155, ..., 33224, 88457, 24576]),
values=tensor([0.5842, 0.8218, 0.6188, ..., 0.3932, 0.6826, 0.0146]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.9733, 0.2979, 0.3395, ..., 0.2786, 0.7488, 0.6423])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 1.063995361328125 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 9868 -ss 100000 -sd 1e-05 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 8.779469966888428}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 99997, 99999,
100000]),
col_indices=tensor([14435, 22527, 43950, ..., 8583, 8872, 18967]),
values=tensor([0.6873, 0.0224, 0.4938, ..., 0.6581, 0.7037, 0.6316]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.2290, 0.1645, 0.1242, ..., 0.3445, 0.2954, 0.7059])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 8.779469966888428 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 11801 -ss 100000 -sd 1e-05 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.278456687927246}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 100000, 100000,
100000]),
col_indices=tensor([88946, 66534, 50450, ..., 63020, 21924, 98776]),
values=tensor([0.0165, 0.3102, 0.5959, ..., 0.2885, 0.2555, 0.6064]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.1483, 0.9193, 0.9702, ..., 0.6151, 0.3023, 0.2526])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 10.278456687927246 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 100000, 100000,
100000]),
col_indices=tensor([88946, 66534, 50450, ..., 63020, 21924, 98776]),
values=tensor([0.0165, 0.3102, 0.5959, ..., 0.2885, 0.2555, 0.6064]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.1483, 0.9193, 0.9702, ..., 0.6151, 0.3023, 0.2526])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 10.278456687927246 seconds
[20.52, 20.48, 20.76, 20.96, 20.96, 21.16, 21.32, 21.28, 21.28, 21.2]
[21.36, 21.64, 21.64, 23.32, 23.96, 29.24, 34.28, 39.64, 43.16, 45.96, 45.88, 46.84, 47.12]
13.589435577392578
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 11801, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.278456687927246, 'TIME_S_1KI': 0.8709818394989616, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 437.5742044067384, 'W': 32.19958635623455}
[20.52, 20.48, 20.76, 20.96, 20.96, 21.16, 21.32, 21.28, 21.28, 21.2, 21.04, 20.92, 20.64, 20.52, 20.52, 20.4, 20.72, 20.96, 21.24, 21.32]
376.16
18.808
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 11801, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.278456687927246, 'TIME_S_1KI': 0.8709818394989616, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 437.5742044067384, 'W': 32.19958635623455, 'J_1KI': 37.0794173719802, 'W_1KI': 2.728547271945983, 'W_D': 13.391586356234548, 'J_D': 181.9841000671388, 'W_D_1KI': 1.1347840315426274, 'J_D_1KI': 0.09615998911470446}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 33464, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.751937627792358, "TIME_S_1KI": 0.321298638172136, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 327.9264199829101, "W": 23.099563679174377, "J_1KI": 9.799379033675296, "W_1KI": 0.6902810088206544, "W_D": 4.345563679174376, "J_D": 61.690565237998875, "W_D_1KI": 0.12985786753449605, "J_D_1KI": 0.0038805243705025113}

81
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.0001.output Normal file
View File

@ -0,0 +1,81 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 10000 -sd 0.0001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 0.358994722366333}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9999, 9999, 10000]),
col_indices=tensor([4769, 2640, 4731, ..., 7727, 9096, 344]),
values=tensor([0.5549, 0.8764, 0.0270, ..., 0.0575, 0.5131, 0.9423]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.2724, 0.3491, 0.1026, ..., 0.4580, 0.8295, 0.5142])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 0.358994722366333 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 29248 -ss 10000 -sd 0.0001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 9.177036046981812}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9997, 9998, 10000]),
col_indices=tensor([8143, 7461, 5162, ..., 7740, 5053, 9684]),
values=tensor([0.7267, 0.3238, 0.0105, ..., 0.5150, 0.5465, 0.0983]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.8883, 0.6326, 0.2674, ..., 0.1564, 0.2088, 0.8392])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 9.177036046981812 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 33464 -ss 10000 -sd 0.0001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.751937627792358}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9994, 9997, 10000]),
col_indices=tensor([1608, 4931, 8613, ..., 2107, 3637, 7054]),
values=tensor([0.4097, 0.1049, 0.8257, ..., 0.2263, 0.1754, 0.1229]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.9092, 0.1064, 0.7261, ..., 0.1695, 0.8231, 0.3389])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 10.751937627792358 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9994, 9997, 10000]),
col_indices=tensor([1608, 4931, 8613, ..., 2107, 3637, 7054]),
values=tensor([0.4097, 0.1049, 0.8257, ..., 0.2263, 0.1754, 0.1229]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.9092, 0.1064, 0.7261, ..., 0.1695, 0.8231, 0.3389])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 10.751937627792358 seconds
[20.16, 20.16, 20.16, 20.32, 20.36, 20.88, 21.6, 22.28, 22.28, 22.28]
[21.52, 20.68, 23.48, 24.56, 27.0, 27.0, 27.6, 28.4, 25.44, 25.08, 23.88, 23.84, 23.72, 23.68]
14.196217060089111
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 33464, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.751937627792358, 'TIME_S_1KI': 0.321298638172136, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 327.9264199829101, 'W': 23.099563679174377}
[20.16, 20.16, 20.16, 20.32, 20.36, 20.88, 21.6, 22.28, 22.28, 22.28, 20.28, 20.68, 20.64, 20.84, 20.84, 20.88, 20.6, 20.6, 20.48, 20.24]
375.08000000000004
18.754
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 33464, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.751937627792358, 'TIME_S_1KI': 0.321298638172136, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 327.9264199829101, 'W': 23.099563679174377, 'J_1KI': 9.799379033675296, 'W_1KI': 0.6902810088206544, 'W_D': 4.345563679174376, 'J_D': 61.690565237998875, 'W_D_1KI': 0.12985786753449605, 'J_D_1KI': 0.0038805243705025113}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 4693, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 10.608984231948853, "TIME_S_1KI": 2.260597535041307, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 333.61959093093867, "W": 23.443356307834602, "J_1KI": 71.08876857680346, "W_1KI": 4.995388090312082, "W_D": 4.929356307834599, "J_D": 70.14907820272437, "W_D_1KI": 1.0503635857307905, "J_D_1KI": 0.223814955408223}

65
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.001.output Normal file
View File

@ -0,0 +1,65 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 10000 -sd 0.001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 2.2371175289154053}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 13, ..., 99981, 99991,
100000]),
col_indices=tensor([ 11, 880, 2486, ..., 7621, 8410, 9572]),
values=tensor([0.7919, 0.7111, 0.9252, ..., 0.0051, 0.9566, 0.6694]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.8227, 0.5043, 0.0669, ..., 0.5765, 0.9663, 0.4234])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 2.2371175289154053 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 4693 -ss 10000 -sd 0.001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 10.608984231948853}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 14, 27, ..., 99982, 99994,
100000]),
col_indices=tensor([ 135, 2132, 2413, ..., 7244, 7277, 8789]),
values=tensor([0.8089, 0.0016, 0.7063, ..., 0.2204, 0.7876, 0.4440]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.2483, 0.7850, 0.0043, ..., 0.4009, 0.1492, 0.4510])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 10.608984231948853 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 14, 27, ..., 99982, 99994,
100000]),
col_indices=tensor([ 135, 2132, 2413, ..., 7244, 7277, 8789]),
values=tensor([0.8089, 0.0016, 0.7063, ..., 0.2204, 0.7876, 0.4440]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.2483, 0.7850, 0.0043, ..., 0.4009, 0.1492, 0.4510])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 10.608984231948853 seconds
[20.32, 20.32, 20.36, 20.6, 20.68, 20.44, 20.64, 20.8, 20.88, 20.84]
[20.84, 20.52, 23.32, 24.96, 27.48, 27.48, 28.36, 28.96, 25.92, 25.2, 24.36, 24.56, 24.48, 24.08]
14.23088002204895
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 4693, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 10.608984231948853, 'TIME_S_1KI': 2.260597535041307, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 333.61959093093867, 'W': 23.443356307834602}
[20.32, 20.32, 20.36, 20.6, 20.68, 20.44, 20.64, 20.8, 20.88, 20.84, 20.68, 20.8, 20.52, 20.64, 20.64, 20.68, 20.4, 20.48, 20.36, 20.24]
370.28000000000003
18.514000000000003
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 4693, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 10.608984231948853, 'TIME_S_1KI': 2.260597535041307, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 333.61959093093867, 'W': 23.443356307834602, 'J_1KI': 71.08876857680346, 'W_1KI': 4.995388090312082, 'W_D': 4.929356307834599, 'J_D': 70.14907820272437, 'W_D_1KI': 1.0503635857307905, 'J_D_1KI': 0.223814955408223}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.01.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 21.223905086517334, "TIME_S_1KI": 21.223905086517334, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 606.5871645927429, "W": 23.902485146880146, "J_1KI": 606.5871645927429, "W_1KI": 23.902485146880146, "W_D": 5.469485146880146, "J_D": 138.80228213262555, "W_D_1KI": 5.469485146880146, "J_D_1KI": 5.469485146880146}

45
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.01.output Normal file
View File

@ -0,0 +1,45 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 10000 -sd 0.01 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 21.223905086517334}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 100, 193, ..., 999807,
999898, 1000000]),
col_indices=tensor([ 45, 67, 78, ..., 9873, 9905, 9941]),
values=tensor([0.2793, 0.5501, 0.9236, ..., 0.0106, 0.8963, 0.7259]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.2312, 0.2281, 0.2895, ..., 0.4123, 0.5947, 0.5960])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 21.223905086517334 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 100, 193, ..., 999807,
999898, 1000000]),
col_indices=tensor([ 45, 67, 78, ..., 9873, 9905, 9941]),
values=tensor([0.2793, 0.5501, 0.9236, ..., 0.0106, 0.8963, 0.7259]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.2312, 0.2281, 0.2895, ..., 0.4123, 0.5947, 0.5960])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 21.223905086517334 seconds
[20.16, 20.16, 20.16, 20.04, 20.28, 20.72, 20.6, 20.64, 20.6, 20.44]
[20.44, 20.64, 23.68, 24.76, 27.96, 27.96, 29.28, 30.08, 27.32, 27.04, 23.96, 23.92, 23.72, 23.6, 23.72, 23.92, 24.08, 24.24, 24.24, 24.36, 24.24, 24.12, 24.4, 23.96, 24.12]
25.377577304840088
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.01, 'TIME_S': 21.223905086517334, 'TIME_S_1KI': 21.223905086517334, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 606.5871645927429, 'W': 23.902485146880146}
[20.16, 20.16, 20.16, 20.04, 20.28, 20.72, 20.6, 20.64, 20.6, 20.44, 20.2, 20.32, 20.32, 20.52, 20.52, 20.8, 20.8, 20.72, 20.68, 20.76]
368.65999999999997
18.433
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.01, 'TIME_S': 21.223905086517334, 'TIME_S_1KI': 21.223905086517334, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 606.5871645927429, 'W': 23.902485146880146, 'J_1KI': 606.5871645927429, 'W_1KI': 23.902485146880146, 'W_D': 5.469485146880146, 'J_D': 138.80228213262555, 'W_D_1KI': 5.469485146880146, 'J_D_1KI': 5.469485146880146}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 112.2527105808258, "TIME_S_1KI": 112.2527105808258, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2847.1031341934195, "W": 24.02975891792854, "J_1KI": 2847.1031341934195, "W_1KI": 24.02975891792854, "W_D": 5.456758917928539, "J_D": 646.5298079283226, "W_D_1KI": 5.456758917928539, "J_D_1KI": 5.456758917928539}

45
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.05.output Normal file
View File

@ -0,0 +1,45 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 10000 -sd 0.05 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 112.2527105808258}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 493, 999, ..., 4999078,
4999538, 5000000]),
col_indices=tensor([ 9, 32, 79, ..., 9948, 9954, 9975]),
values=tensor([0.7230, 0.3394, 0.4856, ..., 0.5860, 0.3031, 0.1676]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.5227, 0.7065, 0.1059, ..., 0.0574, 0.9985, 0.1783])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 112.2527105808258 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 493, 999, ..., 4999078,
4999538, 5000000]),
col_indices=tensor([ 9, 32, 79, ..., 9948, 9954, 9975]),
values=tensor([0.7230, 0.3394, 0.4856, ..., 0.5860, 0.3031, 0.1676]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.5227, 0.7065, 0.1059, ..., 0.0574, 0.9985, 0.1783])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 112.2527105808258 seconds
[20.36, 20.76, 20.76, 20.64, 20.76, 20.64, 20.44, 20.2, 20.64, 20.52]
[20.84, 20.72, 21.32, 21.96, 24.12, 27.04, 27.04, 28.68, 28.56, 28.24, 25.72, 24.44, 24.36, 24.24, 24.24, 24.56, 24.28, 24.4, 24.4, 24.44, 24.56, 24.2, 24.24, 24.04, 24.28, 24.12, 24.12, 24.28, 24.32, 24.24, 24.56, 24.56, 24.6, 24.44, 24.6, 24.6, 24.44, 24.44, 24.44, 24.4, 24.36, 24.36, 24.28, 24.28, 24.32, 24.24, 24.28, 24.08, 24.04, 24.04, 24.2, 24.24, 24.32, 24.6, 24.68, 24.36, 24.36, 24.28, 24.24, 24.08, 24.24, 24.32, 24.36, 24.6, 24.6, 24.64, 24.68, 24.6, 24.6, 24.4, 24.28, 24.4, 24.4, 24.2, 24.32, 24.36, 24.4, 24.44, 24.56, 24.44, 24.44, 24.4, 24.28, 24.4, 24.56, 24.56, 24.64, 24.76, 24.68, 24.44, 24.44, 24.36, 24.32, 24.32, 24.16, 24.24, 24.2, 24.12, 23.8, 23.88, 23.88, 23.76, 24.08, 24.24, 24.4, 24.4, 24.6, 24.52, 24.4, 24.56, 24.48, 24.4, 24.68, 24.72, 24.68, 24.8, 24.8]
118.48238444328308
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.05, 'TIME_S': 112.2527105808258, 'TIME_S_1KI': 112.2527105808258, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2847.1031341934195, 'W': 24.02975891792854}
[20.36, 20.76, 20.76, 20.64, 20.76, 20.64, 20.44, 20.2, 20.64, 20.52, 20.52, 20.56, 20.56, 20.56, 20.8, 20.88, 20.8, 20.8, 20.68, 20.56]
371.46000000000004
18.573
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.05, 'TIME_S': 112.2527105808258, 'TIME_S_1KI': 112.2527105808258, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2847.1031341934195, 'W': 24.02975891792854, 'J_1KI': 2847.1031341934195, 'W_1KI': 24.02975891792854, 'W_D': 5.456758917928539, 'J_D': 646.5298079283226, 'W_D_1KI': 5.456758917928539, 'J_D_1KI': 5.456758917928539}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 141369, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.808244943618774, "TIME_S_1KI": 0.0764541373541496, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 314.87872554779057, "W": 22.096174468711904, "J_1KI": 2.2273534194044706, "W_1KI": 0.15630141310125914, "W_D": 3.7551744687119033, "J_D": 53.51263643360139, "W_D_1KI": 0.02656292729461129, "J_D_1KI": 0.00018789782268114857}

1521
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_1e-05.output Normal file
View File

File diff suppressed because it is too large Load Diff

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_500000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 1458, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [500000, 500000], "MATRIX_ROWS": 500000, "MATRIX_SIZE": 250000000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.73922610282898, "TIME_S_1KI": 7.365724350362812, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 524.544666223526, "W": 35.90986855086994, "J_1KI": 359.77000426853635, "W_1KI": 24.629539472475955, "W_D": 17.579868550869936, "J_D": 256.7936518120765, "W_D_1KI": 12.0575230115706, "J_D_1KI": 8.269906043601233}

68
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_500000_1e-05.output Normal file
View File

@ -0,0 +1,68 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 500000 -sd 1e-05 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [500000, 500000], "MATRIX_ROWS": 500000, "MATRIX_SIZE": 250000000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 1e-05, "TIME_S": 7.201478004455566}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 2499994,
2499998, 2500000]),
col_indices=tensor([111852, 327751, 365150, ..., 493517, 11445,
207886]),
values=tensor([0.9407, 0.2669, 0.8671, ..., 0.7942, 0.4760, 0.2816]),
size=(500000, 500000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.4423, 0.4635, 0.1741, ..., 0.0346, 0.7600, 0.4318])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([500000, 500000])
Rows: 500000
Size: 250000000000
NNZ: 2500000
Density: 1e-05
Time: 7.201478004455566 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1458 -ss 500000 -sd 1e-05 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [500000, 500000], "MATRIX_ROWS": 500000, "MATRIX_SIZE": 250000000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.73922610282898}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 2499994,
2500000, 2500000]),
col_indices=tensor([198857, 399888, 193187, ..., 179513, 216653,
450880]),
values=tensor([0.4554, 0.7901, 0.7135, ..., 0.0158, 0.9399, 0.2709]),
size=(500000, 500000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.8645, 0.3649, 0.9819, ..., 0.4118, 0.2155, 0.1417])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([500000, 500000])
Rows: 500000
Size: 250000000000
NNZ: 2500000
Density: 1e-05
Time: 10.73922610282898 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 2499994,
2500000, 2500000]),
col_indices=tensor([198857, 399888, 193187, ..., 179513, 216653,
450880]),
values=tensor([0.4554, 0.7901, 0.7135, ..., 0.0158, 0.9399, 0.2709]),
size=(500000, 500000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.8645, 0.3649, 0.9819, ..., 0.4118, 0.2155, 0.1417])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([500000, 500000])
Rows: 500000
Size: 250000000000
NNZ: 2500000
Density: 1e-05
Time: 10.73922610282898 seconds
[20.76, 20.72, 20.4, 20.4, 20.36, 20.36, 20.12, 20.36, 20.28, 20.32]
[20.56, 20.48, 21.52, 22.84, 24.72, 30.76, 37.24, 43.6, 43.6, 49.32, 53.6, 53.68, 53.6, 53.56]
14.607256650924683
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1458, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [500000, 500000], 'MATRIX_ROWS': 500000, 'MATRIX_SIZE': 250000000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.73922610282898, 'TIME_S_1KI': 7.365724350362812, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 524.544666223526, 'W': 35.90986855086994}
[20.76, 20.72, 20.4, 20.4, 20.36, 20.36, 20.12, 20.36, 20.28, 20.32, 20.48, 20.4, 20.32, 20.04, 20.2, 20.4, 20.36, 20.36, 20.48, 20.52]
366.6
18.330000000000002
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1458, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [500000, 500000], 'MATRIX_ROWS': 500000, 'MATRIX_SIZE': 250000000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.73922610282898, 'TIME_S_1KI': 7.365724350362812, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 524.544666223526, 'W': 35.90986855086994, 'J_1KI': 359.77000426853635, 'W_1KI': 24.629539472475955, 'W_D': 17.579868550869936, 'J_D': 256.7936518120765, 'W_D_1KI': 12.0575230115706, 'J_D_1KI': 8.269906043601233}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_50000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 3515, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.52223539352417, "TIME_S_1KI": 2.9935235827949276, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 470.9832169723511, "W": 32.170385708153674, "J_1KI": 133.99238036197755, "W_1KI": 9.15231456846477, "W_D": 13.629385708153674, "J_D": 199.53792237424858, "W_D_1KI": 3.8774923778531076, "J_D_1KI": 1.1031272767718656}

65
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_50000_0.0001.output Normal file
View File

@ -0,0 +1,65 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 50000 -sd 0.0001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 2.9865975379943848}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 13, ..., 249989, 249995,
250000]),
col_indices=tensor([12071, 16957, 24871, ..., 32088, 41674, 47752]),
values=tensor([0.0278, 0.4403, 0.7542, ..., 0.8727, 0.3256, 0.0294]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.9906, 0.0790, 0.7013, ..., 0.2118, 0.2385, 0.3873])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 2.9865975379943848 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 3515 -ss 50000 -sd 0.0001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.52223539352417}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 10, ..., 249992, 249995,
250000]),
col_indices=tensor([ 9701, 11138, 26862, ..., 20273, 37187, 48197]),
values=tensor([0.8537, 0.5403, 0.1220, ..., 0.0155, 0.7712, 0.8752]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.5658, 0.7328, 0.9479, ..., 0.1014, 0.1582, 0.5663])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 10.52223539352417 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 10, ..., 249992, 249995,
250000]),
col_indices=tensor([ 9701, 11138, 26862, ..., 20273, 37187, 48197]),
values=tensor([0.8537, 0.5403, 0.1220, ..., 0.0155, 0.7712, 0.8752]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.5658, 0.7328, 0.9479, ..., 0.1014, 0.1582, 0.5663])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 10.52223539352417 seconds
[20.72, 20.68, 20.52, 20.64, 20.64, 20.68, 20.68, 20.72, 20.72, 20.72]
[20.64, 20.72, 21.28, 22.68, 24.8, 29.24, 34.6, 38.2, 42.72, 43.84, 43.84, 44.32, 44.16, 44.08]
14.640272617340088
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 3515, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 250000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.52223539352417, 'TIME_S_1KI': 2.9935235827949276, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 470.9832169723511, 'W': 32.170385708153674}
[20.72, 20.68, 20.52, 20.64, 20.64, 20.68, 20.68, 20.72, 20.72, 20.72, 20.32, 20.24, 20.44, 20.32, 20.48, 20.64, 20.52, 20.64, 20.96, 20.84]
370.82
18.541
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 3515, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 250000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.52223539352417, 'TIME_S_1KI': 2.9935235827949276, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 470.9832169723511, 'W': 32.170385708153674, 'J_1KI': 133.99238036197755, 'W_1KI': 9.15231456846477, 'W_D': 13.629385708153674, 'J_D': 199.53792237424858, 'W_D_1KI': 3.8774923778531076, 'J_D_1KI': 1.1031272767718656}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_50000_0.001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.001, "TIME_S": 27.268765687942505, "TIME_S_1KI": 27.268765687942505, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1234.21608165741, "W": 36.82999610504039, "J_1KI": 1234.21608165741, "W_1KI": 36.82999610504039, "W_D": 18.278996105040388, "J_D": 612.5504571070677, "W_D_1KI": 18.278996105040388, "J_D_1KI": 18.278996105040388}

45
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_50000_0.001.output Normal file
View File

@ -0,0 +1,45 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 50000 -sd 0.001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.001, "TIME_S": 27.268765687942505}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 46, 102, ..., 2499892,
2499945, 2500000]),
col_indices=tensor([ 987, 1836, 5791, ..., 47187, 47558, 49789]),
values=tensor([0.1085, 0.8855, 0.3536, ..., 0.4174, 0.4340, 0.6085]),
size=(50000, 50000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.6397, 0.2759, 0.9232, ..., 0.5725, 0.2810, 0.6127])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 2500000
Density: 0.001
Time: 27.268765687942505 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 46, 102, ..., 2499892,
2499945, 2500000]),
col_indices=tensor([ 987, 1836, 5791, ..., 47187, 47558, 49789]),
values=tensor([0.1085, 0.8855, 0.3536, ..., 0.4174, 0.4340, 0.6085]),
size=(50000, 50000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.6397, 0.2759, 0.9232, ..., 0.5725, 0.2810, 0.6127])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 2500000
Density: 0.001
Time: 27.268765687942505 seconds
[20.68, 20.56, 20.48, 20.12, 20.08, 20.28, 20.44, 20.68, 20.96, 20.96]
[20.84, 20.76, 20.6, 24.92, 26.28, 30.48, 35.16, 37.64, 40.0, 42.72, 43.28, 43.52, 43.36, 43.36, 43.52, 42.92, 43.08, 42.76, 42.76, 42.52, 42.68, 42.8, 42.88, 43.04, 43.16, 42.96, 42.88, 42.76, 42.52, 42.72, 42.64, 42.64]
33.511165142059326
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 27.268765687942505, 'TIME_S_1KI': 27.268765687942505, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1234.21608165741, 'W': 36.82999610504039}
[20.68, 20.56, 20.48, 20.12, 20.08, 20.28, 20.44, 20.68, 20.96, 20.96, 20.44, 20.76, 20.8, 20.8, 20.72, 20.84, 20.84, 20.56, 20.68, 20.76]
371.02
18.551
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 27.268765687942505, 'TIME_S_1KI': 27.268765687942505, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1234.21608165741, 'W': 36.82999610504039, 'J_1KI': 1234.21608165741, 'W_1KI': 36.82999610504039, 'W_D': 18.278996105040388, 'J_D': 612.5504571070677, 'W_D_1KI': 18.278996105040388, 'J_D_1KI': 18.278996105040388}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_50000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 19539, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.163734436035156, "TIME_S_1KI": 0.5201767969719615, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 495.2794138240814, "W": 33.936722825817846, "J_1KI": 25.348247803064712, "W_1KI": 1.7368710182618274, "W_D": 13.302722825817849, "J_D": 194.14263413858416, "W_D_1KI": 0.6808292556332386, "J_D_1KI": 0.03484463153862729}

62
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_50000_1e-05.output Normal file
View File

@ -0,0 +1,62 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 50000 -sd 1e-05 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.5373842716217041}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 24999, 25000, 25000]),
col_indices=tensor([13933, 723, 18387, ..., 22194, 38514, 2158]),
values=tensor([0.9124, 0.6353, 0.3193, ..., 0.0372, 0.2371, 0.8076]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.6534, 0.7497, 0.2436, ..., 0.0965, 0.5741, 0.5754])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 0.5373842716217041 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 19539 -ss 50000 -sd 1e-05 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.163734436035156}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 24996, 24998, 25000]),
col_indices=tensor([44723, 48345, 32100, ..., 22467, 28064, 29572]),
values=tensor([0.7283, 0.2640, 0.9583, ..., 0.2460, 0.4237, 0.5300]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.7754, 0.9311, 0.4703, ..., 0.3816, 0.8788, 0.3934])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 10.163734436035156 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 24996, 24998, 25000]),
col_indices=tensor([44723, 48345, 32100, ..., 22467, 28064, 29572]),
values=tensor([0.7283, 0.2640, 0.9583, ..., 0.2460, 0.4237, 0.5300]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.7754, 0.9311, 0.4703, ..., 0.3816, 0.8788, 0.3934])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 10.163734436035156 seconds
[25.08, 25.2, 25.2, 25.4, 25.08, 25.4, 25.12, 25.0, 25.28, 25.36]
[25.56, 25.6, 26.0, 26.32, 28.76, 32.48, 32.48, 37.24, 41.24, 45.24, 45.88, 45.72, 45.6, 45.64]
14.594202756881714
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 19539, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 25000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.163734436035156, 'TIME_S_1KI': 0.5201767969719615, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 495.2794138240814, 'W': 33.936722825817846}
[25.08, 25.2, 25.2, 25.4, 25.08, 25.4, 25.12, 25.0, 25.28, 25.36, 20.68, 20.68, 20.6, 20.6, 20.52, 20.72, 20.52, 20.72, 20.72, 20.72]
412.67999999999995
20.633999999999997
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 19539, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 25000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.163734436035156, 'TIME_S_1KI': 0.5201767969719615, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 495.2794138240814, 'W': 33.936722825817846, 'J_1KI': 25.348247803064712, 'W_1KI': 1.7368710182618274, 'W_D': 13.302722825817849, 'J_D': 194.14263413858416, 'W_D_1KI': 0.6808292556332386, 'J_D_1KI': 0.03484463153862729}

1
pytorch/output_synthetic_16core/altra_16_csr_20_10_10_synthetic_30000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 9519, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 21.79372525215149, "TIME_S_1KI": 2.2894973476364626, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 607.9340667724609, "W": 24.009013571692563, "J_1KI": 63.86532900225454, "W_1KI": 2.5222201462015508, "W_D": 5.522013571692561, "J_D": 139.82332749271384, "W_D_1KI": 0.5801043777384768, "J_D_1KI": 0.06094173523883567}

62
pytorch/output_synthetic_16core/altra_16_csr_20_10_10_synthetic_30000_0.0001.output Normal file
View File

@ -0,0 +1,62 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 30000 -sd 0.0001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 2.2060210704803467}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 6, ..., 89996, 89998, 90000]),
col_indices=tensor([ 2876, 4713, 6957, ..., 29701, 15647, 23288]),
values=tensor([0.6297, 0.3832, 0.4268, ..., 0.4020, 0.1713, 0.6526]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.2297, 0.3740, 0.0656, ..., 0.6156, 0.3028, 0.9303])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 2.2060210704803467 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 9519 -ss 30000 -sd 0.0001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 21.79372525215149}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 89997, 89999, 90000]),
col_indices=tensor([15244, 15936, 9998, ..., 16898, 18863, 20836]),
values=tensor([0.4356, 0.9410, 0.0325, ..., 0.8568, 0.9195, 0.8628]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.3669, 0.2405, 0.0914, ..., 0.8449, 0.6451, 0.3598])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 21.79372525215149 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 89997, 89999, 90000]),
col_indices=tensor([15244, 15936, 9998, ..., 16898, 18863, 20836]),
values=tensor([0.4356, 0.9410, 0.0325, ..., 0.8568, 0.9195, 0.8628]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.3669, 0.2405, 0.0914, ..., 0.8449, 0.6451, 0.3598])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 21.79372525215149 seconds
[20.48, 20.6, 20.6, 20.52, 20.52, 20.76, 20.88, 21.04, 21.0, 21.0]
[20.96, 20.88, 23.92, 25.0, 27.36, 27.36, 28.16, 28.88, 25.92, 25.64, 24.4, 24.24, 24.2, 24.16, 24.24, 24.84, 24.92, 24.92, 24.6, 24.64, 24.64, 24.68, 24.56, 24.8, 24.96]
25.32107639312744
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 9519, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 90000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 21.79372525215149, 'TIME_S_1KI': 2.2894973476364626, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 607.9340667724609, 'W': 24.009013571692563}
[20.48, 20.6, 20.6, 20.52, 20.52, 20.76, 20.88, 21.04, 21.0, 21.0, 20.24, 20.28, 20.44, 20.36, 20.2, 20.24, 20.24, 20.44, 20.52, 20.48]
369.74
18.487000000000002
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 9519, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 90000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 21.79372525215149, 'TIME_S_1KI': 2.2894973476364626, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 607.9340667724609, 'W': 24.009013571692563, 'J_1KI': 63.86532900225454, 'W_1KI': 2.5222201462015508, 'W_D': 5.522013571692561, 'J_D': 139.82332749271384, 'W_D_1KI': 0.5801043777384768, 'J_D_1KI': 0.06094173523883567}

1
pytorch/output_synthetic_16core/altra_16_csr_20_10_10_synthetic_30000_0.001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 900000, "MATRIX_DENSITY": 0.001, "TIME_S": 20.699798345565796, "TIME_S_1KI": 20.699798345565796, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 537.3931253051758, "W": 23.043866328673488, "J_1KI": 537.3931253051758, "W_1KI": 23.043866328673488, "W_D": 4.700866328673488, "J_D": 109.62627590250972, "W_D_1KI": 4.700866328673488, "J_D_1KI": 4.700866328673488}

45
pytorch/output_synthetic_16core/altra_16_csr_20_10_10_synthetic_30000_0.001.output Normal file
View File

@ -0,0 +1,45 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 30000 -sd 0.001 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 900000, "MATRIX_DENSITY": 0.001, "TIME_S": 20.699798345565796}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 33, 63, ..., 899945, 899974,
900000]),
col_indices=tensor([ 547, 1664, 1767, ..., 28485, 29124, 29514]),
values=tensor([0.5453, 0.3696, 0.4974, ..., 0.8638, 0.8625, 0.2546]),
size=(30000, 30000), nnz=900000, layout=torch.sparse_csr)
tensor([0.4458, 0.9665, 0.0852, ..., 0.0100, 0.1262, 0.9671])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 900000
Density: 0.001
Time: 20.699798345565796 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 33, 63, ..., 899945, 899974,
900000]),
col_indices=tensor([ 547, 1664, 1767, ..., 28485, 29124, 29514]),
values=tensor([0.5453, 0.3696, 0.4974, ..., 0.8638, 0.8625, 0.2546]),
size=(30000, 30000), nnz=900000, layout=torch.sparse_csr)
tensor([0.4458, 0.9665, 0.0852, ..., 0.0100, 0.1262, 0.9671])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 900000
Density: 0.001
Time: 20.699798345565796 seconds
[20.6, 20.52, 20.4, 20.52, 20.64, 20.44, 20.4, 20.36, 20.16, 20.16]
[20.16, 20.08, 20.2, 21.68, 22.88, 25.96, 26.92, 27.04, 26.52, 24.64, 24.28, 23.92, 24.12, 24.12, 24.48, 24.6, 24.4, 24.32, 24.28, 24.36, 24.16, 24.4, 24.32]
23.320441007614136
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 900000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 20.699798345565796, 'TIME_S_1KI': 20.699798345565796, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 537.3931253051758, 'W': 23.043866328673488}
[20.6, 20.52, 20.4, 20.52, 20.64, 20.44, 20.4, 20.36, 20.16, 20.16, 20.44, 20.2, 20.16, 20.16, 20.4, 20.44, 20.56, 20.52, 20.28, 20.2]
366.86
18.343
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 900000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 20.699798345565796, 'TIME_S_1KI': 20.699798345565796, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 537.3931253051758, 'W': 23.043866328673488, 'J_1KI': 537.3931253051758, 'W_1KI': 23.043866328673488, 'W_D': 4.700866328673488, 'J_D': 109.62627590250972, 'W_D_1KI': 4.700866328673488, 'J_D_1KI': 4.700866328673488}

0
pytorch/output_synthetic_16core/altra_16_csr_20_10_10_synthetic_30000_0.01.json Normal file
View File

1
pytorch/output_synthetic_16core/altra_16_csr_20_10_10_synthetic_30000_0.01.output Normal file
View File

@ -0,0 +1 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 30000 -sd 0.01 -c 16']

1
pytorch/output_synthetic_16core/altra_16_csr_20_10_10_synthetic_30000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 52473, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 21.229777336120605, "TIME_S_1KI": 0.40458478333849035, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 561.8060250091554, "W": 23.114903822014995, "J_1KI": 10.706573380770212, "W_1KI": 0.4405104305455185, "W_D": 4.660903822014994, "J_D": 113.28292210769662, "W_D_1KI": 0.08882480174594543, "J_D_1KI": 0.0016927715538647577}

62
pytorch/output_synthetic_16core/altra_16_csr_20_10_10_synthetic_30000_1e-05.output Normal file
View File

@ -0,0 +1,62 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 1000 -ss 30000 -sd 1e-05 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.4002048969268799}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 8999, 9000, 9000]),
col_indices=tensor([17165, 27151, 23572, ..., 25119, 9148, 7528]),
values=tensor([0.4884, 0.2785, 0.9649, ..., 0.5831, 0.3229, 0.8447]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.9734, 0.5614, 0.1566, ..., 0.4974, 0.8204, 0.0911])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 0.4002048969268799 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-altra.sif', '-c', 'numactl --cpunodebind=0 --membind=0 python3 spmv.py synthetic csr 52473 -ss 30000 -sd 1e-05 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 21.229777336120605}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 8998, 9000, 9000]),
col_indices=tensor([25247, 22356, 16191, ..., 29211, 15014, 22819]),
values=tensor([0.9864, 0.6356, 0.7247, ..., 0.1858, 0.6120, 0.1833]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.8210, 0.2318, 0.4195, ..., 0.3881, 0.9911, 0.4380])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 21.229777336120605 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 8998, 9000, 9000]),
col_indices=tensor([25247, 22356, 16191, ..., 29211, 15014, 22819]),
values=tensor([0.9864, 0.6356, 0.7247, ..., 0.1858, 0.6120, 0.1833]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.8210, 0.2318, 0.4195, ..., 0.3881, 0.9911, 0.4380])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 21.229777336120605 seconds
[20.6, 20.72, 20.72, 20.72, 20.72, 20.6, 20.28, 20.32, 20.44, 20.48]
[20.48, 20.52, 20.6, 21.72, 23.52, 25.24, 26.32, 26.68, 25.64, 24.84, 24.92, 24.56, 24.64, 24.64, 24.44, 24.12, 24.32, 24.4, 24.32, 24.44, 24.28, 24.08, 24.04, 23.96]
24.3049259185791
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 52473, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 9000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 21.229777336120605, 'TIME_S_1KI': 0.40458478333849035, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 561.8060250091554, 'W': 23.114903822014995}
[20.6, 20.72, 20.72, 20.72, 20.72, 20.6, 20.28, 20.32, 20.44, 20.48, 20.76, 20.68, 20.68, 20.44, 20.48, 20.24, 20.24, 20.32, 20.36, 20.4]
369.08000000000004
18.454
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 52473, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 9000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 21.229777336120605, 'TIME_S_1KI': 0.40458478333849035, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 561.8060250091554, 'W': 23.114903822014995, 'J_1KI': 10.706573380770212, 'W_1KI': 0.4405104305455185, 'W_D': 4.660903822014994, 'J_D': 113.28292210769662, 'W_D_1KI': 0.08882480174594543, 'J_D_1KI': 0.0016927715538647577}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_100000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 66220, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.006447315216064, "TIME_S_1KI": 0.15110914097275843, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1856.6265121269225, "W": 145.81, "J_1KI": 28.037247238401125, "W_1KI": 2.2019027484143763, "W_D": 109.5725, "J_D": 1395.2075200605393, "W_D_1KI": 1.6546738145575355, "J_D_1KI": 0.024987523626661668}

85
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_100000_0.0001.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '100000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 0.20569086074829102}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 9, 18, ..., 999978,
999986, 1000000]),
col_indices=tensor([ 4321, 11912, 13631, ..., 82074, 92560, 99324]),
values=tensor([0.9071, 0.2919, 0.8193, ..., 0.7739, 0.0445, 0.1624]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.6567, 0.9688, 0.9697, ..., 0.6873, 0.4864, 0.9023])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 0.20569086074829102 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '51047', '-ss', '100000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 8.09404468536377}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 17, ..., 999979,
999988, 1000000]),
col_indices=tensor([15686, 48109, 49313, ..., 51931, 56127, 66767]),
values=tensor([0.4545, 0.6496, 0.9508, ..., 0.7270, 0.9957, 0.0621]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.3660, 0.6002, 0.9317, ..., 0.1977, 0.4107, 0.4541])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 8.09404468536377 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '66220', '-ss', '100000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.006447315216064}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 999980,
999990, 1000000]),
col_indices=tensor([ 4776, 21129, 24622, ..., 75160, 86654, 97411]),
values=tensor([0.8410, 0.1609, 0.8553, ..., 0.3742, 0.0938, 0.8797]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.6086, 0.7634, 0.6649, ..., 0.3430, 0.9091, 0.5785])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 10.006447315216064 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 999980,
999990, 1000000]),
col_indices=tensor([ 4776, 21129, 24622, ..., 75160, 86654, 97411]),
values=tensor([0.8410, 0.1609, 0.8553, ..., 0.3742, 0.0938, 0.8797]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.6086, 0.7634, 0.6649, ..., 0.3430, 0.9091, 0.5785])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 10.006447315216064 seconds
[41.04, 40.85, 39.33, 39.23, 39.35, 39.32, 44.72, 39.63, 39.87, 39.35]
[145.81]
12.733190536499023
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 66220, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.006447315216064, 'TIME_S_1KI': 0.15110914097275843, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1856.6265121269225, 'W': 145.81}
[41.04, 40.85, 39.33, 39.23, 39.35, 39.32, 44.72, 39.63, 39.87, 39.35, 40.83, 39.21, 39.3, 44.69, 39.3, 39.36, 39.77, 39.25, 41.13, 39.66]
724.75
36.2375
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 66220, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.006447315216064, 'TIME_S_1KI': 0.15110914097275843, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1856.6265121269225, 'W': 145.81, 'J_1KI': 28.037247238401125, 'W_1KI': 2.2019027484143763, 'W_D': 109.5725, 'J_D': 1395.2075200605393, 'W_D_1KI': 1.6546738145575355, 'J_D_1KI': 0.024987523626661668}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_100000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 101854, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 13.331042528152466, "TIME_S_1KI": 0.13088383890816724, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1504.6270782256126, "W": 115.31, "J_1KI": 14.77239065943029, "W_1KI": 1.1321106682113615, "W_D": 79.84075000000001, "J_D": 1041.8051721085908, "W_D_1KI": 0.7838744673748701, "J_D_1KI": 0.007696059726420858}

85
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_100000_1e-05.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '100000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.1346125602722168}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 3, ..., 99997, 99999,
100000]),
col_indices=tensor([50727, 53996, 86356, ..., 6143, 63321, 22305]),
values=tensor([0.4164, 0.0014, 0.4337, ..., 0.6487, 0.2549, 0.7487]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.9720, 0.1729, 0.4503, ..., 0.2850, 0.8795, 0.9664])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 0.1346125602722168 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '78001', '-ss', '100000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 8.040945768356323}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 100000, 100000,
100000]),
col_indices=tensor([16049, 52557, 57673, ..., 90883, 73385, 65676]),
values=tensor([0.2845, 0.3961, 0.0285, ..., 0.0101, 0.6896, 0.8511]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.5851, 0.1832, 0.4128, ..., 0.6645, 0.1519, 0.8981])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 8.040945768356323 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '101854', '-ss', '100000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 13.331042528152466}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 100000, 100000,
100000]),
col_indices=tensor([ 641, 46150, 85524, ..., 87101, 55219, 61785]),
values=tensor([0.2560, 0.7953, 0.3517, ..., 0.8505, 0.5170, 0.2719]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.1951, 0.6662, 0.1969, ..., 0.4780, 0.9904, 0.5617])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 13.331042528152466 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 100000, 100000,
100000]),
col_indices=tensor([ 641, 46150, 85524, ..., 87101, 55219, 61785]),
values=tensor([0.2560, 0.7953, 0.3517, ..., 0.8505, 0.5170, 0.2719]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.1951, 0.6662, 0.1969, ..., 0.4780, 0.9904, 0.5617])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 13.331042528152466 seconds
[42.16, 39.6, 39.44, 39.17, 39.28, 39.19, 39.17, 39.11, 40.48, 39.44]
[115.31]
13.048539400100708
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 101854, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 13.331042528152466, 'TIME_S_1KI': 0.13088383890816724, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1504.6270782256126, 'W': 115.31}
[42.16, 39.6, 39.44, 39.17, 39.28, 39.19, 39.17, 39.11, 40.48, 39.44, 39.93, 39.1, 39.15, 39.29, 39.15, 39.23, 39.19, 39.11, 39.43, 39.06]
709.3849999999999
35.469249999999995
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 101854, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 13.331042528152466, 'TIME_S_1KI': 0.13088383890816724, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1504.6270782256126, 'W': 115.31, 'J_1KI': 14.77239065943029, 'W_1KI': 1.1321106682113615, 'W_D': 79.84075000000001, 'J_D': 1041.8051721085908, 'W_D_1KI': 0.7838744673748701, 'J_D_1KI': 0.007696059726420858}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 282693, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.381328821182251, "TIME_S_1KI": 0.0367229780050523, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1282.42685277462, "W": 97.9, "J_1KI": 4.5364648320779795, "W_1KI": 0.3463120770588589, "W_D": 62.39075000000001, "J_D": 817.2785818666817, "W_D_1KI": 0.22070143229581213, "J_D_1KI": 0.00078071063767342}

81
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.0001.output Normal file
View File

@ -0,0 +1,81 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 0.05349230766296387}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9999, 10000, 10000]),
col_indices=tensor([3626, 2250, 5764, ..., 7539, 8316, 7972]),
values=tensor([0.1411, 0.7419, 0.4018, ..., 0.4202, 0.3955, 0.4235]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.9736, 0.6802, 0.3390, ..., 0.1575, 0.6861, 0.0446])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 0.05349230766296387 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '196289', '-ss', '10000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 7.290691137313843}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 6, ..., 9998, 10000, 10000]),
col_indices=tensor([ 763, 7857, 9582, ..., 1442, 6306, 9133]),
values=tensor([0.7701, 0.8887, 0.1796, ..., 0.1701, 0.0666, 0.3737]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.4503, 0.2095, 0.3791, ..., 0.5528, 0.9269, 0.0093])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 7.290691137313843 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '282693', '-ss', '10000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.381328821182251}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9997, 9998, 10000]),
col_indices=tensor([4956, 145, 658, ..., 4096, 6098, 6574]),
values=tensor([0.3279, 0.7076, 0.5307, ..., 0.3493, 0.0702, 0.3289]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.0837, 0.8046, 0.5398, ..., 0.6704, 0.0489, 0.7610])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 10.381328821182251 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9997, 9998, 10000]),
col_indices=tensor([4956, 145, 658, ..., 4096, 6098, 6574]),
values=tensor([0.3279, 0.7076, 0.5307, ..., 0.3493, 0.0702, 0.3289]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.0837, 0.8046, 0.5398, ..., 0.6704, 0.0489, 0.7610])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 10.381328821182251 seconds
[39.48, 38.98, 44.27, 38.82, 39.35, 38.99, 39.28, 39.0, 38.75, 39.31]
[97.9]
13.099354982376099
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 282693, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.381328821182251, 'TIME_S_1KI': 0.0367229780050523, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1282.42685277462, 'W': 97.9}
[39.48, 38.98, 44.27, 38.82, 39.35, 38.99, 39.28, 39.0, 38.75, 39.31, 39.89, 39.14, 38.98, 38.75, 41.57, 38.58, 39.15, 38.62, 39.12, 38.99]
710.185
35.509249999999994
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 282693, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.381328821182251, 'TIME_S_1KI': 0.0367229780050523, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1282.42685277462, 'W': 97.9, 'J_1KI': 4.5364648320779795, 'W_1KI': 0.3463120770588589, 'W_D': 62.39075000000001, 'J_D': 817.2785818666817, 'W_D_1KI': 0.22070143229581213, 'J_D_1KI': 0.00078071063767342}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 189141, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 10.465899229049683, "TIME_S_1KI": 0.05533384738924761, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1388.9149661660194, "W": 107.77, "J_1KI": 7.343278116146259, "W_1KI": 0.5697865613484121, "W_D": 72.38875, "J_D": 932.9295560643077, "W_D_1KI": 0.3827237352028381, "J_D_1KI": 0.002023483724855204}

85
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.001.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 0.06988883018493652}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 9, 21, ..., 99977, 99988,
100000]),
col_indices=tensor([ 768, 2423, 2910, ..., 9615, 9787, 9788]),
values=tensor([0.1330, 0.2030, 0.8709, ..., 0.6786, 0.0798, 0.8357]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.0016, 0.6011, 0.7478, ..., 0.9565, 0.9755, 0.4110])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 0.06988883018493652 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '150238', '-ss', '10000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 8.34029221534729}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 9, 21, ..., 99979, 99987,
100000]),
col_indices=tensor([ 978, 1327, 2112, ..., 8470, 8534, 8708]),
values=tensor([0.4296, 0.3021, 0.5865, ..., 0.4657, 0.4173, 0.7957]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.7639, 0.4914, 0.7736, ..., 0.7926, 0.8542, 0.7117])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 8.34029221534729 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '189141', '-ss', '10000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 10.465899229049683}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 15, ..., 99985, 99995,
100000]),
col_indices=tensor([ 277, 3135, 4455, ..., 4161, 8684, 9934]),
values=tensor([0.4295, 0.8999, 0.7885, ..., 0.8935, 0.6648, 0.4808]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.1477, 0.6711, 0.3568, ..., 0.3604, 0.6617, 0.9866])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 10.465899229049683 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 15, ..., 99985, 99995,
100000]),
col_indices=tensor([ 277, 3135, 4455, ..., 4161, 8684, 9934]),
values=tensor([0.4295, 0.8999, 0.7885, ..., 0.8935, 0.6648, 0.4808]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.1477, 0.6711, 0.3568, ..., 0.3604, 0.6617, 0.9866])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 10.465899229049683 seconds
[39.55, 39.91, 39.11, 39.43, 38.85, 39.56, 39.33, 39.4, 38.87, 38.91]
[107.77]
12.887769937515259
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 189141, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 10.465899229049683, 'TIME_S_1KI': 0.05533384738924761, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1388.9149661660194, 'W': 107.77}
[39.55, 39.91, 39.11, 39.43, 38.85, 39.56, 39.33, 39.4, 38.87, 38.91, 39.56, 38.88, 39.03, 38.89, 39.75, 39.07, 39.18, 39.11, 38.8, 42.89]
707.6249999999999
35.381249999999994
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 189141, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 10.465899229049683, 'TIME_S_1KI': 0.05533384738924761, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1388.9149661660194, 'W': 107.77, 'J_1KI': 7.343278116146259, 'W_1KI': 0.5697865613484121, 'W_D': 72.38875, 'J_D': 932.9295560643077, 'W_D_1KI': 0.3827237352028381, 'J_D_1KI': 0.002023483724855204}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.01.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 105256, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 10.995163202285767, "TIME_S_1KI": 0.10446115378017184, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1749.1652185320856, "W": 133.05, "J_1KI": 16.61819961362854, "W_1KI": 1.264060956145018, "W_D": 97.78125000000001, "J_D": 1285.4983955249193, "W_D_1KI": 0.928985045983127, "J_D_1KI": 0.008825958101990642}

85
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.01.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '0.01', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 0.13490986824035645}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 107, 208, ..., 999789,
999899, 1000000]),
col_indices=tensor([ 114, 296, 309, ..., 9749, 9750, 9977]),
values=tensor([0.3507, 0.7412, 0.8612, ..., 0.2456, 0.4049, 0.8296]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.8457, 0.6850, 0.0016, ..., 0.7234, 0.0569, 0.9899])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 0.13490986824035645 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '77829', '-ss', '10000', '-sd', '0.01', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 7.763918876647949}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 109, 200, ..., 999783,
999885, 1000000]),
col_indices=tensor([ 5, 70, 184, ..., 9826, 9903, 9930]),
values=tensor([0.4822, 0.0560, 0.4645, ..., 0.7540, 0.5324, 0.2081]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.2499, 0.5119, 0.0857, ..., 0.6236, 0.3822, 0.7230])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 7.763918876647949 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '105256', '-ss', '10000', '-sd', '0.01', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 10.995163202285767}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 112, 220, ..., 999823,
999909, 1000000]),
col_indices=tensor([ 16, 85, 154, ..., 9645, 9832, 9858]),
values=tensor([0.5111, 0.0405, 0.8270, ..., 0.3072, 0.2885, 0.2472]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.6255, 0.9183, 0.8326, ..., 0.9246, 0.2373, 0.5392])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 10.995163202285767 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 112, 220, ..., 999823,
999909, 1000000]),
col_indices=tensor([ 16, 85, 154, ..., 9645, 9832, 9858]),
values=tensor([0.5111, 0.0405, 0.8270, ..., 0.3072, 0.2885, 0.2472]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.6255, 0.9183, 0.8326, ..., 0.9246, 0.2373, 0.5392])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 10.995163202285767 seconds
[40.1, 38.91, 39.01, 39.37, 38.97, 38.99, 39.2, 38.87, 38.93, 38.8]
[133.05]
13.146675825119019
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 105256, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.01, 'TIME_S': 10.995163202285767, 'TIME_S_1KI': 0.10446115378017184, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1749.1652185320856, 'W': 133.05}
[40.1, 38.91, 39.01, 39.37, 38.97, 38.99, 39.2, 38.87, 38.93, 38.8, 40.08, 39.28, 39.91, 38.88, 38.99, 38.83, 39.08, 39.35, 39.95, 38.73]
705.375
35.26875
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 105256, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.01, 'TIME_S': 10.995163202285767, 'TIME_S_1KI': 0.10446115378017184, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1749.1652185320856, 'W': 133.05, 'J_1KI': 16.61819961362854, 'W_1KI': 1.264060956145018, 'W_D': 97.78125000000001, 'J_D': 1285.4983955249193, 'W_D_1KI': 0.928985045983127, 'J_D_1KI': 0.008825958101990642}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 27486, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 10.233055591583252, "TIME_S_1KI": 0.3723006472961963, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2077.8737717533113, "W": 151.69, "J_1KI": 75.59753226199925, "W_1KI": 5.518809575784036, "W_D": 115.92275, "J_D": 1587.9282864692211, "W_D_1KI": 4.217519828276212, "J_D_1KI": 0.1534424735602202}

85
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.05.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '0.05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 0.45975399017333984}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 512, 1006, ..., 4999034,
4999489, 5000000]),
col_indices=tensor([ 23, 40, 103, ..., 9927, 9976, 9991]),
values=tensor([0.6183, 0.2980, 0.3566, ..., 0.0352, 0.5258, 0.0852]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.4623, 0.5953, 0.6862, ..., 0.1082, 0.6720, 0.4260])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 0.45975399017333984 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '22838', '-ss', '10000', '-sd', '0.05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 8.724292278289795}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 513, 1006, ..., 4998953,
4999498, 5000000]),
col_indices=tensor([ 69, 83, 128, ..., 9917, 9953, 9972]),
values=tensor([0.6637, 0.2623, 0.2360, ..., 0.3507, 0.8119, 0.6229]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.8552, 0.8520, 0.0158, ..., 0.2551, 0.9127, 0.4905])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 8.724292278289795 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '27486', '-ss', '10000', '-sd', '0.05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 10.233055591583252}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 518, 1031, ..., 4999018,
4999521, 5000000]),
col_indices=tensor([ 2, 29, 76, ..., 9919, 9923, 9942]),
values=tensor([0.8327, 0.8899, 0.2406, ..., 0.0134, 0.9622, 0.2874]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.5989, 0.5463, 0.4311, ..., 0.3886, 0.2295, 0.1764])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 10.233055591583252 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 518, 1031, ..., 4999018,
4999521, 5000000]),
col_indices=tensor([ 2, 29, 76, ..., 9919, 9923, 9942]),
values=tensor([0.8327, 0.8899, 0.2406, ..., 0.0134, 0.9622, 0.2874]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.5989, 0.5463, 0.4311, ..., 0.3886, 0.2295, 0.1764])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 10.233055591583252 seconds
[45.15, 39.18, 39.48, 39.74, 39.27, 40.48, 39.56, 39.63, 39.62, 39.49]
[151.69]
13.698159217834473
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 27486, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.05, 'TIME_S': 10.233055591583252, 'TIME_S_1KI': 0.3723006472961963, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2077.8737717533113, 'W': 151.69}
[45.15, 39.18, 39.48, 39.74, 39.27, 40.48, 39.56, 39.63, 39.62, 39.49, 40.16, 40.86, 39.65, 39.13, 39.26, 39.59, 39.63, 39.11, 39.14, 39.23]
715.345
35.767250000000004
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 27486, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.05, 'TIME_S': 10.233055591583252, 'TIME_S_1KI': 0.3723006472961963, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2077.8737717533113, 'W': 151.69, 'J_1KI': 75.59753226199925, 'W_1KI': 5.518809575784036, 'W_D': 115.92275, 'J_D': 1587.9282864692211, 'W_D_1KI': 4.217519828276212, 'J_D_1KI': 0.1534424735602202}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 375977, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.999524116516113, "TIME_S_1KI": 0.029255843087518954, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1323.755545105934, "W": 96.41999999999999, "J_1KI": 3.5208418203930933, "W_1KI": 0.25645185742744897, "W_D": 61.132499999999986, "J_D": 839.2914941006896, "W_D_1KI": 0.16259638222550846, "J_D_1KI": 0.0004324636406628822}

1900
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_1e-05.output Normal file
View File

File diff suppressed because it is too large Load Diff

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_500000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 21375, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [500000, 500000], "MATRIX_ROWS": 500000, "MATRIX_SIZE": 250000000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.636817216873169, "TIME_S_1KI": 0.4976288756431892, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2053.7217113614083, "W": 152.01000000000002, "J_1KI": 96.08054789994893, "W_1KI": 7.111578947368422, "W_D": 116.32275000000001, "J_D": 1571.5713255724313, "W_D_1KI": 5.442000000000001, "J_D_1KI": 0.2545964912280702}

89
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_500000_1e-05.output Normal file
View File

@ -0,0 +1,89 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '500000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [500000, 500000], "MATRIX_ROWS": 500000, "MATRIX_SIZE": 250000000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.5442898273468018}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 11, ..., 2499988,
2499996, 2500000]),
col_indices=tensor([ 37839, 98870, 148404, ..., 161688, 445826,
487462]),
values=tensor([0.2708, 0.4230, 0.0396, ..., 0.5012, 0.9237, 0.4084]),
size=(500000, 500000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.6604, 0.4578, 0.9008, ..., 0.1692, 0.6250, 0.2013])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([500000, 500000])
Rows: 500000
Size: 250000000000
NNZ: 2500000
Density: 1e-05
Time: 0.5442898273468018 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '19291', '-ss', '500000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [500000, 500000], "MATRIX_ROWS": 500000, "MATRIX_SIZE": 250000000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 1e-05, "TIME_S": 9.475887298583984}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 11, ..., 2499990,
2499994, 2500000]),
col_indices=tensor([ 2997, 16168, 106256, ..., 284595, 359619,
400100]),
values=tensor([0.5956, 0.5098, 0.7367, ..., 0.1293, 0.8182, 0.3844]),
size=(500000, 500000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.4741, 0.3124, 0.4103, ..., 0.8230, 0.7925, 0.1055])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([500000, 500000])
Rows: 500000
Size: 250000000000
NNZ: 2500000
Density: 1e-05
Time: 9.475887298583984 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '21375', '-ss', '500000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [500000, 500000], "MATRIX_ROWS": 500000, "MATRIX_SIZE": 250000000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.636817216873169}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 7, ..., 2499987,
2499990, 2500000]),
col_indices=tensor([ 69634, 109368, 119504, ..., 397654, 413765,
480494]),
values=tensor([0.1977, 0.6347, 0.9236, ..., 0.5996, 0.0558, 0.7507]),
size=(500000, 500000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.2633, 0.4244, 0.4182, ..., 0.0717, 0.3446, 0.9616])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([500000, 500000])
Rows: 500000
Size: 250000000000
NNZ: 2500000
Density: 1e-05
Time: 10.636817216873169 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 7, ..., 2499987,
2499990, 2500000]),
col_indices=tensor([ 69634, 109368, 119504, ..., 397654, 413765,
480494]),
values=tensor([0.1977, 0.6347, 0.9236, ..., 0.5996, 0.0558, 0.7507]),
size=(500000, 500000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.2633, 0.4244, 0.4182, ..., 0.0717, 0.3446, 0.9616])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([500000, 500000])
Rows: 500000
Size: 250000000000
NNZ: 2500000
Density: 1e-05
Time: 10.636817216873169 seconds
[39.92, 39.18, 39.71, 40.61, 40.15, 39.79, 39.41, 39.75, 39.22, 39.64]
[152.01]
13.510438203811646
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 21375, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [500000, 500000], 'MATRIX_ROWS': 500000, 'MATRIX_SIZE': 250000000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.636817216873169, 'TIME_S_1KI': 0.4976288756431892, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2053.7217113614083, 'W': 152.01000000000002}
[39.92, 39.18, 39.71, 40.61, 40.15, 39.79, 39.41, 39.75, 39.22, 39.64, 39.92, 39.78, 39.31, 39.34, 39.44, 40.15, 39.73, 39.14, 39.7, 39.19]
713.7450000000001
35.687250000000006
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 21375, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [500000, 500000], 'MATRIX_ROWS': 500000, 'MATRIX_SIZE': 250000000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.636817216873169, 'TIME_S_1KI': 0.4976288756431892, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2053.7217113614083, 'W': 152.01000000000002, 'J_1KI': 96.08054789994893, 'W_1KI': 7.111578947368422, 'W_D': 116.32275000000001, 'J_D': 1571.5713255724313, 'W_D_1KI': 5.442000000000001, 'J_D_1KI': 0.2545964912280702}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_50000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 88993, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.460110664367676, "TIME_S_1KI": 0.11753857791475371, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1554.5358743476868, "W": 116.24, "J_1KI": 17.468069110465844, "W_1KI": 1.3061701482139043, "W_D": 80.32, "J_D": 1074.1596819305419, "W_D_1KI": 0.9025428966323192, "J_D_1KI": 0.010141729086920537}

105
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_50000_0.0001.output Normal file
View File

@ -0,0 +1,105 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '50000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 0.1613328456878662}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 9, ..., 249987, 249990,
250000]),
col_indices=tensor([ 2831, 11435, 18332, ..., 36257, 39398, 40541]),
values=tensor([0.1158, 0.5239, 0.2299, ..., 0.2166, 0.7808, 0.4412]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.7586, 0.4736, 0.7326, ..., 0.5631, 0.8162, 0.2413])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 0.1613328456878662 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '65082', '-ss', '50000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 8.079791784286499}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 7, ..., 249989, 249995,
250000]),
col_indices=tensor([ 9506, 10457, 11174, ..., 14178, 16522, 25750]),
values=tensor([0.5729, 0.5279, 0.3744, ..., 0.1961, 0.5511, 0.6709]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.0404, 0.4787, 0.7701, ..., 0.8815, 0.0868, 0.4305])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 8.079791784286499 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '84576', '-ss', '50000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 9.978835582733154}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 7, ..., 249990, 249996,
250000]),
col_indices=tensor([26217, 28400, 13678, ..., 15637, 35417, 48424]),
values=tensor([0.3837, 0.9571, 0.9616, ..., 0.3970, 0.1960, 0.8766]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.6737, 0.6555, 0.0878, ..., 0.0726, 0.6482, 0.1469])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 9.978835582733154 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '88993', '-ss', '50000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.460110664367676}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 12, ..., 249992, 249993,
250000]),
col_indices=tensor([ 7470, 20811, 24121, ..., 36968, 38743, 40607]),
values=tensor([0.2685, 0.7271, 0.6618, ..., 0.0403, 0.7886, 0.4035]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.0454, 0.0390, 0.3317, ..., 0.3195, 0.9524, 0.5758])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 10.460110664367676 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 12, ..., 249992, 249993,
250000]),
col_indices=tensor([ 7470, 20811, 24121, ..., 36968, 38743, 40607]),
values=tensor([0.2685, 0.7271, 0.6618, ..., 0.0403, 0.7886, 0.4035]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.0454, 0.0390, 0.3317, ..., 0.3195, 0.9524, 0.5758])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 10.460110664367676 seconds
[40.38, 39.72, 39.57, 39.21, 40.98, 39.68, 39.65, 39.04, 39.23, 43.6]
[116.24]
13.373502016067505
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 88993, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 250000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.460110664367676, 'TIME_S_1KI': 0.11753857791475371, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1554.5358743476868, 'W': 116.24}
[40.38, 39.72, 39.57, 39.21, 40.98, 39.68, 39.65, 39.04, 39.23, 43.6, 39.84, 40.48, 39.37, 39.22, 39.24, 38.98, 44.18, 39.09, 39.18, 39.34]
718.4
35.92
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 88993, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 250000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.460110664367676, 'TIME_S_1KI': 0.11753857791475371, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1554.5358743476868, 'W': 116.24, 'J_1KI': 17.468069110465844, 'W_1KI': 1.3061701482139043, 'W_D': 80.32, 'J_D': 1074.1596819305419, 'W_D_1KI': 0.9025428966323192, 'J_D_1KI': 0.010141729086920537}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_50000_0.001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 46287, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.001, "TIME_S": 11.077528476715088, "TIME_S_1KI": 0.23932267108940064, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2013.3453536748884, "W": 148.1, "J_1KI": 43.49699383573981, "W_1KI": 3.1996024801780196, "W_D": 112.52425, "J_D": 1529.7108434385657, "W_D_1KI": 2.4310119471989973, "J_D_1KI": 0.052520404156652996}

85
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_50000_0.001.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '50000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.001, "TIME_S": 0.2967829704284668}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 60, 108, ..., 2499894,
2499951, 2500000]),
col_indices=tensor([ 368, 1693, 4088, ..., 44885, 46596, 47442]),
values=tensor([0.5982, 0.3592, 0.7042, ..., 0.6155, 0.2314, 0.2925]),
size=(50000, 50000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.7227, 0.5816, 0.4934, ..., 0.3583, 0.6407, 0.9822])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 2500000
Density: 0.001
Time: 0.2967829704284668 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '35379', '-ss', '50000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.001, "TIME_S": 8.025555610656738}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 48, 96, ..., 2499914,
2499959, 2500000]),
col_indices=tensor([ 123, 723, 909, ..., 47588, 48779, 49819]),
values=tensor([0.6654, 0.3505, 0.8901, ..., 0.8476, 0.5107, 0.1185]),
size=(50000, 50000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.6658, 0.6242, 0.4020, ..., 0.5009, 0.1451, 0.6481])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 2500000
Density: 0.001
Time: 8.025555610656738 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '46287', '-ss', '50000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.001, "TIME_S": 11.077528476715088}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 45, 94, ..., 2499903,
2499951, 2500000]),
col_indices=tensor([ 506, 1320, 4404, ..., 49283, 49651, 49966]),
values=tensor([0.0094, 0.0130, 0.0811, ..., 0.5846, 0.7695, 0.1584]),
size=(50000, 50000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.8952, 0.2999, 0.6108, ..., 0.3758, 0.9662, 0.9596])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 2500000
Density: 0.001
Time: 11.077528476715088 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 45, 94, ..., 2499903,
2499951, 2500000]),
col_indices=tensor([ 506, 1320, 4404, ..., 49283, 49651, 49966]),
values=tensor([0.0094, 0.0130, 0.0811, ..., 0.5846, 0.7695, 0.1584]),
size=(50000, 50000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.8952, 0.2999, 0.6108, ..., 0.3758, 0.9662, 0.9596])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 2500000
Density: 0.001
Time: 11.077528476715088 seconds
[40.89, 39.29, 39.33, 39.28, 39.26, 39.19, 39.19, 40.42, 39.41, 39.34]
[148.1]
13.594499349594116
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 46287, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 11.077528476715088, 'TIME_S_1KI': 0.23932267108940064, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2013.3453536748884, 'W': 148.1}
[40.89, 39.29, 39.33, 39.28, 39.26, 39.19, 39.19, 40.42, 39.41, 39.34, 40.01, 40.0, 39.66, 39.57, 39.25, 39.46, 39.33, 39.3, 39.25, 40.41]
711.5149999999999
35.57574999999999
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 46287, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 11.077528476715088, 'TIME_S_1KI': 0.23932267108940064, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2013.3453536748884, 'W': 148.1, 'J_1KI': 43.49699383573981, 'W_1KI': 3.1996024801780196, 'W_D': 112.52425, 'J_D': 1529.7108434385657, 'W_D_1KI': 2.4310119471989973, 'J_D_1KI': 0.052520404156652996}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_50000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 126164, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.496079683303833, "TIME_S_1KI": 0.08319393553869434, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1339.4068178844452, "W": 103.41, "J_1KI": 10.616394675854009, "W_1KI": 0.8196474430106845, "W_D": 67.667, "J_D": 876.4494840517044, "W_D_1KI": 0.5363415871405472, "J_D_1KI": 0.0042511460253364455}

81
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_50000_1e-05.output Normal file
View File

@ -0,0 +1,81 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '50000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.13353276252746582}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 24999, 25000, 25000]),
col_indices=tensor([10989, 5739, 28866, ..., 21823, 4005, 34886]),
values=tensor([0.4353, 0.4497, 0.0871, ..., 0.0925, 0.2903, 0.5435]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.7562, 0.8922, 0.4564, ..., 0.1486, 0.4797, 0.4813])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 0.13353276252746582 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '78632', '-ss', '50000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 6.544129848480225}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 24999, 24999, 25000]),
col_indices=tensor([34114, 35224, 10296, ..., 13464, 985, 3770]),
values=tensor([0.2384, 0.3975, 0.4000, ..., 0.4541, 0.7785, 0.5313]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.2311, 0.0634, 0.6873, ..., 0.2883, 0.1765, 0.0650])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 6.544129848480225 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '126164', '-ss', '50000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.496079683303833}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 25000, 25000, 25000]),
col_indices=tensor([ 3707, 41195, 46820, ..., 24919, 16438, 24153]),
values=tensor([0.4077, 0.2091, 0.6369, ..., 0.9924, 0.1508, 0.5036]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.4831, 0.5861, 0.9166, ..., 0.7031, 0.1228, 0.1244])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 10.496079683303833 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 25000, 25000, 25000]),
col_indices=tensor([ 3707, 41195, 46820, ..., 24919, 16438, 24153]),
values=tensor([0.4077, 0.2091, 0.6369, ..., 0.9924, 0.1508, 0.5036]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.4831, 0.5861, 0.9166, ..., 0.7031, 0.1228, 0.1244])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 10.496079683303833 seconds
[41.91, 39.37, 39.94, 39.02, 39.81, 39.1, 40.39, 38.98, 39.08, 39.0]
[103.41]
12.952391624450684
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 126164, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 25000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.496079683303833, 'TIME_S_1KI': 0.08319393553869434, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1339.4068178844452, 'W': 103.41}
[41.91, 39.37, 39.94, 39.02, 39.81, 39.1, 40.39, 38.98, 39.08, 39.0, 39.67, 39.09, 39.31, 39.83, 39.31, 39.25, 39.43, 38.89, 44.16, 39.22]
714.86
35.743
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 126164, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 25000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.496079683303833, 'TIME_S_1KI': 0.08319393553869434, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1339.4068178844452, 'W': 103.41, 'J_1KI': 10.616394675854009, 'W_1KI': 0.8196474430106845, 'W_D': 67.667, 'J_D': 876.4494840517044, 'W_D_1KI': 0.5363415871405472, 'J_D_1KI': 0.0042511460253364455}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_20_10_10_synthetic_30000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 250038, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 22.432795524597168, "TIME_S_1KI": 0.08971754503154387, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2606.0890457463265, "W": 109.03, "J_1KI": 10.422771921653215, "W_1KI": 0.4360537198345852, "W_D": 73.72525, "J_D": 1762.2174302477242, "W_D_1KI": 0.2948561818603573, "J_D_1KI": 0.0011792454821281456}

81
pytorch/output_synthetic_16core/epyc_7313p_16_csr_20_10_10_synthetic_30000_0.0001.output Normal file
View File

@ -0,0 +1,81 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '30000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 0.10453343391418457}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 3, ..., 89993, 89997, 90000]),
col_indices=tensor([20651, 24290, 28771, ..., 10287, 15356, 24487]),
values=tensor([0.1253, 0.8320, 0.5079, ..., 0.2152, 0.2753, 0.6533]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.9310, 0.8886, 0.9050, ..., 0.7990, 0.2751, 0.5722])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 0.10453343391418457 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '200892', '-ss', '30000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 16.872318267822266}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 5, ..., 89991, 89992, 90000]),
col_indices=tensor([ 9009, 16842, 24312, ..., 27764, 28622, 29005]),
values=tensor([0.8393, 0.9269, 0.8193, ..., 0.0379, 0.8842, 0.8625]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.6604, 0.9619, 0.4104, ..., 0.2632, 0.2079, 0.2105])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 16.872318267822266 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '250038', '-ss', '30000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 22.432795524597168}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 6, ..., 89997, 89998, 90000]),
col_indices=tensor([12588, 20450, 20704, ..., 21668, 10676, 12342]),
values=tensor([0.6372, 0.0652, 0.9949, ..., 0.3492, 0.9239, 0.3604]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.0127, 0.8502, 0.1682, ..., 0.0608, 0.3685, 0.2970])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 22.432795524597168 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 6, ..., 89997, 89998, 90000]),
col_indices=tensor([12588, 20450, 20704, ..., 21668, 10676, 12342]),
values=tensor([0.6372, 0.0652, 0.9949, ..., 0.3492, 0.9239, 0.3604]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.0127, 0.8502, 0.1682, ..., 0.0608, 0.3685, 0.2970])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 22.432795524597168 seconds
[40.69, 39.01, 39.44, 38.94, 38.95, 39.35, 38.98, 39.05, 38.97, 38.87]
[109.03]
23.90249514579773
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 250038, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 90000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 22.432795524597168, 'TIME_S_1KI': 0.08971754503154387, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2606.0890457463265, 'W': 109.03}
[40.69, 39.01, 39.44, 38.94, 38.95, 39.35, 38.98, 39.05, 38.97, 38.87, 40.47, 39.15, 39.52, 39.41, 39.16, 39.78, 39.02, 38.95, 38.92, 38.96]
706.095
35.30475
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 250038, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 90000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 22.432795524597168, 'TIME_S_1KI': 0.08971754503154387, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2606.0890457463265, 'W': 109.03, 'J_1KI': 10.422771921653215, 'W_1KI': 0.4360537198345852, 'W_D': 73.72525, 'J_D': 1762.2174302477242, 'W_D_1KI': 0.2948561818603573, 'J_D_1KI': 0.0011792454821281456}

0
pytorch/output_synthetic_16core/epyc_7313p_16_csr_20_10_10_synthetic_30000_0.001.json Normal file
View File

21
pytorch/output_synthetic_16core/epyc_7313p_16_csr_20_10_10_synthetic_30000_0.001.output Normal file
View File

@ -0,0 +1,21 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '30000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 900000, "MATRIX_DENSITY": 0.001, "TIME_S": 0.1439976692199707}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 29, 67, ..., 899936, 899964,
900000]),
col_indices=tensor([ 58, 341, 3959, ..., 27670, 28034, 29816]),
values=tensor([0.8286, 0.0691, 0.1730, ..., 0.2645, 0.7295, 0.5386]),
size=(30000, 30000), nnz=900000, layout=torch.sparse_csr)
tensor([0.0558, 0.4553, 0.9674, ..., 0.2366, 0.6209, 0.6160])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 900000
Density: 0.001
Time: 0.1439976692199707 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '145835', '-ss', '30000', '-sd', '0.001', '-c', '16']

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_20_10_10_synthetic_30000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 321850, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 20.594725370407104, "TIME_S_1KI": 0.06398858278827747, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2327.1387594389917, "W": 101.41000000000001, "J_1KI": 7.230507253189348, "W_1KI": 0.3150846667702346, "W_D": 65.9145, "J_D": 1512.5942979887725, "W_D_1KI": 0.20479881932577287, "J_D_1KI": 0.0006363175992722476}

81
pytorch/output_synthetic_16core/epyc_7313p_16_csr_20_10_10_synthetic_30000_1e-05.output Normal file
View File

@ -0,0 +1,81 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '30000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.08333611488342285}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9000, 9000, 9000]),
col_indices=tensor([13464, 15002, 12998, ..., 1674, 7890, 9839]),
values=tensor([0.3937, 0.5826, 0.6728, ..., 0.2443, 0.0810, 0.3168]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.3767, 0.3322, 0.0921, ..., 0.4449, 0.8687, 0.6223])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 0.08333611488342285 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '251991', '-ss', '30000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 16.441835403442383}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9000, 9000, 9000]),
col_indices=tensor([ 1592, 26221, 2007, ..., 5499, 7511, 18290]),
values=tensor([0.1009, 0.0773, 0.0762, ..., 0.6540, 0.2265, 0.9524]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.5719, 0.1239, 0.1698, ..., 0.8424, 0.3509, 0.9636])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 16.441835403442383 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '321850', '-ss', '30000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 20.594725370407104}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9000, 9000, 9000]),
col_indices=tensor([28655, 14046, 22660, ..., 19793, 14001, 26576]),
values=tensor([0.0604, 0.3035, 0.4856, ..., 0.8323, 0.7946, 0.0096]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.2670, 0.6630, 0.3861, ..., 0.4215, 0.9031, 0.7574])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 20.594725370407104 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9000, 9000, 9000]),
col_indices=tensor([28655, 14046, 22660, ..., 19793, 14001, 26576]),
values=tensor([0.0604, 0.3035, 0.4856, ..., 0.8323, 0.7946, 0.0096]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.2670, 0.6630, 0.3861, ..., 0.4215, 0.9031, 0.7574])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 20.594725370407104 seconds
[39.56, 39.04, 39.2, 38.58, 39.16, 39.36, 38.83, 40.63, 38.67, 39.16]
[101.41]
22.94782328605652
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 321850, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 9000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 20.594725370407104, 'TIME_S_1KI': 0.06398858278827747, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2327.1387594389917, 'W': 101.41000000000001}
[39.56, 39.04, 39.2, 38.58, 39.16, 39.36, 38.83, 40.63, 38.67, 39.16, 39.61, 44.11, 38.89, 39.27, 38.75, 38.81, 40.71, 38.62, 38.81, 38.61]
709.91
35.4955
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 321850, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 9000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 20.594725370407104, 'TIME_S_1KI': 0.06398858278827747, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2327.1387594389917, 'W': 101.41000000000001, 'J_1KI': 7.230507253189348, 'W_1KI': 0.3150846667702346, 'W_D': 65.9145, 'J_D': 1512.5942979887725, 'W_D_1KI': 0.20479881932577287, 'J_D_1KI': 0.0006363175992722476}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_100000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 33012, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.519692420959473, "TIME_S_1KI": 0.318662680872394, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1224.664799463749, "W": 88.39, "J_1KI": 37.09756450574788, "W_1KI": 2.677511208045559, "W_D": 72.108, "J_D": 999.0737567567826, "W_D_1KI": 2.184296619411123, "J_D_1KI": 0.06616674601390778}

65
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_100000_0.0001.output Normal file
View File

@ -0,0 +1,65 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '100000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 0.3180568218231201}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 999979,
999991, 1000000]),
col_indices=tensor([10691, 12782, 14246, ..., 70658, 88202, 93324]),
values=tensor([0.3844, 0.6658, 0.7124, ..., 0.3153, 0.8920, 0.6509]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.9202, 0.9151, 0.8232, ..., 0.5628, 0.6151, 0.8368])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 0.3180568218231201 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '33012', '-ss', '100000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.519692420959473}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 15, ..., 999984,
999990, 1000000]),
col_indices=tensor([ 7405, 49048, 69982, ..., 87685, 98650, 99933]),
values=tensor([0.6053, 0.2022, 0.4562, ..., 0.3977, 0.5709, 0.7435]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.0490, 0.1129, 0.5767, ..., 0.3037, 0.9982, 0.0194])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 10.519692420959473 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 15, ..., 999984,
999990, 1000000]),
col_indices=tensor([ 7405, 49048, 69982, ..., 87685, 98650, 99933]),
values=tensor([0.6053, 0.2022, 0.4562, ..., 0.3977, 0.5709, 0.7435]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.0490, 0.1129, 0.5767, ..., 0.3037, 0.9982, 0.0194])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 10.519692420959473 seconds
[18.25, 17.99, 18.13, 17.81, 18.01, 17.82, 18.18, 18.25, 17.98, 18.75]
[88.39]
13.855241537094116
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 33012, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.519692420959473, 'TIME_S_1KI': 0.318662680872394, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1224.664799463749, 'W': 88.39}
[18.25, 17.99, 18.13, 17.81, 18.01, 17.82, 18.18, 18.25, 17.98, 18.75, 18.55, 17.94, 18.05, 17.81, 18.35, 17.79, 18.28, 18.36, 18.2, 17.83]
325.64
16.282
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 33012, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.519692420959473, 'TIME_S_1KI': 0.318662680872394, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1224.664799463749, 'W': 88.39, 'J_1KI': 37.09756450574788, 'W_1KI': 2.677511208045559, 'W_D': 72.108, 'J_D': 999.0737567567826, 'W_D_1KI': 2.184296619411123, 'J_D_1KI': 0.06616674601390778}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_100000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 64591, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.66994047164917, "TIME_S_1KI": 0.16519237156336286, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1159.8280617713929, "W": 82.8, "J_1KI": 17.9564964433341, "W_1KI": 1.2819123407285846, "W_D": 66.57124999999999, "J_D": 932.5024620434641, "W_D_1KI": 1.0306582960474369, "J_D_1KI": 0.015956685854800777}

85
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_100000_1e-05.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '100000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.17906904220581055}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 99999, 99999,
100000]),
col_indices=tensor([85471, 5444, 13434, ..., 17615, 87992, 83918]),
values=tensor([0.7119, 0.1219, 0.2242, ..., 0.7199, 0.3920, 0.9751]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.8861, 0.1716, 0.8373, ..., 0.2826, 0.6276, 0.0027])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 0.17906904220581055 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '58636', '-ss', '100000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 9.531909704208374}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 99999, 100000,
100000]),
col_indices=tensor([28875, 86601, 1118, ..., 53659, 98581, 89346]),
values=tensor([0.0170, 0.0837, 0.6677, ..., 0.0775, 0.7543, 0.4196]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.4702, 0.4277, 0.7376, ..., 0.9470, 0.3873, 0.6416])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 9.531909704208374 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '64591', '-ss', '100000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.66994047164917}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 100000, 100000,
100000]),
col_indices=tensor([32373, 45973, 94969, ..., 5823, 12968, 35562]),
values=tensor([0.6698, 0.7885, 0.1863, ..., 0.4943, 0.2796, 0.7613]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.4737, 0.5533, 0.8139, ..., 0.3662, 0.3156, 0.7007])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 10.66994047164917 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 100000, 100000,
100000]),
col_indices=tensor([32373, 45973, 94969, ..., 5823, 12968, 35562]),
values=tensor([0.6698, 0.7885, 0.1863, ..., 0.4943, 0.2796, 0.7613]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.4737, 0.5533, 0.8139, ..., 0.3662, 0.3156, 0.7007])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 10.66994047164917 seconds
[18.59, 17.9, 18.34, 17.96, 18.14, 17.94, 18.32, 17.79, 17.82, 17.71]
[82.8]
14.007585287094116
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 64591, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.66994047164917, 'TIME_S_1KI': 0.16519237156336286, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1159.8280617713929, 'W': 82.8}
[18.59, 17.9, 18.34, 17.96, 18.14, 17.94, 18.32, 17.79, 17.82, 17.71, 18.29, 17.85, 18.6, 17.8, 18.13, 17.74, 17.99, 17.83, 18.16, 17.94]
324.57500000000005
16.22875
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 64591, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.66994047164917, 'TIME_S_1KI': 0.16519237156336286, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1159.8280617713929, 'W': 82.8, 'J_1KI': 17.9564964433341, 'W_1KI': 1.2819123407285846, 'W_D': 66.57124999999999, 'J_D': 932.5024620434641, 'W_D_1KI': 1.0306582960474369, 'J_D_1KI': 0.015956685854800777}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 250193, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.988550901412964, "TIME_S_1KI": 0.043920297136262665, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1067.7657339859009, "W": 74.48, "J_1KI": 4.267768218878628, "W_1KI": 0.29769018317858614, "W_D": 58.048, "J_D": 832.1920693664551, "W_D_1KI": 0.23201288605196788, "J_D_1KI": 0.0009273356410929478}

81
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.0001.output Normal file
View File

@ -0,0 +1,81 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 0.06029987335205078}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9998, 9999, 10000]),
col_indices=tensor([9584, 2249, 9621, ..., 267, 2843, 1232]),
values=tensor([0.1887, 0.8280, 0.8733, ..., 0.6422, 0.8241, 0.9503]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.2203, 0.8610, 0.9153, ..., 0.2931, 0.9983, 0.3156])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 0.06029987335205078 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '174129', '-ss', '10000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 7.307769536972046}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 10000, 10000, 10000]),
col_indices=tensor([5050, 9096, 467, ..., 6460, 6547, 2963]),
values=tensor([0.3312, 0.9984, 0.8182, ..., 0.5509, 0.3722, 0.7285]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.0543, 0.3720, 0.3677, ..., 0.5280, 0.6433, 0.3148])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 7.307769536972046 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '250193', '-ss', '10000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.988550901412964}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9997, 9999, 10000]),
col_indices=tensor([4233, 4275, 7541, ..., 2248, 7833, 717]),
values=tensor([0.0347, 0.7995, 0.4404, ..., 0.0217, 0.2651, 0.9390]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.4739, 0.4789, 0.6628, ..., 0.7267, 0.9323, 0.5704])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 10.988550901412964 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9997, 9999, 10000]),
col_indices=tensor([4233, 4275, 7541, ..., 2248, 7833, 717]),
values=tensor([0.0347, 0.7995, 0.4404, ..., 0.0217, 0.2651, 0.9390]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.4739, 0.4789, 0.6628, ..., 0.7267, 0.9323, 0.5704])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 10.988550901412964 seconds
[18.5, 18.04, 18.08, 20.55, 18.03, 18.27, 18.34, 17.92, 18.14, 18.0]
[74.48]
14.33627462387085
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 250193, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.988550901412964, 'TIME_S_1KI': 0.043920297136262665, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1067.7657339859009, 'W': 74.48}
[18.5, 18.04, 18.08, 20.55, 18.03, 18.27, 18.34, 17.92, 18.14, 18.0, 18.31, 18.29, 18.5, 18.09, 18.0, 17.95, 17.89, 18.08, 18.14, 17.85]
328.64
16.432
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 250193, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.988550901412964, 'TIME_S_1KI': 0.043920297136262665, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1067.7657339859009, 'W': 74.48, 'J_1KI': 4.267768218878628, 'W_1KI': 0.29769018317858614, 'W_D': 58.048, 'J_D': 832.1920693664551, 'W_D_1KI': 0.23201288605196788, 'J_D_1KI': 0.0009273356410929478}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 186516, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 10.228839635848999, "TIME_S_1KI": 0.054841620214078145, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1075.814607758522, "W": 79.58, "J_1KI": 5.767948099672533, "W_1KI": 0.4266658088314139, "W_D": 63.054, "J_D": 852.4053063282967, "W_D_1KI": 0.338062150164061, "J_D_1KI": 0.0018125101876732344}

85
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.001.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 0.0709388256072998}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 9, 17, ..., 99974, 99988,
100000]),
col_indices=tensor([1106, 1398, 2518, ..., 6886, 7547, 8173]),
values=tensor([0.5902, 0.0057, 0.8492, ..., 0.2608, 0.7269, 0.6940]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.8144, 0.0674, 0.1585, ..., 0.0850, 0.2846, 0.5370])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 0.0709388256072998 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '148014', '-ss', '10000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 8.332475900650024}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 17, ..., 99965, 99978,
100000]),
col_indices=tensor([ 77, 628, 3642, ..., 8176, 8481, 9600]),
values=tensor([0.7580, 0.3721, 0.0885, ..., 0.9345, 0.1388, 0.5730]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.9678, 0.5744, 0.4262, ..., 0.2115, 0.3242, 0.5272])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 8.332475900650024 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '186516', '-ss', '10000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 10.228839635848999}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 30, ..., 99982, 99990,
100000]),
col_indices=tensor([ 298, 367, 1190, ..., 3689, 6850, 7173]),
values=tensor([0.7086, 0.6908, 0.8648, ..., 0.4576, 0.3199, 0.8368]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.7366, 0.0593, 0.8663, ..., 0.2557, 0.4256, 0.5242])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 10.228839635848999 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 30, ..., 99982, 99990,
100000]),
col_indices=tensor([ 298, 367, 1190, ..., 3689, 6850, 7173]),
values=tensor([0.7086, 0.6908, 0.8648, ..., 0.4576, 0.3199, 0.8368]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.7366, 0.0593, 0.8663, ..., 0.2557, 0.4256, 0.5242])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 10.228839635848999 seconds
[18.35, 18.1, 18.16, 18.05, 17.94, 18.34, 18.01, 17.89, 17.93, 17.71]
[79.58]
13.51865553855896
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 186516, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 10.228839635848999, 'TIME_S_1KI': 0.054841620214078145, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1075.814607758522, 'W': 79.58}
[18.35, 18.1, 18.16, 18.05, 17.94, 18.34, 18.01, 17.89, 17.93, 17.71, 19.07, 18.27, 18.23, 17.94, 18.27, 18.14, 21.36, 18.53, 18.63, 18.33]
330.52
16.526
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 186516, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 10.228839635848999, 'TIME_S_1KI': 0.054841620214078145, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1075.814607758522, 'W': 79.58, 'J_1KI': 5.767948099672533, 'W_1KI': 0.4266658088314139, 'W_D': 63.054, 'J_D': 852.4053063282967, 'W_D_1KI': 0.338062150164061, 'J_D_1KI': 0.0018125101876732344}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.01.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 57497, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 10.399010181427002, "TIME_S_1KI": 0.18086178724849997, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1216.1904487371446, "W": 87.17000000000002, "J_1KI": 21.15224183413299, "W_1KI": 1.5160790997791191, "W_D": 70.89300000000001, "J_D": 989.0947514319422, "W_D_1KI": 1.2329860688383745, "J_D_1KI": 0.021444354815701245}

85
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.01.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '0.01', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 0.1964414119720459}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 102, 210, ..., 999804,
999909, 1000000]),
col_indices=tensor([ 4, 297, 328, ..., 9417, 9717, 9744]),
values=tensor([0.3827, 0.2830, 0.2497, ..., 0.1291, 0.2102, 0.5312]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.7948, 0.9855, 0.6473, ..., 0.4205, 0.5296, 0.9253])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 0.1964414119720459 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '53451', '-ss', '10000', '-sd', '0.01', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 9.761078357696533}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 112, 220, ..., 999796,
999898, 1000000]),
col_indices=tensor([ 465, 658, 715, ..., 9500, 9653, 9927]),
values=tensor([0.9513, 0.9158, 0.4499, ..., 0.0775, 0.2496, 0.9759]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.5799, 0.5098, 0.6156, ..., 0.8166, 0.2331, 0.2979])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 9.761078357696533 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '57497', '-ss', '10000', '-sd', '0.01', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 10.399010181427002}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 104, 198, ..., 999802,
999905, 1000000]),
col_indices=tensor([ 124, 157, 187, ..., 9539, 9601, 9680]),
values=tensor([0.6532, 0.0603, 0.0418, ..., 0.1935, 0.1125, 0.4778]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.5307, 0.8097, 0.3092, ..., 0.4937, 0.1856, 0.7516])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 10.399010181427002 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 104, 198, ..., 999802,
999905, 1000000]),
col_indices=tensor([ 124, 157, 187, ..., 9539, 9601, 9680]),
values=tensor([0.6532, 0.0603, 0.0418, ..., 0.1935, 0.1125, 0.4778]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.5307, 0.8097, 0.3092, ..., 0.4937, 0.1856, 0.7516])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 10.399010181427002 seconds
[18.23, 18.13, 17.96, 18.03, 17.94, 17.9, 17.91, 17.8, 18.15, 18.93]
[87.17]
13.951938152313232
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 57497, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.01, 'TIME_S': 10.399010181427002, 'TIME_S_1KI': 0.18086178724849997, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1216.1904487371446, 'W': 87.17000000000002}
[18.23, 18.13, 17.96, 18.03, 17.94, 17.9, 17.91, 17.8, 18.15, 18.93, 18.34, 17.87, 18.21, 18.15, 18.42, 17.87, 18.22, 18.25, 18.03, 17.9]
325.54
16.277
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 57497, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.01, 'TIME_S': 10.399010181427002, 'TIME_S_1KI': 0.18086178724849997, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1216.1904487371446, 'W': 87.17000000000002, 'J_1KI': 21.15224183413299, 'W_1KI': 1.5160790997791191, 'W_D': 70.89300000000001, 'J_D': 989.0947514319422, 'W_D_1KI': 1.2329860688383745, 'J_D_1KI': 0.021444354815701245}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 9007, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 10.744792222976685, "TIME_S_1KI": 1.192937961915919, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1359.414791688919, "W": 84.86, "J_1KI": 150.92869897734195, "W_1KI": 9.421561008104806, "W_D": 68.55725000000001, "J_D": 1098.2528838971855, "W_D_1KI": 7.611552126124127, "J_D_1KI": 0.8450707367740786}

65
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.05.output Normal file
View File

@ -0,0 +1,65 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '10000', '-sd', '0.05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 1.1656646728515625}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 506, 991, ..., 4998989,
4999492, 5000000]),
col_indices=tensor([ 25, 30, 53, ..., 9970, 9993, 9995]),
values=tensor([0.0157, 0.5603, 0.3033, ..., 0.4419, 0.2413, 0.9606]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.4291, 0.9468, 0.9558, ..., 0.3375, 0.0455, 0.9666])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 1.1656646728515625 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '9007', '-ss', '10000', '-sd', '0.05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 10.744792222976685}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 491, 1013, ..., 4998981,
4999517, 5000000]),
col_indices=tensor([ 61, 62, 77, ..., 9979, 9982, 9988]),
values=tensor([0.6511, 0.9070, 0.7175, ..., 0.4257, 0.4784, 0.0096]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.7046, 0.2172, 0.5779, ..., 0.4690, 0.0165, 0.6122])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 10.744792222976685 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 491, 1013, ..., 4998981,
4999517, 5000000]),
col_indices=tensor([ 61, 62, 77, ..., 9979, 9982, 9988]),
values=tensor([0.6511, 0.9070, 0.7175, ..., 0.4257, 0.4784, 0.0096]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.7046, 0.2172, 0.5779, ..., 0.4690, 0.0165, 0.6122])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 10.744792222976685 seconds
[18.37, 18.68, 18.13, 17.9, 18.06, 18.22, 18.04, 18.49, 17.9, 18.1]
[84.86]
16.019500255584717
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 9007, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.05, 'TIME_S': 10.744792222976685, 'TIME_S_1KI': 1.192937961915919, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1359.414791688919, 'W': 84.86}
[18.37, 18.68, 18.13, 17.9, 18.06, 18.22, 18.04, 18.49, 17.9, 18.1, 18.25, 17.97, 18.05, 17.84, 17.92, 18.2, 17.96, 17.89, 18.19, 18.51]
326.05499999999995
16.302749999999996
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 9007, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.05, 'TIME_S': 10.744792222976685, 'TIME_S_1KI': 1.192937961915919, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1359.414791688919, 'W': 84.86, 'J_1KI': 150.92869897734195, 'W_1KI': 9.421561008104806, 'W_D': 68.55725000000001, 'J_D': 1098.2528838971855, 'W_D_1KI': 7.611552126124127, 'J_D_1KI': 0.8450707367740786}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 279705, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.192691802978516, "TIME_S_1KI": 0.03644086377783206, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1002.2923643112182, "W": 73.24, "J_1KI": 3.5833909451429835, "W_1KI": 0.2618473034089487, "W_D": 56.983999999999995, "J_D": 779.8283463668822, "W_D_1KI": 0.20372892869272982, "J_D_1KI": 0.0007283707073263969}

1521
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_1e-05.output Normal file
View File

File diff suppressed because it is too large Load Diff

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_500000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 8355, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [500000, 500000], "MATRIX_ROWS": 500000, "MATRIX_SIZE": 250000000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.90480637550354, "TIME_S_1KI": 1.305183288510298, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1307.563270778656, "W": 87.44, "J_1KI": 156.5006906976249, "W_1KI": 10.4655894673848, "W_D": 70.932, "J_D": 1060.7053742322921, "W_D_1KI": 8.489766606822261, "J_D_1KI": 1.0161300546765126}

68
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_500000_1e-05.output Normal file
View File

@ -0,0 +1,68 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '500000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [500000, 500000], "MATRIX_ROWS": 500000, "MATRIX_SIZE": 250000000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 1e-05, "TIME_S": 1.2567212581634521}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 2499992,
2499996, 2500000]),
col_indices=tensor([164554, 277712, 289036, ..., 389470, 409865,
491502]),
values=tensor([0.0126, 0.9348, 0.8595, ..., 0.3584, 0.7345, 0.5238]),
size=(500000, 500000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.0175, 0.7668, 0.4852, ..., 0.2657, 0.5513, 0.9738])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([500000, 500000])
Rows: 500000
Size: 250000000000
NNZ: 2500000
Density: 1e-05
Time: 1.2567212581634521 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '8355', '-ss', '500000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [500000, 500000], "MATRIX_ROWS": 500000, "MATRIX_SIZE": 250000000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.90480637550354}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 12, ..., 2499995,
2499997, 2500000]),
col_indices=tensor([ 72448, 73110, 121261, ..., 13350, 176428,
278854]),
values=tensor([0.1918, 0.3445, 0.8471, ..., 0.5873, 0.4603, 0.6922]),
size=(500000, 500000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.8369, 0.9252, 0.2721, ..., 0.2352, 0.7861, 0.2173])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([500000, 500000])
Rows: 500000
Size: 250000000000
NNZ: 2500000
Density: 1e-05
Time: 10.90480637550354 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 12, ..., 2499995,
2499997, 2500000]),
col_indices=tensor([ 72448, 73110, 121261, ..., 13350, 176428,
278854]),
values=tensor([0.1918, 0.3445, 0.8471, ..., 0.5873, 0.4603, 0.6922]),
size=(500000, 500000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.8369, 0.9252, 0.2721, ..., 0.2352, 0.7861, 0.2173])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([500000, 500000])
Rows: 500000
Size: 250000000000
NNZ: 2500000
Density: 1e-05
Time: 10.90480637550354 seconds
[18.48, 21.5, 18.34, 18.43, 18.27, 18.03, 18.28, 18.13, 18.03, 18.78]
[87.44]
14.953834295272827
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 8355, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [500000, 500000], 'MATRIX_ROWS': 500000, 'MATRIX_SIZE': 250000000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.90480637550354, 'TIME_S_1KI': 1.305183288510298, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1307.563270778656, 'W': 87.44}
[18.48, 21.5, 18.34, 18.43, 18.27, 18.03, 18.28, 18.13, 18.03, 18.78, 18.48, 18.09, 18.1, 17.9, 18.11, 18.02, 17.99, 17.84, 18.15, 18.16]
330.15999999999997
16.508
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 8355, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [500000, 500000], 'MATRIX_ROWS': 500000, 'MATRIX_SIZE': 250000000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.90480637550354, 'TIME_S_1KI': 1.305183288510298, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1307.563270778656, 'W': 87.44, 'J_1KI': 156.5006906976249, 'W_1KI': 10.4655894673848, 'W_D': 70.932, 'J_D': 1060.7053742322921, 'W_D_1KI': 8.489766606822261, 'J_D_1KI': 1.0161300546765126}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_50000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 77922, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.570462703704834, "TIME_S_1KI": 0.13565440701861906, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1209.7638923931122, "W": 83.24, "J_1KI": 15.525318811030418, "W_1KI": 1.0682477349144015, "W_D": 66.53899999999999, "J_D": 967.0408413736818, "W_D_1KI": 0.8539180205846871, "J_D_1KI": 0.010958625556129042}

85
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_50000_0.0001.output Normal file
View File

@ -0,0 +1,85 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '50000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 0.14919304847717285}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 9, ..., 249989, 249995,
250000]),
col_indices=tensor([ 8787, 10800, 12548, ..., 22776, 32520, 35593]),
values=tensor([0.0395, 0.0216, 0.0459, ..., 0.9233, 0.0886, 0.1442]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.0084, 0.2765, 0.2672, ..., 0.0856, 0.1416, 0.8826])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 0.14919304847717285 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '70378', '-ss', '50000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 9.483437538146973}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 9, ..., 249988, 249997,
250000]),
col_indices=tensor([ 1665, 9567, 9654, ..., 4112, 18670, 38091]),
values=tensor([0.4890, 0.0494, 0.7903, ..., 0.9513, 0.0590, 0.1377]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.5003, 0.9747, 0.2176, ..., 0.9666, 0.4758, 0.9002])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 9.483437538146973 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '77922', '-ss', '50000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 250000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.570462703704834}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 7, ..., 249995, 249999,
250000]),
col_indices=tensor([18420, 40988, 3727, ..., 33621, 36384, 44487]),
values=tensor([0.1861, 0.8144, 0.1628, ..., 0.4774, 0.5715, 0.3216]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.6272, 0.7644, 0.0884, ..., 0.9496, 0.3089, 0.8679])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 10.570462703704834 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 7, ..., 249995, 249999,
250000]),
col_indices=tensor([18420, 40988, 3727, ..., 33621, 36384, 44487]),
values=tensor([0.1861, 0.8144, 0.1628, ..., 0.4774, 0.5715, 0.3216]),
size=(50000, 50000), nnz=250000, layout=torch.sparse_csr)
tensor([0.6272, 0.7644, 0.0884, ..., 0.9496, 0.3089, 0.8679])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 250000
Density: 0.0001
Time: 10.570462703704834 seconds
[18.59, 17.97, 19.57, 18.3, 18.09, 17.87, 18.02, 21.06, 18.56, 17.89]
[83.24]
14.533444166183472
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 77922, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 250000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.570462703704834, 'TIME_S_1KI': 0.13565440701861906, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1209.7638923931122, 'W': 83.24}
[18.59, 17.97, 19.57, 18.3, 18.09, 17.87, 18.02, 21.06, 18.56, 17.89, 18.2, 17.85, 17.83, 21.57, 17.96, 18.06, 18.27, 18.3, 18.38, 18.04]
334.02
16.701
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 77922, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 250000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.570462703704834, 'TIME_S_1KI': 0.13565440701861906, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1209.7638923931122, 'W': 83.24, 'J_1KI': 15.525318811030418, 'W_1KI': 1.0682477349144015, 'W_D': 66.53899999999999, 'J_D': 967.0408413736818, 'W_D_1KI': 0.8539180205846871, 'J_D_1KI': 0.010958625556129042}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_50000_0.001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 17357, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.001, "TIME_S": 10.690638303756714, "TIME_S_1KI": 0.6159266177194627, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1295.2395059108735, "W": 87.56, "J_1KI": 74.62346637730447, "W_1KI": 5.044650573255747, "W_D": 71.326, "J_D": 1055.0965394997595, "W_D_1KI": 4.109350694244396, "J_D_1KI": 0.23675466349279234}

65
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_50000_0.001.output Normal file
View File

@ -0,0 +1,65 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '50000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.001, "TIME_S": 0.6049323081970215}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 46, 103, ..., 2499893,
2499950, 2500000]),
col_indices=tensor([ 214, 217, 3424, ..., 47339, 47927, 48505]),
values=tensor([0.8463, 0.5755, 0.1058, ..., 0.4565, 0.0843, 0.4040]),
size=(50000, 50000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.2070, 0.0126, 0.4112, ..., 0.3463, 0.8132, 0.3234])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 2500000
Density: 0.001
Time: 0.6049323081970215 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '17357', '-ss', '50000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.001, "TIME_S": 10.690638303756714}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 47, 101, ..., 2499901,
2499949, 2500000]),
col_indices=tensor([ 511, 725, 819, ..., 47217, 48788, 49222]),
values=tensor([0.0511, 0.3894, 0.2647, ..., 0.8233, 0.9615, 0.4045]),
size=(50000, 50000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.7118, 0.9063, 0.7110, ..., 0.7333, 0.4959, 0.7807])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 2500000
Density: 0.001
Time: 10.690638303756714 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 47, 101, ..., 2499901,
2499949, 2500000]),
col_indices=tensor([ 511, 725, 819, ..., 47217, 48788, 49222]),
values=tensor([0.0511, 0.3894, 0.2647, ..., 0.8233, 0.9615, 0.4045]),
size=(50000, 50000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.7118, 0.9063, 0.7110, ..., 0.7333, 0.4959, 0.7807])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 2500000
Density: 0.001
Time: 10.690638303756714 seconds
[18.27, 17.74, 18.29, 17.77, 18.01, 17.85, 17.88, 17.79, 18.2, 17.88]
[87.56]
14.792593717575073
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 17357, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 10.690638303756714, 'TIME_S_1KI': 0.6159266177194627, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1295.2395059108735, 'W': 87.56}
[18.27, 17.74, 18.29, 17.77, 18.01, 17.85, 17.88, 17.79, 18.2, 17.88, 18.2, 18.34, 18.01, 17.89, 18.13, 18.46, 18.22, 17.9, 17.95, 18.15]
324.68
16.234
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 17357, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 10.690638303756714, 'TIME_S_1KI': 0.6159266177194627, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1295.2395059108735, 'W': 87.56, 'J_1KI': 74.62346637730447, 'W_1KI': 5.044650573255747, 'W_D': 71.326, 'J_D': 1055.0965394997595, 'W_D_1KI': 4.109350694244396, 'J_D_1KI': 0.23675466349279234}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_50000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 112508, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.720516443252563, "TIME_S_1KI": 0.09528670355221462, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1090.2722121667862, "W": 75.82, "J_1KI": 9.69061944187779, "W_1KI": 0.6739076332349699, "W_D": 59.38549999999999, "J_D": 853.9483046113252, "W_D_1KI": 0.5278335762790201, "J_D_1KI": 0.004691520392141182}

81
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_50000_1e-05.output Normal file
View File

@ -0,0 +1,81 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '50000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.11333847045898438}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 2, ..., 24996, 24999, 25000]),
col_indices=tensor([ 9502, 18497, 7204, ..., 33396, 45910, 109]),
values=tensor([0.5325, 0.6011, 0.4727, ..., 0.6967, 0.0269, 0.7415]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.7210, 0.8240, 0.5786, ..., 0.5702, 0.4441, 0.2533])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 0.11333847045898438 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '92642', '-ss', '50000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 8.645956993103027}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 25000, 25000, 25000]),
col_indices=tensor([35285, 1305, 12700, ..., 6399, 17561, 45264]),
values=tensor([0.6896, 0.7157, 0.5414, ..., 0.3157, 0.2585, 0.8046]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.8892, 0.5178, 0.0901, ..., 0.0600, 0.1718, 0.0275])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 8.645956993103027 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '112508', '-ss', '50000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [50000, 50000], "MATRIX_ROWS": 50000, "MATRIX_SIZE": 2500000000, "MATRIX_NNZ": 25000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.720516443252563}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 25000, 25000, 25000]),
col_indices=tensor([27684, 39939, 2715, ..., 47308, 11944, 42221]),
values=tensor([0.6561, 0.5911, 0.5622, ..., 0.2806, 0.4491, 0.6100]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.8957, 0.2813, 0.1993, ..., 0.7019, 0.9944, 0.8970])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 10.720516443252563 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 25000, 25000, 25000]),
col_indices=tensor([27684, 39939, 2715, ..., 47308, 11944, 42221]),
values=tensor([0.6561, 0.5911, 0.5622, ..., 0.2806, 0.4491, 0.6100]),
size=(50000, 50000), nnz=25000, layout=torch.sparse_csr)
tensor([0.8957, 0.2813, 0.1993, ..., 0.7019, 0.9944, 0.8970])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([50000, 50000])
Rows: 50000
Size: 2500000000
NNZ: 25000
Density: 1e-05
Time: 10.720516443252563 seconds
[18.48, 18.01, 18.17, 19.64, 18.03, 18.42, 18.45, 18.1, 18.26, 18.04]
[75.82]
14.379744291305542
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 112508, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 25000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.720516443252563, 'TIME_S_1KI': 0.09528670355221462, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1090.2722121667862, 'W': 75.82}
[18.48, 18.01, 18.17, 19.64, 18.03, 18.42, 18.45, 18.1, 18.26, 18.04, 18.62, 18.47, 18.04, 18.06, 17.8, 18.08, 18.08, 18.67, 17.82, 18.04]
328.69000000000005
16.434500000000003
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 112508, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [50000, 50000], 'MATRIX_ROWS': 50000, 'MATRIX_SIZE': 2500000000, 'MATRIX_NNZ': 25000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.720516443252563, 'TIME_S_1KI': 0.09528670355221462, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1090.2722121667862, 'W': 75.82, 'J_1KI': 9.69061944187779, 'W_1KI': 0.6739076332349699, 'W_D': 59.38549999999999, 'J_D': 853.9483046113252, 'W_D_1KI': 0.5278335762790201, 'J_D_1KI': 0.004691520392141182}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_20_10_10_synthetic_30000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 234425, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 21.312235116958618, "TIME_S_1KI": 0.09091280843322434, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2054.109435153008, "W": 83.45000000000002, "J_1KI": 8.762330959381499, "W_1KI": 0.3559773914898156, "W_D": 67.21450000000002, "J_D": 1654.4749985511307, "W_D_1KI": 0.2867206995840888, "J_D_1KI": 0.0012230807276702091}

81
pytorch/output_synthetic_16core/xeon_4216_16_csr_20_10_10_synthetic_30000_0.0001.output Normal file
View File

@ -0,0 +1,81 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '30000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 0.10643196105957031}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 9, ..., 89992, 89998, 90000]),
col_indices=tensor([ 7924, 12206, 12582, ..., 21107, 10373, 19571]),
values=tensor([0.8274, 0.6462, 0.9289, ..., 0.2542, 0.4328, 0.6143]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.4141, 0.4229, 0.5665, ..., 0.1440, 0.7095, 0.1472])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 0.10643196105957031 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '197309', '-ss', '30000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 17.675063133239746}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 9, ..., 89992, 89999, 90000]),
col_indices=tensor([ 929, 2315, 11088, ..., 21381, 23338, 19838]),
values=tensor([0.3872, 0.2873, 0.0227, ..., 0.4746, 0.4839, 0.3522]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.1013, 0.5431, 0.3309, ..., 0.2751, 0.1147, 0.0007])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 17.675063133239746 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '234425', '-ss', '30000', '-sd', '0.0001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 90000, "MATRIX_DENSITY": 0.0001, "TIME_S": 21.312235116958618}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 10, ..., 89994, 89997, 90000]),
col_indices=tensor([ 6200, 14122, 21980, ..., 11781, 19689, 21155]),
values=tensor([0.5859, 0.7824, 0.3581, ..., 0.7747, 0.1479, 0.5181]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.3266, 0.0767, 0.6789, ..., 0.9087, 0.9799, 0.8849])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 21.312235116958618 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 10, ..., 89994, 89997, 90000]),
col_indices=tensor([ 6200, 14122, 21980, ..., 11781, 19689, 21155]),
values=tensor([0.5859, 0.7824, 0.3581, ..., 0.7747, 0.1479, 0.5181]),
size=(30000, 30000), nnz=90000, layout=torch.sparse_csr)
tensor([0.3266, 0.0767, 0.6789, ..., 0.9087, 0.9799, 0.8849])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 90000
Density: 0.0001
Time: 21.312235116958618 seconds
[18.37, 18.49, 18.06, 17.95, 17.95, 17.71, 18.07, 17.8, 17.94, 17.81]
[83.45]
24.61485242843628
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 234425, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 90000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 21.312235116958618, 'TIME_S_1KI': 0.09091280843322434, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2054.109435153008, 'W': 83.45000000000002}
[18.37, 18.49, 18.06, 17.95, 17.95, 17.71, 18.07, 17.8, 17.94, 17.81, 18.28, 18.06, 18.16, 17.86, 18.06, 17.97, 18.34, 17.89, 18.12, 18.1]
324.71
16.2355
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 234425, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 90000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 21.312235116958618, 'TIME_S_1KI': 0.09091280843322434, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2054.109435153008, 'W': 83.45000000000002, 'J_1KI': 8.762330959381499, 'W_1KI': 0.3559773914898156, 'W_D': 67.21450000000002, 'J_D': 1654.4749985511307, 'W_D_1KI': 0.2867206995840888, 'J_D_1KI': 0.0012230807276702091}

0
pytorch/output_synthetic_16core/xeon_4216_16_csr_20_10_10_synthetic_30000_0.001.json Normal file
View File

77
pytorch/output_synthetic_16core/xeon_4216_16_csr_20_10_10_synthetic_30000_0.001.output Normal file
View File

@ -0,0 +1,77 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '30000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 900000, "MATRIX_DENSITY": 0.001, "TIME_S": 0.2140212059020996}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 34, 62, ..., 899934, 899967,
900000]),
col_indices=tensor([ 1559, 1711, 3295, ..., 29804, 29893, 29964]),
values=tensor([0.7225, 0.7366, 0.0675, ..., 0.3495, 0.2204, 0.5611]),
size=(30000, 30000), nnz=900000, layout=torch.sparse_csr)
tensor([0.1783, 0.4759, 0.5239, ..., 0.8363, 0.1566, 0.5506])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 900000
Density: 0.001
Time: 0.2140212059020996 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '98121', '-ss', '30000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 900000, "MATRIX_DENSITY": 0.001, "TIME_S": 19.3143093585968}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 29, 64, ..., 899940, 899966,
900000]),
col_indices=tensor([ 612, 701, 1017, ..., 29770, 29777, 29834]),
values=tensor([0.4034, 0.5977, 0.8788, ..., 0.6466, 0.3405, 0.9207]),
size=(30000, 30000), nnz=900000, layout=torch.sparse_csr)
tensor([0.7678, 0.0123, 0.5496, ..., 0.4589, 0.2646, 0.8857])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 900000
Density: 0.001
Time: 19.3143093585968 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '106684', '-ss', '30000', '-sd', '0.001', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 900000, "MATRIX_DENSITY": 0.001, "TIME_S": 20.90600872039795}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 21, 51, ..., 899936, 899963,
900000]),
col_indices=tensor([ 855, 2329, 2453, ..., 28070, 28293, 29379]),
values=tensor([0.2478, 0.6443, 0.3087, ..., 0.2033, 0.4619, 0.6203]),
size=(30000, 30000), nnz=900000, layout=torch.sparse_csr)
tensor([0.3161, 0.0015, 0.4480, ..., 0.6517, 0.7843, 0.6370])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 900000
Density: 0.001
Time: 20.90600872039795 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 21, 51, ..., 899936, 899963,
900000]),
col_indices=tensor([ 855, 2329, 2453, ..., 28070, 28293, 29379]),
values=tensor([0.2478, 0.6443, 0.3087, ..., 0.2033, 0.4619, 0.6203]),
size=(30000, 30000), nnz=900000, layout=torch.sparse_csr)
tensor([0.3161, 0.0015, 0.4480, ..., 0.6517, 0.7843, 0.6370])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 900000
Density: 0.001
Time: 20.90600872039795 seconds

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_20_10_10_synthetic_30000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 303288, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 21.079484939575195, "TIME_S_1KI": 0.06950319478375404, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1922.1091361045835, "W": 78.6, "J_1KI": 6.337570679039671, "W_1KI": 0.2591596106670887, "W_D": 62.18274999999999, "J_D": 1520.636537953019, "W_D_1KI": 0.20502871857772148, "J_D_1KI": 0.0006760198839971298}

81
pytorch/output_synthetic_16core/xeon_4216_16_csr_20_10_10_synthetic_30000_1e-05.output Normal file
View File

@ -0,0 +1,81 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '30000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 0.08539462089538574}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9000, 9000, 9000]),
col_indices=tensor([ 8168, 26166, 15021, ..., 3965, 14348, 3180]),
values=tensor([0.0414, 0.9204, 0.6909, ..., 0.5705, 0.2524, 0.4947]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.9721, 0.7014, 0.8881, ..., 0.4193, 0.5170, 0.9013])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 0.08539462089538574 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '245917', '-ss', '30000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 17.02755308151245}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9000, 9000, 9000]),
col_indices=tensor([ 9352, 11930, 17471, ..., 19597, 20552, 1111]),
values=tensor([0.4298, 0.4908, 0.5157, ..., 0.6454, 0.4570, 0.2738]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.3622, 0.2189, 0.3857, ..., 0.2935, 0.6447, 0.7890])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 17.02755308151245 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-xeon_4216.sif', 'numactl', '--cpunodebind=0', '--membind=0', 'python3', 'spmv.py', 'synthetic', 'csr', '303288', '-ss', '30000', '-sd', '1e-05', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [30000, 30000], "MATRIX_ROWS": 30000, "MATRIX_SIZE": 900000000, "MATRIX_NNZ": 9000, "MATRIX_DENSITY": 1e-05, "TIME_S": 21.079484939575195}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 8999, 9000, 9000]),
col_indices=tensor([20060, 11216, 16521, ..., 22127, 15786, 9820]),
values=tensor([0.3604, 0.7216, 0.9721, ..., 0.8443, 0.2707, 0.5761]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.4467, 0.9917, 0.2567, ..., 0.4911, 0.0150, 0.9779])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 21.079484939575195 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 8999, 9000, 9000]),
col_indices=tensor([20060, 11216, 16521, ..., 22127, 15786, 9820]),
values=tensor([0.3604, 0.7216, 0.9721, ..., 0.8443, 0.2707, 0.5761]),
size=(30000, 30000), nnz=9000, layout=torch.sparse_csr)
tensor([0.4467, 0.9917, 0.2567, ..., 0.4911, 0.0150, 0.9779])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([30000, 30000])
Rows: 30000
Size: 900000000
NNZ: 9000
Density: 1e-05
Time: 21.079484939575195 seconds
[18.69, 17.92, 17.95, 17.83, 18.17, 17.93, 17.96, 17.93, 18.28, 17.85]
[78.6]
24.454314708709717
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 303288, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 9000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 21.079484939575195, 'TIME_S_1KI': 0.06950319478375404, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1922.1091361045835, 'W': 78.6}
[18.69, 17.92, 17.95, 17.83, 18.17, 17.93, 17.96, 17.93, 18.28, 17.85, 18.57, 17.96, 18.0, 18.17, 18.14, 18.05, 18.17, 20.58, 18.66, 18.18]
328.345
16.417250000000003
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 303288, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [30000, 30000], 'MATRIX_ROWS': 30000, 'MATRIX_SIZE': 900000000, 'MATRIX_NNZ': 9000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 21.079484939575195, 'TIME_S_1KI': 0.06950319478375404, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1922.1091361045835, 'W': 78.6, 'J_1KI': 6.337570679039671, 'W_1KI': 0.2591596106670887, 'W_D': 62.18274999999999, 'J_D': 1520.636537953019, 'W_D_1KI': 0.20502871857772148, 'J_D_1KI': 0.0006760198839971298}

1
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_100000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 24.691206455230713, "TIME_S_1KI": 24.691206455230713, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 647.4072245025635, "W": 22.90104020202067, "J_1KI": 647.4072245025635, "W_1KI": 22.90104020202067, "W_D": 3.140040202020675, "J_D": 88.76822598814977, "W_D_1KI": 3.140040202020675, "J_D_1KI": 3.140040202020675}

45
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_100000_0.0001.output Normal file
View File

@ -0,0 +1,45 @@
['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_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.0001, "TIME_S": 24.691206455230713}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 21, ..., 999980,
999990, 1000000]),
col_indices=tensor([ 5106, 13656, 15471, ..., 68202, 79637, 95576]),
values=tensor([0.9862, 0.5796, 0.7870, ..., 0.3201, 0.7080, 0.2748]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.6105, 0.8083, 0.7150, ..., 0.7011, 0.0810, 0.6416])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 24.691206455230713 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 21, ..., 999980,
999990, 1000000]),
col_indices=tensor([ 5106, 13656, 15471, ..., 68202, 79637, 95576]),
values=tensor([0.9862, 0.5796, 0.7870, ..., 0.3201, 0.7080, 0.2748]),
size=(100000, 100000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.6105, 0.8083, 0.7150, ..., 0.7011, 0.0810, 0.6416])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 1000000
Density: 0.0001
Time: 24.691206455230713 seconds
[20.64, 20.52, 20.48, 20.68, 20.48, 20.48, 20.52, 20.48, 20.28, 20.36]
[20.64, 20.84, 21.24, 24.56, 26.2, 27.16, 28.16, 26.08, 25.68, 24.72, 24.72, 24.48, 24.6, 24.6, 24.72, 24.68, 24.6, 24.52, 24.52, 24.8, 24.72, 24.6, 24.48, 24.48, 24.52, 24.44, 24.64]
28.269773721694946
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 24.691206455230713, 'TIME_S_1KI': 24.691206455230713, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 647.4072245025635, 'W': 22.90104020202067}
[20.64, 20.52, 20.48, 20.68, 20.48, 20.48, 20.52, 20.48, 20.28, 20.36, 20.6, 20.56, 20.64, 22.68, 24.64, 25.4, 25.4, 25.36, 24.48, 22.68]
395.2199999999999
19.760999999999996
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 24.691206455230713, 'TIME_S_1KI': 24.691206455230713, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 647.4072245025635, 'W': 22.90104020202067, 'J_1KI': 647.4072245025635, 'W_1KI': 22.90104020202067, 'W_D': 3.140040202020675, 'J_D': 88.76822598814977, 'W_D_1KI': 3.140040202020675, 'J_D_1KI': 3.140040202020675}

1
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_100000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 3170, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.487157583236694, "TIME_S_1KI": 3.3082516035446985, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 325.83638416290285, "W": 22.27932516765204, "J_1KI": 102.78750289050564, "W_1KI": 7.028178286325565, "W_D": 3.710325167652041, "J_D": 54.263714344978354, "W_D_1KI": 1.170449579700959, "J_D_1KI": 0.36922699675109116}

65
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_100000_1e-05.output Normal file
View File

@ -0,0 +1,65 @@
['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_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 3.3119447231292725}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 99999, 99999,
100000]),
col_indices=tensor([34080, 20424, 38945, ..., 64155, 47978, 44736]),
values=tensor([0.5824, 0.7466, 0.8758, ..., 0.8278, 0.8938, 0.7712]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.9015, 0.6308, 0.7799, ..., 0.6045, 0.4908, 0.8218])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 3.3119447231292725 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 3170 -ss 100000 -sd 1e-05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [100000, 100000], "MATRIX_ROWS": 100000, "MATRIX_SIZE": 10000000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.487157583236694}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 99997, 99999,
100000]),
col_indices=tensor([62540, 50524, 43651, ..., 12394, 59846, 74659]),
values=tensor([0.6601, 0.8101, 0.4564, ..., 0.4320, 0.9061, 0.8749]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.2783, 0.8812, 0.6091, ..., 0.5557, 0.0745, 0.6879])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 10.487157583236694 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 99997, 99999,
100000]),
col_indices=tensor([62540, 50524, 43651, ..., 12394, 59846, 74659]),
values=tensor([0.6601, 0.8101, 0.4564, ..., 0.4320, 0.9061, 0.8749]),
size=(100000, 100000), nnz=100000, layout=torch.sparse_csr)
tensor([0.2783, 0.8812, 0.6091, ..., 0.5557, 0.0745, 0.6879])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([100000, 100000])
Rows: 100000
Size: 10000000000
NNZ: 100000
Density: 1e-05
Time: 10.487157583236694 seconds
[20.76, 20.76, 20.76, 20.8, 20.84, 20.72, 20.6, 20.36, 20.36, 20.32]
[20.32, 20.36, 20.48, 22.0, 23.24, 25.44, 26.04, 26.48, 26.08, 24.6, 24.44, 24.44, 24.4, 24.6]
14.625056266784668
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 3170, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.487157583236694, 'TIME_S_1KI': 3.3082516035446985, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 325.83638416290285, 'W': 22.27932516765204}
[20.76, 20.76, 20.76, 20.8, 20.84, 20.72, 20.6, 20.36, 20.36, 20.32, 20.04, 20.16, 20.24, 20.6, 20.72, 20.72, 20.88, 20.72, 21.08, 21.0]
371.38
18.569
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 3170, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [100000, 100000], 'MATRIX_ROWS': 100000, 'MATRIX_SIZE': 10000000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 1e-05, 'TIME_S': 10.487157583236694, 'TIME_S_1KI': 3.3082516035446985, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 325.83638416290285, 'W': 22.27932516765204, 'J_1KI': 102.78750289050564, 'W_1KI': 7.028178286325565, 'W_D': 3.710325167652041, 'J_D': 54.263714344978354, 'W_D_1KI': 1.170449579700959, 'J_D_1KI': 0.36922699675109116}

1
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_10000_0.0001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 32170, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.42804479598999, "TIME_S_1KI": 0.32415432999658034, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 291.8528582954407, "W": 21.456480473872652, "J_1KI": 9.07220572879828, "W_1KI": 0.666971727506144, "W_D": 3.1474804738726547, "J_D": 42.81229504752161, "W_D_1KI": 0.09783899514680307, "J_D_1KI": 0.0030413116303016183}

62
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_10000_0.0001.output Normal file
View File

@ -0,0 +1,62 @@
['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_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 0.3263826370239258}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 10000, 10000, 10000]),
col_indices=tensor([1982, 558, 3662, ..., 629, 5634, 6549]),
values=tensor([0.5250, 0.9307, 0.0448, ..., 0.0150, 0.4421, 0.4831]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.5546, 0.0630, 0.8785, ..., 0.4779, 0.8090, 0.6189])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 0.3263826370239258 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 32170 -ss 10000 -sd 0.0001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000, "MATRIX_DENSITY": 0.0001, "TIME_S": 10.42804479598999}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9996, 9999, 10000]),
col_indices=tensor([8155, 9480, 4094, ..., 6796, 6921, 3902]),
values=tensor([0.0915, 0.3699, 0.5728, ..., 0.9057, 0.8661, 0.7356]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.3327, 0.6532, 0.3155, ..., 0.1421, 0.0155, 0.6755])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 10.42804479598999 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, ..., 9996, 9999, 10000]),
col_indices=tensor([8155, 9480, 4094, ..., 6796, 6921, 3902]),
values=tensor([0.0915, 0.3699, 0.5728, ..., 0.9057, 0.8661, 0.7356]),
size=(10000, 10000), nnz=10000, layout=torch.sparse_csr)
tensor([0.3327, 0.6532, 0.3155, ..., 0.1421, 0.0155, 0.6755])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 10000
Density: 0.0001
Time: 10.42804479598999 seconds
[20.24, 20.16, 19.96, 20.2, 20.32, 20.28, 20.4, 20.32, 20.32, 20.36]
[20.36, 20.24, 20.48, 22.52, 23.04, 24.76, 25.6, 25.52, 24.28, 23.12, 23.12, 23.16, 23.44]
13.602084398269653
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 32170, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.42804479598999, 'TIME_S_1KI': 0.32415432999658034, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 291.8528582954407, 'W': 21.456480473872652}
[20.24, 20.16, 19.96, 20.2, 20.32, 20.28, 20.4, 20.32, 20.32, 20.36, 20.52, 20.48, 20.64, 20.48, 20.48, 20.32, 20.48, 20.36, 20.32, 20.2]
366.17999999999995
18.308999999999997
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 32170, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000, 'MATRIX_DENSITY': 0.0001, 'TIME_S': 10.42804479598999, 'TIME_S_1KI': 0.32415432999658034, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 291.8528582954407, 'W': 21.456480473872652, 'J_1KI': 9.07220572879828, 'W_1KI': 0.666971727506144, 'W_D': 3.1474804738726547, 'J_D': 42.81229504752161, 'W_D_1KI': 0.09783899514680307, 'J_D_1KI': 0.0030413116303016183}

1
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_10000_0.001.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 4747, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 10.586360931396484, "TIME_S_1KI": 2.2301160588574858, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 326.0044006347656, "W": 22.3162197944069, "J_1KI": 68.67587963656321, "W_1KI": 4.701120664505352, "W_D": 3.9862197944068996, "J_D": 58.23231742858882, "W_D_1KI": 0.8397345258914893, "J_D_1KI": 0.17689794099251935}

65
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_10000_0.001.output Normal file
View File

@ -0,0 +1,65 @@
['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 0.001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 2.2116076946258545}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 9, 19, ..., 99984, 99990,
100000]),
col_indices=tensor([ 365, 990, 1421, ..., 6204, 7506, 8345]),
values=tensor([0.4012, 0.2163, 0.0214, ..., 0.4427, 0.7190, 0.8381]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.6373, 0.6560, 0.2779, ..., 0.6662, 0.5919, 0.8676])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 2.2116076946258545 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 4747 -ss 10000 -sd 0.001 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 100000, "MATRIX_DENSITY": 0.001, "TIME_S": 10.586360931396484}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 23, ..., 99976, 99989,
100000]),
col_indices=tensor([ 145, 447, 695, ..., 7955, 8009, 9128]),
values=tensor([0.3182, 0.0478, 0.7097, ..., 0.3986, 0.2793, 0.7202]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.3264, 0.5290, 0.7390, ..., 0.4961, 0.6761, 0.4965])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 10.586360931396484 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 23, ..., 99976, 99989,
100000]),
col_indices=tensor([ 145, 447, 695, ..., 7955, 8009, 9128]),
values=tensor([0.3182, 0.0478, 0.7097, ..., 0.3986, 0.2793, 0.7202]),
size=(10000, 10000), nnz=100000, layout=torch.sparse_csr)
tensor([0.3264, 0.5290, 0.7390, ..., 0.4961, 0.6761, 0.4965])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 100000
Density: 0.001
Time: 10.586360931396484 seconds
[20.56, 20.48, 20.36, 20.48, 20.4, 20.2, 20.32, 20.48, 20.52, 20.6]
[20.44, 20.44, 20.44, 21.8, 24.32, 26.12, 27.12, 27.16, 25.36, 24.28, 24.24, 24.12, 23.96, 23.84]
14.608406066894531
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 4747, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 10.586360931396484, 'TIME_S_1KI': 2.2301160588574858, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 326.0044006347656, 'W': 22.3162197944069}
[20.56, 20.48, 20.36, 20.48, 20.4, 20.2, 20.32, 20.48, 20.52, 20.6, 20.28, 20.08, 20.4, 20.32, 20.2, 20.36, 20.36, 20.4, 20.28, 20.48]
366.6
18.330000000000002
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 4747, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 100000, 'MATRIX_DENSITY': 0.001, 'TIME_S': 10.586360931396484, 'TIME_S_1KI': 2.2301160588574858, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 326.0044006347656, 'W': 22.3162197944069, 'J_1KI': 68.67587963656321, 'W_1KI': 4.701120664505352, 'W_D': 3.9862197944068996, 'J_D': 58.23231742858882, 'W_D_1KI': 0.8397345258914893, 'J_D_1KI': 0.17689794099251935}

1
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_10000_0.01.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 21.214847326278687, "TIME_S_1KI": 21.214847326278687, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 593.507265138626, "W": 22.813207511083125, "J_1KI": 593.507265138626, "W_1KI": 22.813207511083125, "W_D": 4.622207511083129, "J_D": 120.25111933398253, "W_D_1KI": 4.622207511083129, "J_D_1KI": 4.622207511083129}

45
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_10000_0.01.output Normal file
View File

@ -0,0 +1,45 @@
['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 0.01 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000000, "MATRIX_DENSITY": 0.01, "TIME_S": 21.214847326278687}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 111, 190, ..., 999805,
999902, 1000000]),
col_indices=tensor([ 3, 255, 407, ..., 9480, 9499, 9966]),
values=tensor([0.6179, 0.1045, 0.6429, ..., 0.5216, 0.7550, 0.7148]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.6572, 0.8503, 0.2699, ..., 0.6176, 0.8577, 0.2518])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 21.214847326278687 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 111, 190, ..., 999805,
999902, 1000000]),
col_indices=tensor([ 3, 255, 407, ..., 9480, 9499, 9966]),
values=tensor([0.6179, 0.1045, 0.6429, ..., 0.5216, 0.7550, 0.7148]),
size=(10000, 10000), nnz=1000000, layout=torch.sparse_csr)
tensor([0.6572, 0.8503, 0.2699, ..., 0.6176, 0.8577, 0.2518])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 1000000
Density: 0.01
Time: 21.214847326278687 seconds
[20.56, 20.64, 20.64, 20.52, 20.36, 20.4, 19.92, 19.96, 20.0, 20.08]
[20.04, 20.08, 23.16, 25.4, 27.72, 28.8, 28.8, 29.48, 25.64, 25.4, 24.0, 23.96, 23.64, 23.72, 23.92, 24.04, 24.32, 24.36, 24.04, 24.0, 23.84, 24.08, 24.28, 24.28, 24.28]
26.015949964523315
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.01, 'TIME_S': 21.214847326278687, 'TIME_S_1KI': 21.214847326278687, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 593.507265138626, 'W': 22.813207511083125}
[20.56, 20.64, 20.64, 20.52, 20.36, 20.4, 19.92, 19.96, 20.0, 20.08, 19.76, 19.76, 19.96, 20.28, 20.4, 20.36, 20.4, 20.0, 19.92, 20.2]
363.81999999999994
18.190999999999995
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 1000000, 'MATRIX_DENSITY': 0.01, 'TIME_S': 21.214847326278687, 'TIME_S_1KI': 21.214847326278687, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 593.507265138626, 'W': 22.813207511083125, 'J_1KI': 593.507265138626, 'W_1KI': 22.813207511083125, 'W_D': 4.622207511083129, 'J_D': 120.25111933398253, 'W_D_1KI': 4.622207511083129, 'J_D_1KI': 4.622207511083129}

1
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_10000_0.05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 106.68757820129395, "TIME_S_1KI": 106.68757820129395, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2614.243714923859, "W": 23.06903562044379, "J_1KI": 2614.243714923859, "W_1KI": 23.06903562044379, "W_D": 4.456035620443789, "J_D": 504.9696617529395, "W_D_1KI": 4.456035620443789, "J_D_1KI": 4.456035620443789}

45
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_10000_0.05.output Normal file
View File

@ -0,0 +1,45 @@
['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 0.05 -c 1']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.05, "TIME_S": 106.68757820129395}
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 529, 1008, ..., 4999026,
4999478, 5000000]),
col_indices=tensor([ 75, 122, 128, ..., 9908, 9909, 9916]),
values=tensor([0.8571, 0.2596, 0.0411, ..., 0.7048, 0.9398, 0.3732]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.2354, 0.1436, 0.6485, ..., 0.5167, 0.9065, 0.2719])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 106.68757820129395 seconds
/nfshomes/vut/ampere_research/pytorch/spmv.py:75: 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, 529, 1008, ..., 4999026,
4999478, 5000000]),
col_indices=tensor([ 75, 122, 128, ..., 9908, 9909, 9916]),
values=tensor([0.8571, 0.2596, 0.0411, ..., 0.7048, 0.9398, 0.3732]),
size=(10000, 10000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.2354, 0.1436, 0.6485, ..., 0.5167, 0.9065, 0.2719])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([10000, 10000])
Rows: 10000
Size: 100000000
NNZ: 5000000
Density: 0.05
Time: 106.68757820129395 seconds
[20.52, 20.6, 20.8, 20.84, 20.8, 20.6, 20.68, 20.68, 20.4, 20.56]
[20.56, 20.64, 20.92, 22.0, 23.8, 25.56, 26.6, 26.56, 26.4, 25.36, 24.52, 24.6, 24.56, 24.6, 24.32, 24.4, 24.4, 24.32, 24.24, 24.44, 24.6, 24.36, 24.32, 24.36, 24.48, 24.6, 24.76, 24.72, 24.64, 24.6, 24.48, 24.56, 24.64, 24.64, 24.44, 24.48, 24.36, 24.16, 24.16, 24.24, 24.28, 24.16, 24.2, 24.36, 24.44, 24.44, 24.32, 24.0, 24.0, 24.0, 24.28, 24.44, 24.56, 24.48, 24.48, 24.32, 24.52, 24.52, 24.36, 24.4, 24.4, 24.32, 24.36, 24.32, 24.68, 24.72, 24.6, 24.6, 24.64, 24.6, 24.72, 24.64, 24.64, 24.68, 24.68, 24.52, 24.4, 24.32, 24.2, 24.16, 24.24, 24.2, 24.2, 24.4, 24.52, 24.56, 24.8, 24.8, 24.56, 24.44, 24.4, 23.84, 23.76, 23.88, 24.0, 24.0, 24.16, 24.2, 24.36, 24.2, 24.16, 24.2, 24.24, 24.16, 24.16, 24.4, 24.32, 24.56]
113.32262682914734
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.05, 'TIME_S': 106.68757820129395, 'TIME_S_1KI': 106.68757820129395, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2614.243714923859, 'W': 23.06903562044379}
[20.52, 20.6, 20.8, 20.84, 20.8, 20.6, 20.68, 20.68, 20.4, 20.56, 20.4, 20.52, 20.72, 20.84, 20.96, 20.84, 20.8, 20.52, 20.64, 20.56]
372.26
18.613
{'CPU': 'Altra', 'CORES': 1, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.05, 'TIME_S': 106.68757820129395, 'TIME_S_1KI': 106.68757820129395, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2614.243714923859, 'W': 23.06903562044379, 'J_1KI': 2614.243714923859, 'W_1KI': 23.06903562044379, 'W_D': 4.456035620443789, 'J_D': 504.9696617529395, 'W_D_1KI': 4.456035620443789, 'J_D_1KI': 4.456035620443789}

1
pytorch/output_synthetic_1core/altra_1_csr_10_10_10_synthetic_10000_1e-05.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 1, "ITERATIONS": 145400, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 1000, "MATRIX_DENSITY": 1e-05, "TIME_S": 10.376285076141357, "TIME_S_1KI": 0.07136372129395707, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 324.9616888427734, "W": 22.159348523127505, "J_1KI": 2.2349497169379187, "W_1KI": 0.15240267209853856, "W_D": 3.711348523127505, "J_D": 54.42606233215331, "W_D_1KI": 0.02552509300637899, "J_D_1KI": 0.00017555084598610036}

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