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

MOAR

Browse Source
This commit is contained in:
cephi 2024-12-17 14:27:55 -05:00
parent a512a4dadc
commit a339f2d2fe
50 changed files with 974 additions and 535 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) parameter_list = enumerate([(size, density)
for size in args.synthetic_size for size in args.synthetic_size
for density in args.synthetic_density for density in args.synthetic_density
if size ** 2 * density <= 20000000]) if size ** 2 * density <= 30000000])
#for i, matrix in enumerate(glob.glob(f'{args.matrix_dir.rstrip("/")}/*.mtx')): #for i, matrix in enumerate(glob.glob(f'{args.matrix_dir.rstrip("/")}/*.mtx')):
for i, parameter in parameter_list: for i, parameter in parameter_list:

2
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.1.json
View File

@ -1 +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": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 215.95656299591064, "TIME_S_1KI": 215.95656299591064, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 4458.621953725812, "W": 20.314604170190528, "J_1KI": 4458.621953725812, "W_1KI": 20.314604170190528, "W_D": 5.2026041701905275, "J_D": 1141.8605538864108, "W_D_1KI": 5.2026041701905275, "J_D_1KI": 5.2026041701905275} {"CPU": "Altra", "CORES": 16, "ITERATIONS": 100, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 22.49751091003418, "TIME_S_1KI": 224.9751091003418, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 714.621188735962, "W": 24.29855166830563, "J_1KI": 7146.211887359621, "W_1KI": 242.98551668305632, "W_D": 5.645551668305629, "J_D": 166.03585676002507, "W_D_1KI": 56.45551668305629, "J_D_1KI": 564.5551668305628}

44
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.1.output
View File

@ -1,14 +1,14 @@
['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 0.1 -c 16'] ['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 100 -ss 10000 -sd 0.1 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 215.95656299591064} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 22.49751091003418}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 987, 1996, ..., 9998019, tensor(crow_indices=tensor([ 0, 1002, 1924, ..., 9998003,
9999013, 10000000]), 9998995, 10000000]),
col_indices=tensor([ 25, 29, 35, ..., 9989, 9993, 9996]), col_indices=tensor([ 0, 2, 31, ..., 9969, 9996, 9997]),
values=tensor([0.8438, 0.2270, 0.6737, ..., 0.5218, 0.6879, 0.5182]), values=tensor([0.9177, 0.7034, 0.7745, ..., 0.5598, 0.0709, 0.5319]),
size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.6846, 0.7185, 0.0206, ..., 0.2576, 0.7966, 0.0945]) tensor([0.7893, 0.1843, 0.8169, ..., 0.5734, 0.3496, 0.5102])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -16,16 +16,16 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 10000000 NNZ: 10000000
Density: 0.1 Density: 0.1
Time: 215.95656299591064 seconds Time: 22.49751091003418 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 987, 1996, ..., 9998019, tensor(crow_indices=tensor([ 0, 1002, 1924, ..., 9998003,
9999013, 10000000]), 9998995, 10000000]),
col_indices=tensor([ 25, 29, 35, ..., 9989, 9993, 9996]), col_indices=tensor([ 0, 2, 31, ..., 9969, 9996, 9997]),
values=tensor([0.8438, 0.2270, 0.6737, ..., 0.5218, 0.6879, 0.5182]), values=tensor([0.9177, 0.7034, 0.7745, ..., 0.5598, 0.0709, 0.5319]),
size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.6846, 0.7185, 0.0206, ..., 0.2576, 0.7966, 0.0945]) tensor([0.7893, 0.1843, 0.8169, ..., 0.5734, 0.3496, 0.5102])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -33,13 +33,13 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 10000000 NNZ: 10000000
Density: 0.1 Density: 0.1
Time: 215.95656299591064 seconds Time: 22.49751091003418 seconds
[16.56, 16.44, 16.64, 16.6, 16.56, 16.92, 16.84, 16.92, 16.96, 16.72] [20.52, 20.56, 20.88, 20.92, 21.04, 21.0, 20.96, 20.72, 20.72, 20.72]
[16.52, 16.4, 17.44, 17.44, 19.24, 21.04, 24.24, 25.48, 25.32, 24.52, 22.6, 22.16, 20.6, 20.6, 21.08, 21.08, 20.92, 21.04, 20.88, 20.64, 20.6, 20.72, 20.52, 20.76, 20.76, 20.52, 20.44, 20.28, 20.28, 20.28, 20.44, 20.68, 20.72, 20.96, 20.68, 20.72, 20.48, 20.4, 20.2, 20.48, 20.48, 20.24, 20.28, 20.28, 20.2, 20.04, 20.16, 20.28, 20.4, 20.56, 20.6, 20.52, 20.56, 20.56, 20.6, 20.64, 20.72, 20.52, 20.4, 20.28, 20.44, 20.56, 20.52, 20.6, 20.56, 20.56, 20.4, 20.48, 20.28, 20.24, 20.36, 20.44, 20.48, 20.52, 20.36, 20.44, 20.36, 20.36, 20.32, 20.52, 20.52, 20.64, 20.56, 20.52, 20.56, 20.64, 20.36, 20.64, 20.64, 20.72, 20.72, 20.64, 20.8, 20.52, 20.36, 20.32, 20.44, 20.4, 20.56, 20.8, 21.08, 20.84, 20.84, 20.84, 20.76, 20.4, 20.36, 20.48, 20.6, 20.56, 20.76, 20.64, 20.68, 20.72, 20.72, 20.56, 20.56, 20.56, 20.8, 20.6, 20.56, 20.44, 20.44, 20.28, 20.48, 20.56, 20.72, 20.72, 20.56, 20.72, 20.76, 20.68, 20.72, 20.6, 20.64, 20.76, 20.88, 21.08, 20.96, 20.96, 20.72, 20.64, 20.52, 20.44, 20.32, 20.48, 20.6, 20.56, 20.6, 20.84, 20.68, 20.68, 20.64, 20.64, 20.6, 20.44, 20.28, 20.4, 20.16, 20.52, 20.76, 20.92, 20.96, 20.68, 20.68, 20.64, 20.24, 20.16, 20.36, 20.56, 20.6, 20.72, 20.48, 20.48, 20.4, 20.24, 20.24, 20.32, 20.44, 20.28, 20.64, 20.8, 20.88, 21.0, 21.16, 20.76, 20.68, 20.4, 20.4, 20.48, 20.48, 20.52, 20.6, 20.56, 20.32, 20.2, 20.04, 20.04, 20.16, 20.36, 20.4, 20.4, 20.4, 20.4, 20.28, 20.36, 20.4, 20.52, 20.8, 21.04, 21.2, 21.04, 20.72, 20.72, 20.6, 20.6, 20.52] [20.52, 20.44, 20.56, 24.12, 26.0, 27.96, 31.12, 31.04, 30.6, 29.2, 26.72, 25.56, 25.56, 24.04, 24.24, 24.24, 24.04, 24.2, 24.16, 24.08, 24.12, 24.28, 24.16, 24.28, 24.28, 24.28, 24.36, 24.48, 24.44]
219.47865271568298 29.410032272338867
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 215.95656299591064, 'TIME_S_1KI': 215.95656299591064, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 4458.621953725812, 'W': 20.314604170190528} {'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 22.49751091003418, 'TIME_S_1KI': 224.9751091003418, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 714.621188735962, 'W': 24.29855166830563}
[16.56, 16.44, 16.64, 16.6, 16.56, 16.92, 16.84, 16.92, 16.96, 16.72, 16.8, 17.08, 16.92, 17.2, 17.12, 16.88, 16.68, 16.52, 16.6, 16.64] [20.52, 20.56, 20.88, 20.92, 21.04, 21.0, 20.96, 20.72, 20.72, 20.72, 20.32, 20.68, 20.68, 20.64, 20.64, 20.6, 20.6, 20.68, 20.64, 20.64]
302.24 373.06000000000006
15.112 18.653000000000002
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 215.95656299591064, 'TIME_S_1KI': 215.95656299591064, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 4458.621953725812, 'W': 20.314604170190528, 'J_1KI': 4458.621953725812, 'W_1KI': 20.314604170190528, 'W_D': 5.2026041701905275, 'J_D': 1141.8605538864108, 'W_D_1KI': 5.2026041701905275, 'J_D_1KI': 5.2026041701905275} {'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 22.49751091003418, 'TIME_S_1KI': 224.9751091003418, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 714.621188735962, 'W': 24.29855166830563, 'J_1KI': 7146.211887359621, 'W_1KI': 242.98551668305632, 'W_D': 5.645551668305629, 'J_D': 166.03585676002507, 'W_D_1KI': 56.45551668305629, 'J_D_1KI': 564.5551668305628}

2
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.2.json
View File

@ -1 +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": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 424.4422960281372, "TIME_S_1KI": 424.4422960281372, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 10577.914468402869, "W": 24.381232025818772, "J_1KI": 10577.914468402869, "W_1KI": 24.381232025818772, "W_D": 5.810232025818774, "J_D": 2520.797035424717, "W_D_1KI": 5.810232025818774, "J_D_1KI": 5.810232025818774} {"CPU": "Altra", "CORES": 16, "ITERATIONS": 100, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 42.36744213104248, "TIME_S_1KI": 423.6744213104248, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1267.0327228164674, "W": 24.507310404611225, "J_1KI": 12670.327228164673, "W_1KI": 245.07310404611223, "W_D": 5.971310404611227, "J_D": 308.71791134262116, "W_D_1KI": 59.71310404611227, "J_D_1KI": 597.1310404611228}

44
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.2.output
View File

@ -1,14 +1,14 @@
['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 0.2 -c 16'] ['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 100 -ss 10000 -sd 0.2 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 424.4422960281372} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 42.36744213104248}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1997, 3940, ..., 19995978, tensor(crow_indices=tensor([ 0, 1958, 3991, ..., 19996077,
19998013, 20000000]), 19998030, 20000000]),
col_indices=tensor([ 1, 6, 9, ..., 9994, 9996, 9997]), col_indices=tensor([ 7, 14, 15, ..., 9992, 9994, 9996]),
values=tensor([0.5415, 0.5931, 0.4382, ..., 0.2191, 0.0907, 0.5464]), values=tensor([0.5268, 0.0225, 0.1392, ..., 0.0209, 0.3714, 0.4352]),
size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr)
tensor([0.6696, 0.2101, 0.2330, ..., 0.2876, 0.3503, 0.9145]) tensor([0.3357, 0.5966, 0.1485, ..., 0.1370, 0.3281, 0.6890])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -16,16 +16,16 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 20000000 NNZ: 20000000
Density: 0.2 Density: 0.2
Time: 424.4422960281372 seconds Time: 42.36744213104248 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1997, 3940, ..., 19995978, tensor(crow_indices=tensor([ 0, 1958, 3991, ..., 19996077,
19998013, 20000000]), 19998030, 20000000]),
col_indices=tensor([ 1, 6, 9, ..., 9994, 9996, 9997]), col_indices=tensor([ 7, 14, 15, ..., 9992, 9994, 9996]),
values=tensor([0.5415, 0.5931, 0.4382, ..., 0.2191, 0.0907, 0.5464]), values=tensor([0.5268, 0.0225, 0.1392, ..., 0.0209, 0.3714, 0.4352]),
size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr)
tensor([0.6696, 0.2101, 0.2330, ..., 0.2876, 0.3503, 0.9145]) tensor([0.3357, 0.5966, 0.1485, ..., 0.1370, 0.3281, 0.6890])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -33,13 +33,13 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 20000000 NNZ: 20000000
Density: 0.2 Density: 0.2
Time: 424.4422960281372 seconds Time: 42.36744213104248 seconds
[20.28, 20.2, 20.08, 20.08, 20.28, 20.44, 21.04, 20.84, 20.88, 20.76] [20.84, 20.72, 20.6, 20.6, 20.4, 20.36, 20.4, 20.32, 20.32, 20.4]
[20.6, 20.28, 23.28, 24.56, 27.04, 28.72, 32.56, 32.56, 30.16, 30.32, 30.0, 27.84, 26.92, 26.16, 25.36, 24.2, 24.52, 24.6, 24.56, 24.56, 24.44, 24.36, 24.2, 24.16, 24.32, 24.52, 24.8, 24.88, 24.84, 24.96, 24.88, 24.88, 24.88, 24.68, 24.56, 24.52, 24.56, 24.52, 24.56, 24.6, 24.32, 24.28, 24.24, 24.36, 24.36, 24.44, 24.44, 24.36, 24.36, 24.36, 24.36, 24.4, 24.4, 24.56, 24.64, 24.68, 24.68, 24.92, 24.88, 24.92, 24.88, 24.8, 24.56, 24.56, 24.6, 24.4, 24.32, 24.44, 24.4, 24.4, 24.4, 24.52, 24.6, 24.64, 24.64, 24.36, 24.44, 24.32, 24.36, 24.32, 24.36, 24.36, 24.28, 24.36, 24.24, 24.4, 24.52, 24.72, 24.76, 24.68, 24.68, 24.24, 24.44, 24.32, 24.32, 24.48, 24.52, 24.4, 24.32, 24.24, 24.2, 24.4, 24.56, 24.64, 24.84, 24.68, 24.68, 24.68, 24.6, 24.6, 24.64, 25.0, 24.76, 24.56, 24.52, 24.4, 24.44, 24.56, 24.56, 24.72, 25.0, 24.92, 24.88, 25.04, 24.68, 24.52, 24.52, 24.32, 24.16, 24.08, 24.12, 24.12, 24.32, 24.52, 24.64, 24.8, 24.84, 24.72, 25.04, 24.92, 24.76, 24.76, 24.64, 24.64, 24.56, 24.68, 24.64, 24.4, 24.4, 24.36, 24.36, 24.28, 24.56, 24.44, 24.32, 24.32, 24.56, 24.56, 24.4, 24.28, 24.24, 24.12, 24.12, 24.28, 24.52, 24.48, 24.6, 24.64, 24.64, 24.48, 24.52, 24.4, 24.24, 24.08, 24.12, 24.2, 24.44, 24.68, 24.72, 24.96, 24.96, 24.8, 24.64, 24.56, 24.52, 24.6, 24.68, 24.68, 24.68, 24.8, 24.6, 24.52, 24.52, 24.52, 24.52, 24.32, 24.52, 24.44, 24.48, 24.52, 24.64, 24.52, 24.64, 24.8, 24.88, 24.88, 24.76, 24.6, 24.64, 24.52, 24.64, 24.52, 24.64, 24.48, 24.56, 24.44, 24.56, 24.56, 24.48, 24.52, 24.52, 24.6, 24.96, 24.92, 25.12, 25.16, 25.0, 24.92, 24.88, 24.92, 24.92, 24.96, 25.12, 25.08, 24.88, 24.56, 24.52, 24.48, 24.52, 24.68, 24.68, 24.76, 24.76, 24.68, 24.6, 24.56, 24.68, 24.6, 24.44, 24.24, 24.24, 24.36, 24.52, 24.6, 24.6, 24.64, 24.56, 24.36, 24.32, 24.44, 24.6, 24.48, 24.6, 24.56, 24.36, 24.44, 24.44, 24.44, 24.52, 24.84, 24.8, 25.12, 24.88, 24.6, 24.6, 24.64, 24.36, 24.4, 24.24, 24.24, 24.2, 24.2, 24.28, 24.52, 24.48, 24.48, 24.44, 24.4, 24.52, 24.48, 24.56, 24.56, 24.4, 24.56, 24.36, 24.56, 24.48, 24.4, 24.4, 24.36, 24.32, 24.36, 24.48, 24.52, 24.52, 24.72, 24.96, 24.88, 24.88, 24.76, 24.72, 24.64, 24.64, 24.64, 24.72, 24.56, 24.56, 24.56, 24.52, 24.4, 24.4, 24.56, 24.4, 24.64, 24.72, 24.72, 24.68, 24.72, 24.6, 24.6, 24.48, 24.68, 24.72, 24.72, 24.68, 24.68, 24.56, 24.32, 24.28, 24.44, 24.32, 24.32, 24.68, 24.68, 24.64, 24.72, 24.56, 24.52, 24.64, 24.64, 24.56, 24.6, 24.64, 24.64, 24.6, 24.28, 24.36, 24.56, 24.44, 24.76, 24.88, 24.84, 24.76, 24.72, 24.88, 24.72, 24.72, 24.64, 24.52, 24.76, 24.76, 25.0, 24.88, 24.92, 24.6, 24.44, 24.56, 24.72, 24.72, 24.84, 24.96, 25.16, 24.96, 24.96, 24.96, 24.92, 24.88, 24.92, 24.96, 25.04, 24.76, 24.76, 24.64, 24.76, 24.6, 24.68, 24.76, 24.64, 24.68, 24.52, 24.56, 24.68, 24.64, 24.64, 24.56, 24.52, 24.4, 24.28, 24.36, 24.68, 24.76, 24.76, 24.72, 24.6, 24.52, 24.52, 24.4, 24.24, 24.4, 24.44, 24.48, 24.6, 24.44, 24.4, 24.08, 24.16, 24.28, 24.28, 24.28] [20.44, 20.6, 20.96, 23.28, 24.24, 25.88, 29.52, 30.04, 29.56, 29.8, 30.12, 27.4, 27.4, 26.84, 25.8, 24.52, 24.6, 24.56, 24.52, 24.48, 24.4, 24.48, 24.4, 24.52, 24.52, 24.68, 24.84, 24.72, 24.8, 24.72, 24.68, 24.52, 24.56, 24.52, 24.4, 24.32, 24.48, 24.48, 24.2, 24.28, 24.4, 24.12, 24.24, 24.16, 24.32, 24.44, 24.36, 24.32, 24.36, 24.36, 24.24]
433.8547968864441 51.70019483566284
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 424.4422960281372, 'TIME_S_1KI': 424.4422960281372, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 10577.914468402869, 'W': 24.381232025818772} {'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 42.36744213104248, 'TIME_S_1KI': 423.6744213104248, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1267.0327228164674, 'W': 24.507310404611225}
[20.28, 20.2, 20.08, 20.08, 20.28, 20.44, 21.04, 20.84, 20.88, 20.76, 20.28, 20.6, 20.88, 21.24, 21.28, 21.28, 20.88, 20.36, 20.24, 20.32] [20.84, 20.72, 20.6, 20.6, 20.4, 20.36, 20.4, 20.32, 20.32, 20.4, 20.56, 20.36, 20.44, 20.72, 20.76, 20.8, 20.96, 20.92, 20.76, 20.76]
371.41999999999996 370.71999999999997
18.570999999999998 18.535999999999998
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 424.4422960281372, 'TIME_S_1KI': 424.4422960281372, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 10577.914468402869, 'W': 24.381232025818772, 'J_1KI': 10577.914468402869, 'W_1KI': 24.381232025818772, 'W_D': 5.810232025818774, 'J_D': 2520.797035424717, 'W_D_1KI': 5.810232025818774, 'J_D_1KI': 5.810232025818774} {'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 42.36744213104248, 'TIME_S_1KI': 423.6744213104248, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1267.0327228164674, 'W': 24.507310404611225, 'J_1KI': 12670.327228164673, 'W_1KI': 245.07310404611223, 'W_D': 5.971310404611227, 'J_D': 308.71791134262116, 'W_D_1KI': 59.71310404611227, 'J_D_1KI': 597.1310404611228}

2
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.3.json
View File

@ -1 +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": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 644.013694524765, "TIME_S_1KI": 644.013694524765, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 15694.744610672022, "W": 24.244425924397902, "J_1KI": 15694.744610672022, "W_1KI": 24.244425924397902, "W_D": 5.900425924397901, "J_D": 3819.668828884149, "W_D_1KI": 5.900425924397901, "J_D_1KI": 5.900425924397901} {"CPU": "Altra", "CORES": 16, "ITERATIONS": 100, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 64.48762941360474, "TIME_S_1KI": 644.8762941360474, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1866.6232081508633, "W": 24.560001352289554, "J_1KI": 18666.23208150863, "W_1KI": 245.60001352289555, "W_D": 6.134001352289552, "J_D": 466.19986370420406, "W_D_1KI": 61.34001352289552, "J_D_1KI": 613.4001352289552}

44
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_10000_0.3.output
View File

@ -1,14 +1,14 @@
['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 0.3 -c 16'] ['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 100 -ss 10000 -sd 0.3 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 644.013694524765} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 64.48762941360474}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3023, 6063, ..., 29994048, tensor(crow_indices=tensor([ 0, 3018, 6012, ..., 29994075,
29997006, 30000000]), 29997046, 30000000]),
col_indices=tensor([ 0, 6, 7, ..., 9989, 9991, 9992]), col_indices=tensor([ 12, 13, 14, ..., 9975, 9986, 9989]),
values=tensor([0.2093, 0.3818, 0.7570, ..., 0.3871, 0.5537, 0.9186]), values=tensor([0.7172, 0.9331, 0.0150, ..., 0.2187, 0.8326, 0.8808]),
size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr)
tensor([0.8581, 0.1742, 0.9495, ..., 0.9731, 0.2987, 0.5882]) tensor([0.9528, 0.5520, 0.9475, ..., 0.9876, 0.0053, 0.6635])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -16,16 +16,16 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 30000000 NNZ: 30000000
Density: 0.3 Density: 0.3
Time: 644.013694524765 seconds Time: 64.48762941360474 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3023, 6063, ..., 29994048, tensor(crow_indices=tensor([ 0, 3018, 6012, ..., 29994075,
29997006, 30000000]), 29997046, 30000000]),
col_indices=tensor([ 0, 6, 7, ..., 9989, 9991, 9992]), col_indices=tensor([ 12, 13, 14, ..., 9975, 9986, 9989]),
values=tensor([0.2093, 0.3818, 0.7570, ..., 0.3871, 0.5537, 0.9186]), values=tensor([0.7172, 0.9331, 0.0150, ..., 0.2187, 0.8326, 0.8808]),
size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr)
tensor([0.8581, 0.1742, 0.9495, ..., 0.9731, 0.2987, 0.5882]) tensor([0.9528, 0.5520, 0.9475, ..., 0.9876, 0.0053, 0.6635])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -33,13 +33,13 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 30000000 NNZ: 30000000
Density: 0.3 Density: 0.3
Time: 644.013694524765 seconds Time: 64.48762941360474 seconds
[20.6, 20.56, 20.52, 20.4, 20.44, 20.64, 20.76, 20.6, 20.68, 20.6] [20.6, 20.44, 20.32, 20.24, 20.36, 20.64, 20.68, 20.8, 20.92, 20.92]
[20.76, 20.88, 20.88, 21.56, 23.0, 25.4, 26.16, 29.0, 30.28, 29.24, 29.48, 30.24, 28.4, 28.36, 28.36, 27.08, 27.08, 26.28, 24.52, 24.16, 23.96, 24.0, 24.16, 24.28, 24.4, 24.36, 24.16, 24.28, 24.28, 24.24, 24.28, 24.52, 24.56, 24.68, 24.56, 24.44, 24.48, 24.32, 24.12, 24.32, 24.28, 24.28, 24.16, 24.32, 24.2, 24.2, 24.28, 24.36, 24.52, 24.64, 24.8, 24.72, 24.84, 24.84, 24.64, 24.6, 24.68, 24.44, 24.4, 24.44, 24.4, 24.2, 24.28, 24.36, 24.4, 24.4, 24.68, 24.68, 24.6, 24.6, 24.4, 24.44, 24.44, 24.4, 24.4, 24.32, 24.48, 24.8, 24.8, 24.72, 24.72, 24.6, 24.4, 24.24, 24.32, 24.36, 24.4, 24.4, 24.56, 24.8, 24.8, 24.88, 24.96, 25.16, 24.64, 24.8, 24.72, 24.52, 24.56, 24.52, 24.32, 24.48, 24.48, 24.64, 24.8, 24.72, 24.92, 24.96, 24.84, 24.8, 24.88, 24.6, 24.36, 24.36, 24.24, 24.24, 24.32, 24.36, 24.44, 24.36, 24.28, 24.84, 25.24, 25.48, 25.52, 25.32, 24.92, 24.92, 24.76, 24.56, 24.48, 24.56, 24.36, 24.6, 24.52, 24.6, 24.68, 24.64, 24.68, 24.84, 24.84, 24.84, 24.88, 24.92, 24.96, 24.92, 24.36, 24.4, 24.36, 24.28, 24.24, 24.48, 24.48, 24.32, 24.4, 24.2, 24.36, 24.48, 24.4, 24.6, 24.92, 24.6, 24.92, 24.76, 24.76, 24.84, 24.68, 24.52, 24.48, 24.36, 24.44, 24.08, 23.84, 23.96, 23.96, 24.16, 24.24, 24.24, 24.2, 24.04, 24.24, 24.28, 24.32, 24.6, 24.68, 24.44, 24.48, 24.32, 24.0, 24.0, 24.16, 24.04, 24.24, 24.32, 24.28, 24.16, 24.24, 24.32, 24.6, 24.76, 24.84, 24.8, 24.8, 24.8, 24.64, 24.6, 24.68, 24.44, 24.32, 24.28, 24.36, 24.44, 24.56, 24.64, 24.64, 24.48, 24.32, 24.36, 24.2, 24.08, 24.12, 24.36, 24.48, 24.44, 24.52, 24.28, 24.28, 24.16, 24.32, 24.56, 24.64, 24.92, 24.76, 24.48, 24.48, 24.32, 24.08, 24.2, 24.48, 24.48, 24.52, 24.56, 24.92, 24.8, 24.72, 24.84, 24.6, 24.52, 24.8, 24.68, 24.6, 24.6, 24.44, 24.4, 24.44, 24.52, 24.8, 24.92, 24.92, 24.92, 24.92, 24.68, 24.72, 24.6, 24.6, 24.68, 24.64, 24.44, 24.4, 24.28, 24.32, 24.56, 24.8, 24.96, 24.92, 24.64, 24.64, 24.36, 24.32, 24.24, 24.4, 24.52, 24.52, 24.56, 24.52, 24.64, 24.6, 24.84, 24.84, 24.68, 24.96, 24.64, 24.56, 24.4, 24.2, 24.52, 24.64, 24.64, 24.48, 24.36, 24.36, 24.36, 24.36, 24.52, 24.4, 24.36, 24.44, 24.68, 24.64, 24.76, 24.56, 24.48, 24.32, 24.32, 24.36, 24.52, 24.6, 24.72, 24.56, 24.52, 24.48, 24.64, 24.72, 24.72, 24.76, 24.76, 24.76, 24.56, 24.68, 24.76, 24.52, 24.6, 24.64, 24.36, 24.24, 24.2, 24.16, 24.0, 24.0, 24.44, 24.44, 24.64, 24.56, 24.48, 24.44, 24.36, 24.12, 23.96, 23.96, 23.88, 23.88, 24.12, 24.4, 24.64, 24.64, 24.44, 24.56, 24.72, 24.6, 24.72, 24.64, 24.48, 24.4, 24.4, 24.44, 24.4, 24.48, 24.36, 24.28, 24.44, 24.64, 24.64, 24.76, 24.64, 24.44, 24.44, 24.52, 24.4, 24.32, 24.24, 24.36, 24.28, 24.44, 24.52, 24.36, 24.44, 24.56, 24.32, 24.32, 24.56, 24.64, 24.32, 24.56, 24.36, 24.32, 24.4, 24.56, 24.72, 24.96, 24.92, 24.92, 24.68, 24.36, 24.08, 24.16, 24.24, 24.48, 24.44, 24.44, 24.56, 24.36, 24.52, 24.52, 24.4, 24.4, 24.44, 24.56, 24.8, 24.76, 24.6, 24.56, 24.56, 24.56, 24.64, 24.64, 24.64, 24.64, 24.6, 24.76, 24.68, 24.96, 24.88, 24.76, 24.8, 24.76, 24.6, 24.64, 24.64, 24.6, 24.52, 24.48, 24.56, 24.48, 24.4, 24.24, 24.2, 24.36, 24.2, 24.36, 24.44, 24.44, 24.44, 24.32, 24.16, 24.28, 24.24, 24.0, 24.28, 24.44, 24.6, 24.64, 24.6, 24.6, 24.64, 24.36, 24.32, 24.48, 24.52, 24.56, 24.44, 24.56, 24.72, 24.68, 24.68, 24.68, 24.72, 24.56, 24.44, 24.64, 24.64, 24.4, 24.48, 24.36, 24.44, 24.52, 24.56, 24.68, 24.68, 24.8, 24.72, 24.92, 24.8, 24.92, 24.88, 24.92, 24.72, 24.76, 24.6, 24.64, 24.64, 24.56, 24.68, 24.48, 24.44, 24.44, 24.44, 24.44, 24.44, 24.6, 24.8, 24.8, 24.88, 24.88, 24.8, 24.6, 24.56, 24.28, 24.24, 24.2, 24.16, 24.28, 24.36, 24.64, 24.56, 24.56, 24.64, 24.76, 24.88, 24.8, 24.96, 24.64, 24.76, 24.8, 24.72, 24.6, 24.52, 24.52, 24.28, 23.96, 23.92, 24.08, 24.16, 24.24, 24.52, 24.72, 24.64, 24.64, 24.52, 24.28, 24.28, 24.12, 24.16, 24.12, 24.24, 24.24, 24.4, 24.32, 24.64, 24.6, 24.56, 24.44, 24.44, 24.44, 24.32, 24.52, 24.8, 24.88, 24.92, 24.88, 24.6, 24.32, 24.28, 24.36, 24.36, 24.44, 24.64, 24.52, 24.48, 24.36, 24.28, 24.36, 24.48, 24.36, 24.28, 24.44, 24.4, 24.4, 24.0, 24.12, 24.12, 24.0, 24.2, 24.44, 24.44, 24.4, 24.36, 24.32, 24.24, 24.24, 24.36, 24.44, 24.52, 24.6, 24.56, 24.6, 24.28, 24.16, 24.16, 24.12, 23.96, 24.24, 24.24, 24.32, 24.24, 24.4, 24.48, 24.32, 24.24, 24.4, 24.16, 24.2, 24.36, 24.4, 24.4, 24.36, 24.32, 24.24, 24.16, 24.08, 24.32, 24.44, 24.68, 24.64, 24.56, 24.56, 24.56, 24.36, 24.44, 24.56, 24.48] [20.68, 20.44, 20.2, 24.28, 25.52, 26.88, 28.28, 30.32, 31.12, 31.36, 32.28, 31.6, 29.96, 28.72, 28.72, 27.68, 26.76, 25.88, 24.68, 25.0, 25.04, 25.08, 24.72, 24.36, 24.4, 24.2, 24.2, 24.16, 24.36, 24.32, 24.36, 24.28, 24.28, 24.32, 24.44, 24.56, 24.52, 24.84, 24.84, 24.84, 25.0, 24.92, 24.92, 24.6, 24.52, 24.48, 24.4, 24.32, 24.4, 24.0, 24.24, 24.24, 24.32, 24.4, 24.6, 24.72, 24.92, 24.72, 24.64, 24.64, 24.8, 24.88, 24.8, 24.8, 24.92, 24.8, 24.52, 24.44, 24.24, 24.16, 24.16, 24.12, 24.16, 24.32, 24.6]
647.354763507843 76.00256943702698
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 644.013694524765, 'TIME_S_1KI': 644.013694524765, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 15694.744610672022, 'W': 24.244425924397902} {'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 64.48762941360474, 'TIME_S_1KI': 644.8762941360474, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1866.6232081508633, 'W': 24.560001352289554}
[20.6, 20.56, 20.52, 20.4, 20.44, 20.64, 20.76, 20.6, 20.68, 20.6, 20.12, 20.2, 20.36, 20.36, 20.2, 20.08, 20.24, 20.0, 20.12, 20.12] [20.6, 20.44, 20.32, 20.24, 20.36, 20.64, 20.68, 20.8, 20.92, 20.92, 20.28, 20.52, 20.52, 20.56, 20.2, 20.36, 20.4, 20.2, 20.28, 20.36]
366.88 368.52000000000004
18.344 18.426000000000002
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 644.013694524765, 'TIME_S_1KI': 644.013694524765, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 15694.744610672022, 'W': 24.244425924397902, 'J_1KI': 15694.744610672022, 'W_1KI': 24.244425924397902, 'W_D': 5.900425924397901, 'J_D': 3819.668828884149, 'W_D_1KI': 5.900425924397901, 'J_D_1KI': 5.900425924397901} {'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 64.48762941360474, 'TIME_S_1KI': 644.8762941360474, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1866.6232081508633, 'W': 24.560001352289554, 'J_1KI': 18666.23208150863, 'W_1KI': 245.60001352289555, 'W_D': 6.134001352289552, 'J_D': 466.19986370420406, 'W_D_1KI': 61.34001352289552, 'J_D_1KI': 613.4001352289552}

2
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_5000_0.1.json
View File

@ -1 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 53.914990186691284, "TIME_S_1KI": 53.914990186691284, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1144.4719800853732, "W": 20.188888585576834, "J_1KI": 1144.4719800853732, "W_1KI": 20.188888585576834, "W_D": 5.141888585576837, "J_D": 291.48446611952824, "W_D_1KI": 5.141888585576837, "J_D_1KI": 5.141888585576837} {"CPU": "Altra", "CORES": 16, "ITERATIONS": 199, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.763426303863525, "TIME_S_1KI": 54.08756936614837, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 349.11062700271606, "W": 22.846498625739535, "J_1KI": 1754.3247588076183, "W_1KI": 114.80652575748509, "W_D": 4.138498625739533, "J_D": 63.23918048667905, "W_D_1KI": 20.79647550622881, "J_D_1KI": 104.50490204135079}

64
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_5000_0.1.output
View File

@ -1,14 +1,14 @@
['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 5000 -sd 0.1 -c 16'] ['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 100 -ss 5000 -sd 0.1 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 53.914990186691284} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 5.264772415161133}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 478, 982, ..., 2498978, tensor(crow_indices=tensor([ 0, 501, 994, ..., 2498989,
2499487, 2500000]), 2499501, 2500000]),
col_indices=tensor([ 3, 18, 24, ..., 4984, 4986, 4997]), col_indices=tensor([ 0, 2, 18, ..., 4969, 4994, 4999]),
values=tensor([0.0150, 0.0039, 0.1247, ..., 0.8538, 0.3013, 0.1357]), values=tensor([0.5742, 0.5865, 0.7760, ..., 0.1661, 0.0806, 0.2598]),
size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.4451, 0.4302, 0.3190, ..., 0.9031, 0.3775, 0.0047]) tensor([0.0961, 0.5757, 0.9033, ..., 0.7471, 0.5676, 0.5058])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -16,16 +16,19 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 2500000 NNZ: 2500000
Density: 0.1 Density: 0.1
Time: 53.914990186691284 seconds Time: 5.264772415161133 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 199 -ss 5000 -sd 0.1 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.763426303863525}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 478, 982, ..., 2498978, tensor(crow_indices=tensor([ 0, 497, 1003, ..., 2498980,
2499487, 2500000]), 2499516, 2500000]),
col_indices=tensor([ 3, 18, 24, ..., 4984, 4986, 4997]), col_indices=tensor([ 3, 16, 28, ..., 4972, 4988, 4998]),
values=tensor([0.0150, 0.0039, 0.1247, ..., 0.8538, 0.3013, 0.1357]), values=tensor([0.4080, 0.5007, 0.4190, ..., 0.3549, 0.8542, 0.4591]),
size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.4451, 0.4302, 0.3190, ..., 0.9031, 0.3775, 0.0047]) tensor([0.8009, 0.9770, 0.1809, ..., 0.4100, 0.2770, 0.6242])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -33,13 +36,30 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 2500000 NNZ: 2500000
Density: 0.1 Density: 0.1
Time: 53.914990186691284 seconds Time: 10.763426303863525 seconds
[16.52, 16.4, 16.4, 16.56, 16.36, 16.52, 16.8, 16.56, 16.56, 16.76] /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.)
[16.48, 16.28, 19.4, 20.36, 22.04, 22.04, 24.88, 25.48, 22.92, 22.96, 20.08, 20.12, 20.12, 20.04, 19.92, 20.0, 20.24, 20.32, 20.32, 20.4, 20.4, 20.04, 20.08, 20.4, 20.48, 20.68, 20.96, 20.6, 20.48, 20.44, 20.44, 20.12, 20.08, 20.2, 19.96, 20.0, 20.32, 20.44, 20.44, 20.48, 20.64, 20.44, 20.44, 20.56, 20.76, 20.64, 20.8, 20.6, 20.64, 20.44, 20.4, 20.08, 20.08, 20.2, 20.36, 20.36] matrix = matrix.to_sparse_csr().type(torch.float32)
56.68821120262146 tensor(crow_indices=tensor([ 0, 497, 1003, ..., 2498980,
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 53.914990186691284, 'TIME_S_1KI': 53.914990186691284, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1144.4719800853732, 'W': 20.188888585576834} 2499516, 2500000]),
[16.52, 16.4, 16.4, 16.56, 16.36, 16.52, 16.8, 16.56, 16.56, 16.76, 16.72, 16.68, 16.8, 16.84, 16.84, 16.84, 17.0, 17.2, 17.0, 17.16] col_indices=tensor([ 3, 16, 28, ..., 4972, 4988, 4998]),
300.93999999999994 values=tensor([0.4080, 0.5007, 0.4190, ..., 0.3549, 0.8542, 0.4591]),
15.046999999999997 size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr)
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 53.914990186691284, 'TIME_S_1KI': 53.914990186691284, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1144.4719800853732, 'W': 20.188888585576834, 'J_1KI': 1144.4719800853732, 'W_1KI': 20.188888585576834, 'W_D': 5.141888585576837, 'J_D': 291.48446611952824, 'W_D_1KI': 5.141888585576837, 'J_D_1KI': 5.141888585576837} tensor([0.8009, 0.9770, 0.1809, ..., 0.4100, 0.2770, 0.6242])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 2500000
Density: 0.1
Time: 10.763426303863525 seconds
[20.76, 20.6, 20.8, 20.68, 20.68, 20.84, 20.92, 20.84, 20.72, 20.72]
[20.88, 20.92, 22.2, 23.32, 23.32, 25.32, 26.76, 27.36, 26.4, 26.08, 24.52, 24.48, 24.6, 24.56, 24.12]
15.28070592880249
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 199, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.763426303863525, 'TIME_S_1KI': 54.08756936614837, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 349.11062700271606, 'W': 22.846498625739535}
[20.76, 20.6, 20.8, 20.68, 20.68, 20.84, 20.92, 20.84, 20.72, 20.72, 20.68, 20.48, 20.92, 20.92, 20.92, 20.88, 20.92, 20.88, 20.68, 20.8]
374.16
18.708000000000002
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 199, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.763426303863525, 'TIME_S_1KI': 54.08756936614837, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 349.11062700271606, 'W': 22.846498625739535, 'J_1KI': 1754.3247588076183, 'W_1KI': 114.80652575748509, 'W_D': 4.138498625739533, 'J_D': 63.23918048667905, 'W_D_1KI': 20.79647550622881, 'J_D_1KI': 104.50490204135079}

2
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_5000_0.2.json
View File

@ -1 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 105.13774371147156, "TIME_S_1KI": 105.13774371147156, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2659.356458015442, "W": 24.098855953533207, "J_1KI": 2659.356458015442, "W_1KI": 24.098855953533207, "W_D": 5.604855953533207, "J_D": 618.5069492516519, "W_D_1KI": 5.604855953533207, "J_D_1KI": 5.604855953533207} {"CPU": "Altra", "CORES": 16, "ITERATIONS": 100, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 11.224013328552246, "TIME_S_1KI": 112.24013328552246, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 352.91180219650266, "W": 23.07981204778166, "J_1KI": 3529.118021965027, "W_1KI": 230.7981204778166, "W_D": 4.68881204778166, "J_D": 71.69629919505113, "W_D_1KI": 46.8881204778166, "J_D_1KI": 468.88120477816597}

44
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_5000_0.2.output
View File

@ -1,14 +1,14 @@
['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 5000 -sd 0.2 -c 16'] ['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 100 -ss 5000 -sd 0.2 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 105.13774371147156} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 11.224013328552246}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1002, 2008, ..., 4998043, tensor(crow_indices=tensor([ 0, 1006, 2027, ..., 4998032,
4999009, 5000000]), 4998984, 5000000]),
col_indices=tensor([ 2, 3, 6, ..., 4986, 4991, 4996]), col_indices=tensor([ 1, 6, 9, ..., 4994, 4996, 4999]),
values=tensor([0.2244, 0.9982, 0.0019, ..., 0.8643, 0.5557, 0.2909]), values=tensor([0.8034, 0.7589, 0.9109, ..., 0.7227, 0.1781, 0.2537]),
size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr) size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.1159, 0.5733, 0.3535, ..., 0.0131, 0.4436, 0.7693]) tensor([0.0665, 0.6506, 0.6868, ..., 0.4823, 0.5910, 0.0127])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -16,16 +16,16 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 5000000 NNZ: 5000000
Density: 0.2 Density: 0.2
Time: 105.13774371147156 seconds Time: 11.224013328552246 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1002, 2008, ..., 4998043, tensor(crow_indices=tensor([ 0, 1006, 2027, ..., 4998032,
4999009, 5000000]), 4998984, 5000000]),
col_indices=tensor([ 2, 3, 6, ..., 4986, 4991, 4996]), col_indices=tensor([ 1, 6, 9, ..., 4994, 4996, 4999]),
values=tensor([0.2244, 0.9982, 0.0019, ..., 0.8643, 0.5557, 0.2909]), values=tensor([0.8034, 0.7589, 0.9109, ..., 0.7227, 0.1781, 0.2537]),
size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr) size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.1159, 0.5733, 0.3535, ..., 0.0131, 0.4436, 0.7693]) tensor([0.0665, 0.6506, 0.6868, ..., 0.4823, 0.5910, 0.0127])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -33,13 +33,13 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 5000000 NNZ: 5000000
Density: 0.2 Density: 0.2
Time: 105.13774371147156 seconds Time: 11.224013328552246 seconds
[20.56, 20.6, 20.6, 20.56, 20.72, 20.56, 20.56, 20.4, 20.28, 20.32] [20.76, 20.72, 20.88, 20.84, 20.8, 20.64, 20.6, 20.44, 20.64, 20.6]
[20.52, 20.36, 21.08, 21.64, 23.52, 27.0, 27.0, 27.8, 28.08, 27.68, 25.2, 24.2, 24.68, 24.52, 24.52, 24.6, 24.52, 24.04, 24.04, 24.2, 24.04, 24.16, 24.4, 24.28, 24.28, 24.4, 24.68, 24.64, 24.72, 24.8, 24.8, 24.64, 24.52, 24.52, 24.36, 24.28, 24.2, 24.16, 24.32, 24.64, 24.76, 25.0, 25.0, 25.0, 24.96, 24.68, 24.64, 24.64, 24.6, 24.64, 24.6, 24.6, 24.52, 24.6, 24.68, 24.68, 24.68, 24.6, 24.56, 24.44, 24.36, 24.44, 24.48, 24.4, 24.64, 24.6, 24.32, 24.24, 24.24, 24.32, 24.32, 24.2, 24.24, 24.32, 24.28, 24.32, 24.44, 24.48, 24.44, 24.6, 24.6, 24.44, 24.32, 24.64, 24.52, 24.76, 24.48, 24.6, 24.44, 24.28, 24.28, 24.32, 24.32, 24.48, 24.56, 24.68, 24.72, 24.76, 24.88, 24.8, 24.72, 25.04, 25.0, 24.88, 24.72, 24.72, 24.64, 24.56, 24.48] [20.56, 20.4, 21.2, 21.2, 22.36, 25.64, 28.2, 28.88, 28.52, 27.12, 25.64, 24.52, 24.64, 24.76, 24.72]
110.35197949409485 15.29093050956726
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 105.13774371147156, 'TIME_S_1KI': 105.13774371147156, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2659.356458015442, 'W': 24.098855953533207} {'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 11.224013328552246, 'TIME_S_1KI': 112.24013328552246, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 352.91180219650266, 'W': 23.07981204778166}
[20.56, 20.6, 20.6, 20.56, 20.72, 20.56, 20.56, 20.4, 20.28, 20.32, 20.32, 20.48, 20.56, 20.56, 20.76, 20.76, 20.68, 20.48, 20.52, 20.4] [20.76, 20.72, 20.88, 20.84, 20.8, 20.64, 20.6, 20.44, 20.64, 20.6, 20.16, 20.04, 20.04, 20.24, 20.04, 20.16, 20.12, 20.2, 20.4, 20.52]
369.88 367.82000000000005
18.494 18.391000000000002
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 105.13774371147156, 'TIME_S_1KI': 105.13774371147156, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2659.356458015442, 'W': 24.098855953533207, 'J_1KI': 2659.356458015442, 'W_1KI': 24.098855953533207, 'W_D': 5.604855953533207, 'J_D': 618.5069492516519, 'W_D_1KI': 5.604855953533207, 'J_D_1KI': 5.604855953533207} {'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 11.224013328552246, 'TIME_S_1KI': 112.24013328552246, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 352.91180219650266, 'W': 23.07981204778166, 'J_1KI': 3529.118021965027, 'W_1KI': 230.7981204778166, 'W_D': 4.68881204778166, 'J_D': 71.69629919505113, 'W_D_1KI': 46.8881204778166, 'J_D_1KI': 468.88120477816597}

2
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_5000_0.3.json
View File

@ -1 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 161.33362221717834, "TIME_S_1KI": 161.33362221717834, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 4173.615943679809, "W": 24.123159694380117, "J_1KI": 4173.615943679809, "W_1KI": 24.123159694380117, "W_D": 5.260159694380114, "J_D": 910.0750749447333, "W_D_1KI": 5.260159694380114, "J_D_1KI": 5.260159694380114} {"CPU": "Altra", "CORES": 16, "ITERATIONS": 100, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 16.860999822616577, "TIME_S_1KI": 168.60999822616577, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 506.9433403301238, "W": 23.72394119074236, "J_1KI": 5069.433403301237, "W_1KI": 237.2394119074236, "W_D": 5.171941190742359, "J_D": 110.51625537872305, "W_D_1KI": 51.71941190742359, "J_D_1KI": 517.1941190742359}

44
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_5000_0.3.output
View File

@ -1,14 +1,14 @@
['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 5000 -sd 0.3 -c 16'] ['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 100 -ss 5000 -sd 0.3 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 161.33362221717834} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 16.860999822616577}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1457, 2918, ..., 7497005, tensor(crow_indices=tensor([ 0, 1507, 3002, ..., 7496981,
7498478, 7500000]), 7498462, 7500000]),
col_indices=tensor([ 5, 6, 7, ..., 4995, 4996, 4998]), col_indices=tensor([ 0, 3, 8, ..., 4995, 4997, 4998]),
values=tensor([0.1779, 0.8323, 0.7588, ..., 0.2380, 0.4426, 0.6569]), values=tensor([0.0922, 0.0923, 0.7842, ..., 0.1175, 0.1649, 0.6291]),
size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr)
tensor([0.9065, 0.9509, 0.6692, ..., 0.3115, 0.1768, 0.8689]) tensor([0.7654, 0.0824, 0.1261, ..., 0.9395, 0.8981, 0.0893])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -16,16 +16,16 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 7500000 NNZ: 7500000
Density: 0.3 Density: 0.3
Time: 161.33362221717834 seconds Time: 16.860999822616577 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1457, 2918, ..., 7497005, tensor(crow_indices=tensor([ 0, 1507, 3002, ..., 7496981,
7498478, 7500000]), 7498462, 7500000]),
col_indices=tensor([ 5, 6, 7, ..., 4995, 4996, 4998]), col_indices=tensor([ 0, 3, 8, ..., 4995, 4997, 4998]),
values=tensor([0.1779, 0.8323, 0.7588, ..., 0.2380, 0.4426, 0.6569]), values=tensor([0.0922, 0.0923, 0.7842, ..., 0.1175, 0.1649, 0.6291]),
size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr)
tensor([0.9065, 0.9509, 0.6692, ..., 0.3115, 0.1768, 0.8689]) tensor([0.7654, 0.0824, 0.1261, ..., 0.9395, 0.8981, 0.0893])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -33,13 +33,13 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 7500000 NNZ: 7500000
Density: 0.3 Density: 0.3
Time: 161.33362221717834 seconds Time: 16.860999822616577 seconds
[20.88, 20.96, 21.04, 20.92, 21.44, 21.32, 21.2, 21.24, 21.24, 20.44] [20.72, 20.68, 20.76, 20.76, 20.72, 20.56, 20.6, 20.4, 20.4, 20.28]
[20.36, 20.52, 23.44, 23.44, 24.92, 27.36, 30.32, 31.72, 29.36, 27.84, 25.72, 25.0, 24.32, 24.2, 24.2, 24.2, 24.2, 24.12, 24.0, 24.2, 24.36, 24.32, 24.24, 24.08, 24.08, 24.16, 24.28, 24.44, 24.44, 24.64, 24.44, 24.44, 24.8, 24.64, 24.56, 24.6, 24.68, 24.36, 24.52, 24.52, 24.56, 24.56, 24.88, 24.64, 24.56, 24.48, 24.04, 24.12, 24.28, 24.24, 24.28, 24.36, 24.36, 24.36, 24.48, 24.64, 24.6, 24.68, 24.48, 24.44, 24.44, 24.4, 24.28, 24.6, 24.44, 24.32, 24.32, 24.24, 24.2, 24.16, 24.0, 24.08, 24.36, 24.44, 24.64, 24.72, 24.8, 24.6, 24.6, 24.68, 24.6, 24.28, 24.2, 24.08, 24.24, 24.16, 24.28, 24.48, 24.56, 24.88, 24.88, 24.84, 25.04, 24.84, 24.76, 24.56, 24.2, 24.16, 24.24, 24.32, 24.24, 24.24, 24.2, 24.2, 24.2, 24.32, 24.32, 24.4, 24.4, 24.4, 24.64, 24.68, 24.52, 24.48, 24.48, 24.48, 24.56, 24.6, 24.88, 24.8, 24.64, 24.48, 24.6, 24.6, 24.84, 24.76, 24.8, 24.8, 24.72, 24.6, 24.64, 24.28, 24.12, 24.32, 24.2, 24.28, 24.4, 24.48, 24.28, 24.12, 24.12, 24.28, 24.24, 24.2, 24.2, 24.16, 24.04, 24.24, 24.36, 24.48, 24.48, 24.6, 24.6, 24.76, 24.56, 24.6, 24.56, 24.44, 24.4, 24.32, 24.36, 24.52, 24.48, 24.52, 24.52, 24.52, 24.72, 24.6, 24.6, 24.68, 24.64] [20.28, 20.12, 20.36, 22.84, 23.44, 27.24, 29.08, 29.68, 28.6, 28.0, 25.76, 24.4, 24.4, 24.44, 24.48, 24.56, 24.52, 24.52, 24.48, 24.52, 24.6]
173.01282238960266 21.368428468704224
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 161.33362221717834, 'TIME_S_1KI': 161.33362221717834, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 4173.615943679809, 'W': 24.123159694380117} {'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 16.860999822616577, 'TIME_S_1KI': 168.60999822616577, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 506.9433403301238, 'W': 23.72394119074236}
[20.88, 20.96, 21.04, 20.92, 21.44, 21.32, 21.2, 21.24, 21.24, 20.44, 20.6, 20.44, 20.8, 20.72, 20.84, 21.0, 20.96, 20.96, 20.88, 20.68] [20.72, 20.68, 20.76, 20.76, 20.72, 20.56, 20.6, 20.4, 20.4, 20.28, 20.48, 20.36, 20.4, 20.28, 20.48, 20.48, 20.84, 20.96, 21.12, 21.0]
377.26000000000005 371.04
18.863000000000003 18.552
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 161.33362221717834, 'TIME_S_1KI': 161.33362221717834, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 4173.615943679809, 'W': 24.123159694380117, 'J_1KI': 4173.615943679809, 'W_1KI': 24.123159694380117, 'W_D': 5.260159694380114, 'J_D': 910.0750749447333, 'W_D_1KI': 5.260159694380114, 'J_D_1KI': 5.260159694380114} {'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 16.860999822616577, 'TIME_S_1KI': 168.60999822616577, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 506.9433403301238, 'W': 23.72394119074236, 'J_1KI': 5069.433403301237, 'W_1KI': 237.2394119074236, 'W_D': 5.171941190742359, 'J_D': 110.51625537872305, 'W_D_1KI': 51.71941190742359, 'J_D_1KI': 517.1941190742359}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_5000_0.4.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 100, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 20.996277332305908, "TIME_S_1KI": 209.96277332305908, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 668.5575802040102, "W": 24.39593264447633, "J_1KI": 6685.575802040102, "W_1KI": 243.9593264447633, "W_D": 5.836932644476324, "J_D": 159.9580397877693, "W_D_1KI": 58.36932644476324, "J_D_1KI": 583.6932644476324}

45
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_5000_0.4.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 100 -ss 5000 -sd 0.4 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 20.996277332305908}
/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, 2028, 3979, ..., 9995991,
9997998, 10000000]),
col_indices=tensor([ 0, 3, 5, ..., 4997, 4998, 4999]),
values=tensor([0.5630, 0.7878, 0.5063, ..., 0.7425, 0.7666, 0.8643]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.1259, 0.6554, 0.1049, ..., 0.7276, 0.1818, 0.8981])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 20.996277332305908 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, 2028, 3979, ..., 9995991,
9997998, 10000000]),
col_indices=tensor([ 0, 3, 5, ..., 4997, 4998, 4999]),
values=tensor([0.5630, 0.7878, 0.5063, ..., 0.7425, 0.7666, 0.8643]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.1259, 0.6554, 0.1049, ..., 0.7276, 0.1818, 0.8981])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 20.996277332305908 seconds
[20.64, 20.64, 20.68, 20.84, 20.64, 20.56, 20.36, 20.48, 20.44, 20.6]
[20.92, 21.16, 21.52, 23.96, 23.96, 25.64, 28.24, 30.08, 29.68, 29.36, 27.16, 26.72, 25.12, 25.24, 25.48, 25.4, 25.28, 25.28, 25.08, 24.72, 24.36, 24.44, 24.32, 24.44, 24.68, 24.6, 24.52]
27.40446901321411
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.4, 'TIME_S': 20.996277332305908, 'TIME_S_1KI': 209.96277332305908, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 668.5575802040102, 'W': 24.39593264447633}
[20.64, 20.64, 20.68, 20.84, 20.64, 20.56, 20.36, 20.48, 20.44, 20.6, 20.44, 20.32, 20.6, 20.76, 20.76, 20.76, 20.96, 20.76, 20.56, 20.44]
371.18000000000006
18.559000000000005
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.4, 'TIME_S': 20.996277332305908, 'TIME_S_1KI': 209.96277332305908, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 668.5575802040102, 'W': 24.39593264447633, 'J_1KI': 6685.575802040102, 'W_1KI': 243.9593264447633, 'W_D': 5.836932644476324, 'J_D': 159.9580397877693, 'W_D_1KI': 58.36932644476324, 'J_D_1KI': 583.6932644476324}

1
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_5000_0.5.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Altra", "CORES": 16, "ITERATIONS": 100, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 12500000, "MATRIX_DENSITY": 0.5, "TIME_S": 26.35210919380188, "TIME_S_1KI": 263.5210919380188, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 811.2840313434602, "W": 24.192575243793513, "J_1KI": 8112.840313434603, "W_1KI": 241.92575243793513, "W_D": 5.690575243793514, "J_D": 190.8301525540354, "W_D_1KI": 56.90575243793514, "J_D_1KI": 569.0575243793514}

45
pytorch/output_synthetic_16core/altra_16_csr_10_10_10_synthetic_5000_0.5.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 100 -ss 5000 -sd 0.5 -c 16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 12500000, "MATRIX_DENSITY": 0.5, "TIME_S": 26.35210919380188}
/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, 2515, 4982, ..., 12494949,
12497477, 12500000]),
col_indices=tensor([ 0, 2, 3, ..., 4997, 4998, 4999]),
values=tensor([0.8539, 0.6304, 0.9472, ..., 0.9422, 0.2299, 0.9872]),
size=(5000, 5000), nnz=12500000, layout=torch.sparse_csr)
tensor([0.9794, 0.7866, 0.7649, ..., 0.0670, 0.2063, 0.9592])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 12500000
Density: 0.5
Time: 26.35210919380188 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, 2515, 4982, ..., 12494949,
12497477, 12500000]),
col_indices=tensor([ 0, 2, 3, ..., 4997, 4998, 4999]),
values=tensor([0.8539, 0.6304, 0.9472, ..., 0.9422, 0.2299, 0.9872]),
size=(5000, 5000), nnz=12500000, layout=torch.sparse_csr)
tensor([0.9794, 0.7866, 0.7649, ..., 0.0670, 0.2063, 0.9592])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 12500000
Density: 0.5
Time: 26.35210919380188 seconds
[20.32, 20.32, 20.56, 20.52, 20.8, 20.8, 20.84, 20.84, 20.56, 20.4]
[20.56, 20.68, 20.84, 22.32, 23.2, 25.76, 28.56, 30.4, 30.84, 30.24, 27.28, 27.28, 26.24, 24.76, 24.6, 24.68, 24.6, 24.4, 24.44, 24.36, 24.16, 24.28, 24.2, 24.2, 24.24, 24.24, 24.16, 24.04, 24.0, 24.0, 24.24, 24.4, 24.52]
33.534422159194946
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 12500000, 'MATRIX_DENSITY': 0.5, 'TIME_S': 26.35210919380188, 'TIME_S_1KI': 263.5210919380188, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 811.2840313434602, 'W': 24.192575243793513}
[20.32, 20.32, 20.56, 20.52, 20.8, 20.8, 20.84, 20.84, 20.56, 20.4, 20.56, 20.68, 20.6, 20.6, 20.6, 20.52, 20.24, 20.28, 20.48, 20.32]
370.03999999999996
18.502
{'CPU': 'Altra', 'CORES': 16, 'ITERATIONS': 100, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 12500000, 'MATRIX_DENSITY': 0.5, 'TIME_S': 26.35210919380188, 'TIME_S_1KI': 263.5210919380188, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 811.2840313434602, 'W': 24.192575243793513, 'J_1KI': 8112.840313434603, 'W_1KI': 241.92575243793513, 'W_D': 5.690575243793514, 'J_D': 190.8301525540354, 'W_D_1KI': 56.90575243793514, 'J_D_1KI': 569.0575243793514}

2
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.1.json
View File

@ -1 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 4716, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.072229146957397, "TIME_S_1KI": 2.1357568165728154, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1865.214413022995, "W": 124.6, "J_1KI": 395.5077211668777, "W_1KI": 26.42069550466497, "W_D": 88.853, "J_D": 1330.0954754440784, "W_D_1KI": 18.84075487701442, "J_D_1KI": 3.9950710086968657} {"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 4295, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 11.452549695968628, "TIME_S_1KI": 2.6664842132639412, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1764.6434702682495, "W": 122.04, "J_1KI": 410.85994651181596, "W_1KI": 28.41443538998836, "W_D": 80.21600000000001, "J_D": 1159.8872550888063, "W_D_1KI": 18.676600698486613, "J_D_1KI": 4.348451850637163}

58
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.1.output
View File

@ -1,14 +1,14 @@
['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', '0.1', '-c', '16'] ['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', '0.1', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 2.2264370918273926} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 2.4442625045776367}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1050, 2086, ..., 9997973, tensor(crow_indices=tensor([ 0, 1035, 2082, ..., 9998102,
9998977, 10000000]), 9999051, 10000000]),
col_indices=tensor([ 7, 18, 19, ..., 9986, 9989, 9991]), col_indices=tensor([ 19, 30, 32, ..., 9959, 9982, 9985]),
values=tensor([0.4594, 0.3854, 0.2627, ..., 0.1030, 0.9821, 0.5221]), values=tensor([0.5712, 0.3257, 0.7476, ..., 0.5702, 0.2998, 0.6598]),
size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.7604, 0.6286, 0.0952, ..., 0.1700, 0.1146, 0.5013]) tensor([0.5243, 0.8505, 0.9216, ..., 0.4616, 0.9699, 0.2962])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -16,19 +16,19 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 10000000 NNZ: 10000000
Density: 0.1 Density: 0.1
Time: 2.2264370918273926 seconds Time: 2.4442625045776367 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '4716', '-ss', '10000', '-sd', '0.1', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '4295', '-ss', '10000', '-sd', '0.1', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.072229146957397} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 11.452549695968628}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1024, 2032, ..., 9997988, tensor(crow_indices=tensor([ 0, 1025, 1966, ..., 9998059,
9999031, 10000000]), 9999093, 10000000]),
col_indices=tensor([ 8, 19, 25, ..., 9964, 9974, 9989]), col_indices=tensor([ 1, 7, 10, ..., 9966, 9969, 9981]),
values=tensor([0.6442, 0.9503, 0.4324, ..., 0.4734, 0.5264, 0.7582]), values=tensor([0.1039, 0.2120, 0.0962, ..., 0.1651, 0.8455, 0.3269]),
size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.7346, 0.1525, 0.9122, ..., 0.8135, 0.4141, 0.4880]) tensor([0.9433, 0.5217, 0.8143, ..., 0.2476, 0.9485, 0.9722])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -36,16 +36,16 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 10000000 NNZ: 10000000
Density: 0.1 Density: 0.1
Time: 10.072229146957397 seconds Time: 11.452549695968628 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1024, 2032, ..., 9997988, tensor(crow_indices=tensor([ 0, 1025, 1966, ..., 9998059,
9999031, 10000000]), 9999093, 10000000]),
col_indices=tensor([ 8, 19, 25, ..., 9964, 9974, 9989]), col_indices=tensor([ 1, 7, 10, ..., 9966, 9969, 9981]),
values=tensor([0.6442, 0.9503, 0.4324, ..., 0.4734, 0.5264, 0.7582]), values=tensor([0.1039, 0.2120, 0.0962, ..., 0.1651, 0.8455, 0.3269]),
size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.7346, 0.1525, 0.9122, ..., 0.8135, 0.4141, 0.4880]) tensor([0.9433, 0.5217, 0.8143, ..., 0.2476, 0.9485, 0.9722])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -53,13 +53,13 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 10000000 NNZ: 10000000
Density: 0.1 Density: 0.1
Time: 10.072229146957397 seconds Time: 11.452549695968628 seconds
[40.38, 39.18, 39.76, 39.52, 39.78, 39.24, 39.97, 39.14, 39.12, 39.09] [40.21, 39.72, 39.55, 39.95, 39.94, 40.02, 39.57, 39.94, 44.79, 39.53]
[124.6] [122.04]
14.969618082046509 14.459549903869629
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 4716, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.072229146957397, 'TIME_S_1KI': 2.1357568165728154, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1865.214413022995, 'W': 124.6} {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 4295, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 11.452549695968628, 'TIME_S_1KI': 2.6664842132639412, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1764.6434702682495, 'W': 122.04}
[40.38, 39.18, 39.76, 39.52, 39.78, 39.24, 39.97, 39.14, 39.12, 39.09, 39.77, 39.88, 39.63, 39.31, 39.15, 39.05, 39.3, 39.01, 43.67, 41.22] [40.21, 39.72, 39.55, 39.95, 39.94, 40.02, 39.57, 39.94, 44.79, 39.53, 40.97, 39.48, 39.49, 39.81, 60.91, 51.02, 67.71, 65.48, 67.26, 42.97]
714.94 836.48
35.747 41.824
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 4716, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.072229146957397, 'TIME_S_1KI': 2.1357568165728154, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1865.214413022995, 'W': 124.6, 'J_1KI': 395.5077211668777, 'W_1KI': 26.42069550466497, 'W_D': 88.853, 'J_D': 1330.0954754440784, 'W_D_1KI': 18.84075487701442, 'J_D_1KI': 3.9950710086968657} {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 4295, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 11.452549695968628, 'TIME_S_1KI': 2.6664842132639412, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1764.6434702682495, 'W': 122.04, 'J_1KI': 410.85994651181596, 'W_1KI': 28.41443538998836, 'W_D': 80.21600000000001, 'J_D': 1159.8872550888063, 'W_D_1KI': 18.676600698486613, 'J_D_1KI': 4.348451850637163}

2
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.2.json
View File

@ -1 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 2353, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 11.32719349861145, "TIME_S_1KI": 4.813936888487654, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2176.6533761024475, "W": 119.81, "J_1KI": 925.0545584795782, "W_1KI": 50.91797705057373, "W_D": 83.82675, "J_D": 1522.9261196494103, "W_D_1KI": 35.62547811304718, "J_D_1KI": 15.140449686802882} {"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 2191, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.797947645187378, "TIME_S_1KI": 4.928319326876941, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2022.1372631835936, "W": 118.88, "J_1KI": 922.9289197551773, "W_1KI": 54.25832952989502, "W_D": 82.67325, "J_D": 1406.2639594001769, "W_D_1KI": 37.733112733911454, "J_D_1KI": 17.22186797531331}

58
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.2.output
View File

@ -1,14 +1,14 @@
['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', '0.2', '-c', '16'] ['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', '0.2', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 4.462156772613525} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 4.791375398635864}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1973, 4017, ..., 19996005, tensor(crow_indices=tensor([ 0, 1984, 3988, ..., 19996067,
19998026, 20000000]), 19998037, 20000000]),
col_indices=tensor([ 9, 13, 23, ..., 9991, 9994, 9996]), col_indices=tensor([ 4, 5, 9, ..., 9988, 9990, 9993]),
values=tensor([0.6139, 0.9857, 0.8934, ..., 0.0556, 0.4216, 0.1096]), values=tensor([0.6195, 0.8354, 0.3980, ..., 0.7932, 0.7837, 0.6598]),
size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr)
tensor([0.2039, 0.8822, 0.1274, ..., 0.2057, 0.5761, 0.2633]) tensor([0.5668, 0.7998, 0.9844, ..., 0.6200, 0.3521, 0.6910])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -16,19 +16,19 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 20000000 NNZ: 20000000
Density: 0.2 Density: 0.2
Time: 4.462156772613525 seconds Time: 4.791375398635864 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '2353', '-ss', '10000', '-sd', '0.2', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '2191', '-ss', '10000', '-sd', '0.2', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 11.32719349861145} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.797947645187378}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2048, 4126, ..., 19995999, tensor(crow_indices=tensor([ 0, 2053, 4071, ..., 19995990,
19998042, 20000000]), 19998011, 20000000]),
col_indices=tensor([ 2, 9, 12, ..., 9990, 9992, 9998]), col_indices=tensor([ 2, 5, 7, ..., 9978, 9979, 9995]),
values=tensor([0.9897, 0.5727, 0.9448, ..., 0.9783, 0.3588, 0.5443]), values=tensor([0.3718, 0.8252, 0.7364, ..., 0.7436, 0.4903, 0.8833]),
size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr)
tensor([0.6351, 0.0385, 0.6216, ..., 0.2960, 0.7778, 0.2124]) tensor([0.1328, 0.8897, 0.5612, ..., 0.9517, 0.3970, 0.1461])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -36,16 +36,16 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 20000000 NNZ: 20000000
Density: 0.2 Density: 0.2
Time: 11.32719349861145 seconds Time: 10.797947645187378 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2048, 4126, ..., 19995999, tensor(crow_indices=tensor([ 0, 2053, 4071, ..., 19995990,
19998042, 20000000]), 19998011, 20000000]),
col_indices=tensor([ 2, 9, 12, ..., 9990, 9992, 9998]), col_indices=tensor([ 2, 5, 7, ..., 9978, 9979, 9995]),
values=tensor([0.9897, 0.5727, 0.9448, ..., 0.9783, 0.3588, 0.5443]), values=tensor([0.3718, 0.8252, 0.7364, ..., 0.7436, 0.4903, 0.8833]),
size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr)
tensor([0.6351, 0.0385, 0.6216, ..., 0.2960, 0.7778, 0.2124]) tensor([0.1328, 0.8897, 0.5612, ..., 0.9517, 0.3970, 0.1461])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -53,13 +53,13 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 20000000 NNZ: 20000000
Density: 0.2 Density: 0.2
Time: 11.32719349861145 seconds Time: 10.797947645187378 seconds
[40.52, 39.82, 39.98, 39.31, 45.78, 39.49, 39.61, 39.59, 39.44, 39.87] [40.47, 41.35, 39.66, 40.26, 40.03, 39.5, 40.17, 39.67, 39.73, 39.92]
[119.81] [118.88]
18.167543411254883 17.009902954101562
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 2353, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 11.32719349861145, 'TIME_S_1KI': 4.813936888487654, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2176.6533761024475, 'W': 119.81} {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 2191, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.797947645187378, 'TIME_S_1KI': 4.928319326876941, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2022.1372631835936, 'W': 118.88}
[40.52, 39.82, 39.98, 39.31, 45.78, 39.49, 39.61, 39.59, 39.44, 39.87, 41.36, 39.61, 39.47, 39.55, 39.57, 39.39, 39.51, 39.42, 39.39, 39.72] [40.47, 41.35, 39.66, 40.26, 40.03, 39.5, 40.17, 39.67, 39.73, 39.92, 40.22, 39.71, 39.72, 40.09, 39.98, 40.13, 39.63, 39.44, 45.03, 39.46]
719.665 724.135
35.98325 36.20675
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 2353, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 11.32719349861145, 'TIME_S_1KI': 4.813936888487654, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2176.6533761024475, 'W': 119.81, 'J_1KI': 925.0545584795782, 'W_1KI': 50.91797705057373, 'W_D': 83.82675, 'J_D': 1522.9261196494103, 'W_D_1KI': 35.62547811304718, 'J_D_1KI': 15.140449686802882} {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 2191, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.797947645187378, 'TIME_S_1KI': 4.928319326876941, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2022.1372631835936, 'W': 118.88, 'J_1KI': 922.9289197551773, 'W_1KI': 54.25832952989502, 'W_D': 82.67325, 'J_D': 1406.2639594001769, 'W_D_1KI': 37.733112733911454, 'J_D_1KI': 17.22186797531331}

2
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.3.json
View File

@ -1 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 1423, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.225417137145996, "TIME_S_1KI": 7.185816681058324, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2165.9509678459167, "W": 114.32, "J_1KI": 1522.1018748038769, "W_1KI": 80.33731553056921, "W_D": 78.57325, "J_D": 1488.679206475675, "W_D_1KI": 55.21661981728742, "J_D_1KI": 38.802965437306696} {"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 1433, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.876516819000244, "TIME_S_1KI": 7.5900326720169184, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2139.1934151315686, "W": 115.81999999999998, "J_1KI": 1492.8076867631323, "W_1KI": 80.82344731332866, "W_D": 79.70624999999998, "J_D": 1472.173071531951, "W_D_1KI": 55.62194696441031, "J_D_1KI": 38.81503626267293}

58
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_10000_0.3.output
View File

@ -1,14 +1,14 @@
['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', '0.3', '-c', '16'] ['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', '0.3', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 7.375466823577881} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 7.326756000518799}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2918, 5897, ..., 29994058, tensor(crow_indices=tensor([ 0, 3009, 6089, ..., 29994053,
29997053, 30000000]), 29996999, 30000000]),
col_indices=tensor([ 2, 5, 6, ..., 9983, 9996, 9997]), col_indices=tensor([ 1, 9, 11, ..., 9985, 9993, 9999]),
values=tensor([0.7175, 0.6857, 0.0471, ..., 0.3859, 0.1988, 0.0619]), values=tensor([0.2577, 0.0805, 0.4624, ..., 0.3276, 0.0599, 0.1828]),
size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr)
tensor([0.8128, 0.9693, 0.2211, ..., 0.0242, 0.9676, 0.8056]) tensor([0.9046, 0.5378, 0.0670, ..., 0.4938, 0.1002, 0.8451])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -16,19 +16,19 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 30000000 NNZ: 30000000
Density: 0.3 Density: 0.3
Time: 7.375466823577881 seconds Time: 7.326756000518799 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1423', '-ss', '10000', '-sd', '0.3', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1433', '-ss', '10000', '-sd', '0.3', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.225417137145996} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.876516819000244}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3014, 6025, ..., 29993963, tensor(crow_indices=tensor([ 0, 2984, 5966, ..., 29993973,
29996988, 30000000]), 29997036, 30000000]),
col_indices=tensor([ 3, 15, 23, ..., 9994, 9996, 9999]), col_indices=tensor([ 2, 6, 8, ..., 9989, 9994, 9995]),
values=tensor([0.7948, 0.4683, 0.9975, ..., 0.7409, 0.6873, 0.4627]), values=tensor([0.0728, 0.9990, 0.4694, ..., 0.7329, 0.2175, 0.4979]),
size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr)
tensor([0.6847, 0.1599, 0.7266, ..., 0.4640, 0.6124, 0.9614]) tensor([0.9122, 0.6942, 0.6684, ..., 0.3538, 0.9426, 0.8104])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -36,16 +36,16 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 30000000 NNZ: 30000000
Density: 0.3 Density: 0.3
Time: 10.225417137145996 seconds Time: 10.876516819000244 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3014, 6025, ..., 29993963, tensor(crow_indices=tensor([ 0, 2984, 5966, ..., 29993973,
29996988, 30000000]), 29997036, 30000000]),
col_indices=tensor([ 3, 15, 23, ..., 9994, 9996, 9999]), col_indices=tensor([ 2, 6, 8, ..., 9989, 9994, 9995]),
values=tensor([0.7948, 0.4683, 0.9975, ..., 0.7409, 0.6873, 0.4627]), values=tensor([0.0728, 0.9990, 0.4694, ..., 0.7329, 0.2175, 0.4979]),
size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr)
tensor([0.6847, 0.1599, 0.7266, ..., 0.4640, 0.6124, 0.9614]) tensor([0.9122, 0.6942, 0.6684, ..., 0.3538, 0.9426, 0.8104])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -53,13 +53,13 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 30000000 NNZ: 30000000
Density: 0.3 Density: 0.3
Time: 10.225417137145996 seconds Time: 10.876516819000244 seconds
[40.04, 41.53, 39.4, 39.29, 40.65, 39.48, 39.42, 39.79, 39.24, 39.4] [40.68, 39.67, 39.67, 39.75, 39.81, 39.61, 39.83, 39.6, 39.64, 40.06]
[114.32] [115.82]
18.94638705253601 18.469982862472534
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 1423, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.225417137145996, 'TIME_S_1KI': 7.185816681058324, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2165.9509678459167, 'W': 114.32} {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 1433, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.876516819000244, 'TIME_S_1KI': 7.5900326720169184, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2139.1934151315686, 'W': 115.81999999999998}
[40.04, 41.53, 39.4, 39.29, 40.65, 39.48, 39.42, 39.79, 39.24, 39.4, 40.83, 40.13, 39.25, 39.79, 39.29, 39.6, 39.88, 39.16, 39.35, 39.1] [40.68, 39.67, 39.67, 39.75, 39.81, 39.61, 39.83, 39.6, 39.64, 40.06, 40.61, 39.6, 39.69, 39.61, 40.38, 44.77, 40.11, 39.94, 40.18, 39.48]
714.935 722.2750000000001
35.74675 36.11375
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 1423, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.225417137145996, 'TIME_S_1KI': 7.185816681058324, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2165.9509678459167, 'W': 114.32, 'J_1KI': 1522.1018748038769, 'W_1KI': 80.33731553056921, 'W_D': 78.57325, 'J_D': 1488.679206475675, 'W_D_1KI': 55.21661981728742, 'J_D_1KI': 38.802965437306696} {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 1433, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.876516819000244, 'TIME_S_1KI': 7.5900326720169184, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2139.1934151315686, 'W': 115.81999999999998, 'J_1KI': 1492.8076867631323, 'W_1KI': 80.82344731332866, 'W_D': 79.70624999999998, 'J_D': 1472.173071531951, 'W_D_1KI': 55.62194696441031, 'J_D_1KI': 38.81503626267293}

2
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_5000_0.1.json
View File

@ -1 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 53642, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.719228744506836, "TIME_S_1KI": 0.19982902845730652, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1911.0154657673836, "W": 139.47, "J_1KI": 35.62535822242615, "W_1KI": 2.600014913687036, "W_D": 104.0105, "J_D": 1425.1500258277654, "W_D_1KI": 1.9389750568584316, "J_D_1KI": 0.036146583961418885} {"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 53196, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.621366739273071, "TIME_S_1KI": 0.19966476312642062, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1859.8867098927496, "W": 135.45, "J_1KI": 34.962905291614966, "W_1KI": 2.5462440785021427, "W_D": 100.0975, "J_D": 1374.4555920523405, "W_D_1KI": 1.8816734340927888, "J_D_1KI": 0.03537246097625364}

74
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_5000_0.1.output
View File

@ -1,14 +1,14 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '5000', '-sd', '0.1', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '5000', '-sd', '0.1', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 0.2534494400024414} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 0.3697686195373535}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 505, 999, ..., 2498983, tensor(crow_indices=tensor([ 0, 482, 953, ..., 2499004,
2499471, 2500000]), 2499522, 2500000]),
col_indices=tensor([ 5, 18, 40, ..., 4969, 4978, 4986]), col_indices=tensor([ 8, 9, 10, ..., 4989, 4992, 4993]),
values=tensor([0.5163, 0.4412, 0.3185, ..., 0.4202, 0.6886, 0.7408]), values=tensor([0.1808, 0.3744, 0.3617, ..., 0.4915, 0.9116, 0.1304]),
size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.4458, 0.7168, 0.1041, ..., 0.9922, 0.1900, 0.6310]) tensor([0.8787, 0.8438, 0.3305, ..., 0.0934, 0.4772, 0.3202])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -16,19 +16,19 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 2500000 NNZ: 2500000
Density: 0.1 Density: 0.1
Time: 0.2534494400024414 seconds Time: 0.3697686195373535 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '41428', '-ss', '5000', '-sd', '0.1', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '28396', '-ss', '5000', '-sd', '0.1', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 8.109162092208862} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 5.604789972305298}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 495, 993, ..., 2498983, tensor(crow_indices=tensor([ 0, 510, 995, ..., 2498986,
2499503, 2500000]), 2499511, 2500000]),
col_indices=tensor([ 3, 19, 21, ..., 4946, 4955, 4989]), col_indices=tensor([ 1, 2, 8, ..., 4984, 4986, 4997]),
values=tensor([0.8083, 0.9559, 0.4619, ..., 0.5259, 0.1142, 0.6698]), values=tensor([0.0748, 0.9968, 0.9443, ..., 0.0526, 0.3043, 0.5465]),
size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.9615, 0.1263, 0.8854, ..., 0.2773, 0.4703, 0.0965]) tensor([0.9671, 0.4248, 0.6295, ..., 0.7757, 0.3159, 0.1103])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -36,19 +36,19 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 2500000 NNZ: 2500000
Density: 0.1 Density: 0.1
Time: 8.109162092208862 seconds Time: 5.604789972305298 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '53642', '-ss', '5000', '-sd', '0.1', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '53196', '-ss', '5000', '-sd', '0.1', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.719228744506836} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.621366739273071}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 514, 1060, ..., 2499048, tensor(crow_indices=tensor([ 0, 515, 1007, ..., 2498937,
2499512, 2500000]), 2499481, 2500000]),
col_indices=tensor([ 10, 15, 21, ..., 4947, 4988, 4996]), col_indices=tensor([ 6, 15, 19, ..., 4986, 4991, 4993]),
values=tensor([0.5424, 0.5712, 0.8006, ..., 0.9771, 0.7885, 0.2387]), values=tensor([0.5542, 0.8952, 0.8012, ..., 0.8325, 0.7201, 0.7015]),
size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.8250, 0.9268, 0.6213, ..., 0.2000, 0.5207, 0.9721]) tensor([0.4746, 0.7865, 0.3346, ..., 0.0574, 0.7499, 0.4825])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -56,16 +56,16 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 2500000 NNZ: 2500000
Density: 0.1 Density: 0.1
Time: 10.719228744506836 seconds Time: 10.621366739273071 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 514, 1060, ..., 2499048, tensor(crow_indices=tensor([ 0, 515, 1007, ..., 2498937,
2499512, 2500000]), 2499481, 2500000]),
col_indices=tensor([ 10, 15, 21, ..., 4947, 4988, 4996]), col_indices=tensor([ 6, 15, 19, ..., 4986, 4991, 4993]),
values=tensor([0.5424, 0.5712, 0.8006, ..., 0.9771, 0.7885, 0.2387]), values=tensor([0.5542, 0.8952, 0.8012, ..., 0.8325, 0.7201, 0.7015]),
size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.8250, 0.9268, 0.6213, ..., 0.2000, 0.5207, 0.9721]) tensor([0.4746, 0.7865, 0.3346, ..., 0.0574, 0.7499, 0.4825])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -73,13 +73,13 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 2500000 NNZ: 2500000
Density: 0.1 Density: 0.1
Time: 10.719228744506836 seconds Time: 10.621366739273071 seconds
[40.33, 39.84, 39.58, 39.84, 39.14, 39.05, 39.58, 39.29, 39.44, 39.24] [39.76, 40.98, 39.03, 38.9, 39.03, 39.28, 39.39, 38.84, 39.07, 38.79]
[139.47] [135.45]
13.701982259750366 13.731168031692505
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 53642, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.719228744506836, 'TIME_S_1KI': 0.19982902845730652, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1911.0154657673836, 'W': 139.47} {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 53196, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.621366739273071, 'TIME_S_1KI': 0.19966476312642062, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1859.8867098927496, 'W': 135.45}
[40.33, 39.84, 39.58, 39.84, 39.14, 39.05, 39.58, 39.29, 39.44, 39.24, 39.75, 39.09, 39.67, 39.12, 39.04, 39.58, 39.1, 39.15, 39.53, 38.98] [39.76, 40.98, 39.03, 38.9, 39.03, 39.28, 39.39, 38.84, 39.07, 38.79, 39.74, 39.0, 39.18, 39.17, 39.18, 38.97, 39.43, 39.52, 39.42, 39.03]
709.19 707.05
35.459500000000006 35.3525
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 53642, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.719228744506836, 'TIME_S_1KI': 0.19982902845730652, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1911.0154657673836, 'W': 139.47, 'J_1KI': 35.62535822242615, 'W_1KI': 2.600014913687036, 'W_D': 104.0105, 'J_D': 1425.1500258277654, 'W_D_1KI': 1.9389750568584316, 'J_D_1KI': 0.036146583961418885} {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 53196, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.621366739273071, 'TIME_S_1KI': 0.19966476312642062, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1859.8867098927496, 'W': 135.45, 'J_1KI': 34.962905291614966, 'W_1KI': 2.5462440785021427, 'W_D': 100.0975, 'J_D': 1374.4555920523405, 'W_D_1KI': 1.8816734340927888, 'J_D_1KI': 0.03537246097625364}

2
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_5000_0.2.json
View File

@ -1 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 28937, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 14.827638626098633, "TIME_S_1KI": 0.5124110524967561, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 3609.586873168945, "W": 118.08, "J_1KI": 124.73949867536183, "W_1KI": 4.080588865466358, "W_D": 72.93325, "J_D": 2229.496119728565, "W_D_1KI": 2.5204150395687184, "J_D_1KI": 0.08710008085042398} {"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 28983, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.485858678817749, "TIME_S_1KI": 0.36179341955000344, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1935.3792489004136, "W": 138.58, "J_1KI": 66.77636024222522, "W_1KI": 4.781423593140807, "W_D": 102.671, "J_D": 1433.8816774704458, "W_D_1KI": 3.542455922437291, "J_D_1KI": 0.12222530181269332}

94
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_5000_0.2.output
View File

@ -1,14 +1,14 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '5000', '-sd', '0.2', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '5000', '-sd', '0.2', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 0.43933725357055664} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 0.43319129943847656}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1026, 2006, ..., 4997956, tensor(crow_indices=tensor([ 0, 992, 2019, ..., 4998034,
4998952, 5000000]), 4999022, 5000000]),
col_indices=tensor([ 2, 17, 24, ..., 4985, 4989, 4995]), col_indices=tensor([ 6, 7, 25, ..., 4976, 4987, 4995]),
values=tensor([0.8274, 0.9158, 0.7152, ..., 0.4764, 0.4337, 0.1760]), values=tensor([0.8659, 0.6827, 0.8740, ..., 0.4360, 0.6938, 0.9338]),
size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr) size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.9697, 0.2900, 0.2967, ..., 0.8048, 0.6311, 0.8937]) tensor([0.7963, 0.9033, 0.3372, ..., 0.2486, 0.4072, 0.8365])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -16,19 +16,19 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 5000000 NNZ: 5000000
Density: 0.2 Density: 0.2
Time: 0.43933725357055664 seconds Time: 0.43319129943847656 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '23899', '-ss', '5000', '-sd', '0.2', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '24238', '-ss', '5000', '-sd', '0.2', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 9.217516660690308} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 8.780949115753174}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1007, 2063, ..., 4997980, tensor(crow_indices=tensor([ 0, 970, 1988, ..., 4997978,
4998987, 5000000]), 4998978, 5000000]),
col_indices=tensor([ 1, 11, 20, ..., 4979, 4988, 4999]), col_indices=tensor([ 16, 17, 19, ..., 4996, 4997, 4999]),
values=tensor([0.6304, 0.9983, 0.6257, ..., 0.2269, 0.0229, 0.1968]), values=tensor([0.1543, 0.7354, 0.5297, ..., 0.9670, 0.3753, 0.9860]),
size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr) size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.1233, 0.6806, 0.4968, ..., 0.5662, 0.5626, 0.1071]) tensor([0.8318, 0.0236, 0.3656, ..., 0.1748, 0.2631, 0.0655])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -36,19 +36,19 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 5000000 NNZ: 5000000
Density: 0.2 Density: 0.2
Time: 9.217516660690308 seconds Time: 8.780949115753174 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '27224', '-ss', '5000', '-sd', '0.2', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '28983', '-ss', '5000', '-sd', '0.2', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 9.878192901611328} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.485858678817749}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 990, 1993, ..., 4997978, tensor(crow_indices=tensor([ 0, 1046, 2086, ..., 4998016,
4998966, 5000000]), 4999015, 5000000]),
col_indices=tensor([ 3, 4, 10, ..., 4987, 4994, 4995]), col_indices=tensor([ 1, 2, 7, ..., 4963, 4988, 4989]),
values=tensor([0.4002, 0.6649, 0.4467, ..., 0.0548, 0.1613, 0.2598]), values=tensor([0.8714, 0.5660, 0.3384, ..., 0.1998, 0.0405, 0.9940]),
size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr) size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.8115, 0.3462, 0.9342, ..., 0.0919, 0.3442, 0.4570]) tensor([0.6062, 0.0119, 0.7376, ..., 0.3549, 0.0670, 0.7680])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -56,19 +56,16 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 5000000 NNZ: 5000000
Density: 0.2 Density: 0.2
Time: 9.878192901611328 seconds Time: 10.485858678817749 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '28937', '-ss', '5000', '-sd', '0.2', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 14.827638626098633}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 969, 2001, ..., 4997956, tensor(crow_indices=tensor([ 0, 1046, 2086, ..., 4998016,
4998975, 5000000]), 4999015, 5000000]),
col_indices=tensor([ 6, 11, 14, ..., 4991, 4992, 4999]), col_indices=tensor([ 1, 2, 7, ..., 4963, 4988, 4989]),
values=tensor([0.1794, 0.8821, 0.0709, ..., 0.4822, 0.4100, 0.3846]), values=tensor([0.8714, 0.5660, 0.3384, ..., 0.1998, 0.0405, 0.9940]),
size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr) size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.9695, 0.1954, 0.0199, ..., 0.8094, 0.5782, 0.4777]) tensor([0.6062, 0.0119, 0.7376, ..., 0.3549, 0.0670, 0.7680])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -76,30 +73,13 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 5000000 NNZ: 5000000
Density: 0.2 Density: 0.2
Time: 14.827638626098633 seconds Time: 10.485858678817749 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.) [40.02, 39.97, 39.89, 40.7, 39.75, 39.29, 39.24, 40.56, 39.28, 39.34]
matrix = matrix.to_sparse_csr().type(torch.float32) [138.58]
tensor(crow_indices=tensor([ 0, 969, 2001, ..., 4997956, 13.965790510177612
4998975, 5000000]), {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 28983, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.485858678817749, 'TIME_S_1KI': 0.36179341955000344, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1935.3792489004136, 'W': 138.58}
col_indices=tensor([ 6, 11, 14, ..., 4991, 4992, 4999]), [40.02, 39.97, 39.89, 40.7, 39.75, 39.29, 39.24, 40.56, 39.28, 39.34, 40.14, 39.41, 39.41, 40.87, 40.15, 39.38, 39.46, 39.42, 41.94, 39.42]
values=tensor([0.1794, 0.8821, 0.0709, ..., 0.4822, 0.4100, 0.3846]), 718.1800000000001
size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr) 35.909000000000006
tensor([0.9695, 0.1954, 0.0199, ..., 0.8094, 0.5782, 0.4777]) {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 28983, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.485858678817749, 'TIME_S_1KI': 0.36179341955000344, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1935.3792489004136, 'W': 138.58, 'J_1KI': 66.77636024222522, 'W_1KI': 4.781423593140807, 'W_D': 102.671, 'J_D': 1433.8816774704458, 'W_D_1KI': 3.542455922437291, 'J_D_1KI': 0.12222530181269332}
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 5000000
Density: 0.2
Time: 14.827638626098633 seconds
[40.27, 40.67, 45.31, 61.44, 66.02, 68.38, 70.5, 73.12, 64.38, 67.33]
[118.08]
30.568994522094727
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 28937, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 14.827638626098633, 'TIME_S_1KI': 0.5124110524967561, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 3609.586873168945, 'W': 118.08}
[40.27, 40.67, 45.31, 61.44, 66.02, 68.38, 70.5, 73.12, 64.38, 67.33, 41.67, 40.18, 39.68, 39.83, 39.75, 39.7, 39.65, 40.28, 39.58, 39.66]
902.935
45.14675
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 28937, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 14.827638626098633, 'TIME_S_1KI': 0.5124110524967561, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 3609.586873168945, 'W': 118.08, 'J_1KI': 124.73949867536183, 'W_1KI': 4.080588865466358, 'W_D': 72.93325, 'J_D': 2229.496119728565, 'W_D_1KI': 2.5204150395687184, 'J_D_1KI': 0.08710008085042398}

2
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_5000_0.3.json
View File

@ -1 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 19318, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.466439723968506, "TIME_S_1KI": 0.5417972732150588, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1987.1806412553788, "W": 136.19, "J_1KI": 102.86678958770985, "W_1KI": 7.04990164613314, "W_D": 100.84025, "J_D": 1471.3840418485402, "W_D_1KI": 5.2200150119059945, "J_D_1KI": 0.2702150849935808} {"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 19322, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.55522608757019, "TIME_S_1KI": 0.5462802032693401, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1980.0031997823714, "W": 138.03, "J_1KI": 102.47402959229746, "W_1KI": 7.143670427491978, "W_D": 102.176, "J_D": 1465.6872197418213, "W_D_1KI": 5.288065417658628, "J_D_1KI": 0.2736810587754181}

74
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_5000_0.3.output
View File

@ -1,14 +1,14 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '5000', '-sd', '0.3', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '1000', '-ss', '5000', '-sd', '0.3', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 0.6186139583587646} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 0.6236028671264648}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1488, 3016, ..., 7497085, tensor(crow_indices=tensor([ 0, 1508, 3089, ..., 7496976,
7498552, 7500000]), 7498474, 7500000]),
col_indices=tensor([ 6, 10, 18, ..., 4990, 4991, 4992]), col_indices=tensor([ 7, 9, 15, ..., 4993, 4994, 4999]),
values=tensor([0.0319, 0.7689, 0.5437, ..., 0.3456, 0.3746, 0.7534]), values=tensor([0.5303, 0.4048, 0.9163, ..., 0.8482, 0.2205, 0.3341]),
size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr)
tensor([0.3033, 0.5024, 0.5149, ..., 0.3250, 0.9742, 0.7218]) tensor([0.9586, 0.6247, 0.6872, ..., 0.0060, 0.2602, 0.2097])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -16,19 +16,19 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 7500000 NNZ: 7500000
Density: 0.3 Density: 0.3
Time: 0.6186139583587646 seconds Time: 0.6236028671264648 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '16973', '-ss', '5000', '-sd', '0.3', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '16837', '-ss', '5000', '-sd', '0.3', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 9.225086212158203} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 9.149490594863892}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1496, 2988, ..., 7496994, tensor(crow_indices=tensor([ 0, 1490, 2963, ..., 7497013,
7498506, 7500000]), 7498520, 7500000]),
col_indices=tensor([ 3, 6, 16, ..., 4989, 4992, 4998]), col_indices=tensor([ 3, 5, 7, ..., 4995, 4998, 4999]),
values=tensor([0.5767, 0.7515, 0.3496, ..., 0.2965, 0.3271, 0.6066]), values=tensor([0.0553, 0.4159, 0.6036, ..., 0.7243, 0.9131, 0.1926]),
size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr)
tensor([0.6092, 0.5985, 0.2131, ..., 0.5175, 0.3653, 0.9642]) tensor([0.5885, 0.8517, 0.1610, ..., 0.2789, 0.8066, 0.5429])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -36,19 +36,19 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 7500000 NNZ: 7500000
Density: 0.3 Density: 0.3
Time: 9.225086212158203 seconds Time: 9.149490594863892 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '19318', '-ss', '5000', '-sd', '0.3', '-c', '16'] ['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '19322', '-ss', '5000', '-sd', '0.3', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.466439723968506} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.55522608757019}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1490, 3014, ..., 7497019, tensor(crow_indices=tensor([ 0, 1484, 2984, ..., 7497085,
7498487, 7500000]), 7498530, 7500000]),
col_indices=tensor([ 4, 5, 6, ..., 4990, 4995, 4998]), col_indices=tensor([ 5, 8, 9, ..., 4987, 4993, 4996]),
values=tensor([0.5409, 0.4022, 0.8537, ..., 0.7192, 0.8151, 0.0953]), values=tensor([0.5644, 0.8179, 0.2626, ..., 0.6489, 0.4956, 0.4262]),
size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr)
tensor([0.1022, 0.6049, 0.3113, ..., 0.6147, 0.8376, 0.5666]) tensor([0.0893, 0.0032, 0.8731, ..., 0.8525, 0.4366, 0.3134])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -56,16 +56,16 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 7500000 NNZ: 7500000
Density: 0.3 Density: 0.3
Time: 10.466439723968506 seconds Time: 10.55522608757019 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1490, 3014, ..., 7497019, tensor(crow_indices=tensor([ 0, 1484, 2984, ..., 7497085,
7498487, 7500000]), 7498530, 7500000]),
col_indices=tensor([ 4, 5, 6, ..., 4990, 4995, 4998]), col_indices=tensor([ 5, 8, 9, ..., 4987, 4993, 4996]),
values=tensor([0.5409, 0.4022, 0.8537, ..., 0.7192, 0.8151, 0.0953]), values=tensor([0.5644, 0.8179, 0.2626, ..., 0.6489, 0.4956, 0.4262]),
size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr)
tensor([0.1022, 0.6049, 0.3113, ..., 0.6147, 0.8376, 0.5666]) tensor([0.0893, 0.0032, 0.8731, ..., 0.8525, 0.4366, 0.3134])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -73,13 +73,13 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 7500000 NNZ: 7500000
Density: 0.3 Density: 0.3
Time: 10.466439723968506 seconds Time: 10.55522608757019 seconds
[39.82, 39.03, 39.04, 38.94, 39.09, 39.36, 39.4, 38.84, 39.28, 40.17] [40.26, 39.93, 39.81, 39.63, 39.7, 39.44, 39.58, 40.08, 39.61, 39.95]
[136.19] [138.03]
14.591237545013428 14.344730854034424
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 19318, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.466439723968506, 'TIME_S_1KI': 0.5417972732150588, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1987.1806412553788, 'W': 136.19} {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 19322, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.55522608757019, 'TIME_S_1KI': 0.5462802032693401, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1980.0031997823714, 'W': 138.03}
[39.82, 39.03, 39.04, 38.94, 39.09, 39.36, 39.4, 38.84, 39.28, 40.17, 40.47, 39.28, 39.08, 39.29, 39.92, 39.06, 39.2, 39.0, 39.43, 39.05] [40.26, 39.93, 39.81, 39.63, 39.7, 39.44, 39.58, 40.08, 39.61, 39.95, 40.68, 39.57, 40.69, 39.43, 39.89, 39.46, 39.71, 40.58, 39.58, 39.89]
706.9950000000001 717.0799999999999
35.34975000000001 35.854
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 19318, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.466439723968506, 'TIME_S_1KI': 0.5417972732150588, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1987.1806412553788, 'W': 136.19, 'J_1KI': 102.86678958770985, 'W_1KI': 7.04990164613314, 'W_D': 100.84025, 'J_D': 1471.3840418485402, 'W_D_1KI': 5.2200150119059945, 'J_D_1KI': 0.2702150849935808} {'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 19322, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.55522608757019, 'TIME_S_1KI': 0.5462802032693401, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1980.0031997823714, 'W': 138.03, 'J_1KI': 102.47402959229746, 'W_1KI': 7.143670427491978, 'W_D': 102.176, 'J_D': 1465.6872197418213, 'W_D_1KI': 5.288065417658628, 'J_D_1KI': 0.2736810587754181}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_5000_0.4.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 4731, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 11.47292685508728, "TIME_S_1KI": 2.425053235063894, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1834.0436391162873, "W": 121.03, "J_1KI": 387.66511078340466, "W_1KI": 25.582329317269078, "W_D": 85.37275, "J_D": 1293.70692465806, "W_D_1KI": 18.04539209469457, "J_D_1KI": 3.814287062924238}

105
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_5000_0.4.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', '100', '-ss', '5000', '-sd', '0.4', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 0.2614312171936035}
/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, 2042, 4085, ..., 9995953,
9997989, 10000000]),
col_indices=tensor([ 0, 1, 5, ..., 4996, 4997, 4999]),
values=tensor([0.9986, 0.3546, 0.4267, ..., 0.5093, 0.4180, 0.9781]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.6145, 0.7478, 0.4380, ..., 0.5071, 0.9779, 0.3806])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 0.2614312171936035 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '4016', '-ss', '5000', '-sd', '0.4', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 9.667505025863647}
/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, 1990, 3989, ..., 9996010,
9997994, 10000000]),
col_indices=tensor([ 1, 4, 5, ..., 4995, 4997, 4999]),
values=tensor([0.6319, 0.2996, 0.5553, ..., 0.4232, 0.5114, 0.0407]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.4664, 0.6985, 0.4117, ..., 0.0590, 0.5393, 0.9628])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 9.667505025863647 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '4361', '-ss', '5000', '-sd', '0.4', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 9.678797960281372}
/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, 2007, 3964, ..., 9996001,
9997987, 10000000]),
col_indices=tensor([ 1, 7, 8, ..., 4994, 4995, 4997]),
values=tensor([0.8466, 0.2679, 0.7670, ..., 0.5879, 0.3748, 0.0373]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.1526, 0.7358, 0.4701, ..., 0.9960, 0.7430, 0.4616])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 9.678797960281372 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '4731', '-ss', '5000', '-sd', '0.4', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 11.47292685508728}
/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, 1970, 4010, ..., 9995993,
9998023, 10000000]),
col_indices=tensor([ 8, 9, 11, ..., 4991, 4992, 4994]),
values=tensor([0.6655, 0.5354, 0.4656, ..., 0.4879, 0.8481, 0.6598]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.0638, 0.6144, 0.0359, ..., 0.3011, 0.2543, 0.1471])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 11.47292685508728 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, 1970, 4010, ..., 9995993,
9998023, 10000000]),
col_indices=tensor([ 8, 9, 11, ..., 4991, 4992, 4994]),
values=tensor([0.6655, 0.5354, 0.4656, ..., 0.4879, 0.8481, 0.6598]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.0638, 0.6144, 0.0359, ..., 0.3011, 0.2543, 0.1471])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 11.47292685508728 seconds
[39.98, 39.12, 39.4, 39.16, 39.21, 39.21, 39.11, 39.42, 39.73, 39.08]
[121.03]
15.1536283493042
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 4731, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.4, 'TIME_S': 11.47292685508728, 'TIME_S_1KI': 2.425053235063894, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1834.0436391162873, 'W': 121.03}
[39.98, 39.12, 39.4, 39.16, 39.21, 39.21, 39.11, 39.42, 39.73, 39.08, 39.95, 39.19, 39.66, 39.3, 39.21, 39.17, 39.07, 44.88, 39.22, 39.16]
713.145
35.65725
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 4731, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.4, 'TIME_S': 11.47292685508728, 'TIME_S_1KI': 2.425053235063894, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1834.0436391162873, 'W': 121.03, 'J_1KI': 387.66511078340466, 'W_1KI': 25.582329317269078, 'W_D': 85.37275, 'J_D': 1293.70692465806, 'W_D_1KI': 18.04539209469457, 'J_D_1KI': 3.814287062924238}

1
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_5000_0.5.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Epyc 7313P", "CORES": 16, "ITERATIONS": 3442, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 12500000, "MATRIX_DENSITY": 0.5, "TIME_S": 10.042348861694336, "TIME_S_1KI": 2.9175911858496035, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1718.164745798111, "W": 121.03, "J_1KI": 499.1762771057847, "W_1KI": 35.16269610691458, "W_D": 85.32925, "J_D": 1211.350153973341, "W_D_1KI": 24.79060139453806, "J_D_1KI": 7.202382741004667}

65
pytorch/output_synthetic_16core/epyc_7313p_16_csr_10_10_10_synthetic_5000_0.5.output Normal file
View File

@ -0,0 +1,65 @@
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '100', '-ss', '5000', '-sd', '0.5', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 12500000, "MATRIX_DENSITY": 0.5, "TIME_S": 0.3049776554107666}
/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, 2533, 4997, ..., 12495039,
12497522, 12500000]),
col_indices=tensor([ 0, 3, 4, ..., 4997, 4998, 4999]),
values=tensor([0.1496, 0.2452, 0.1750, ..., 0.2661, 0.4996, 0.2189]),
size=(5000, 5000), nnz=12500000, layout=torch.sparse_csr)
tensor([0.2427, 0.7858, 0.7306, ..., 0.9706, 0.6342, 0.3926])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 12500000
Density: 0.5
Time: 0.3049776554107666 seconds
['apptainer', 'run', '--env', 'OMP_PROC_BIND=true', '--env', 'OMP_PLACES={0:16}', 'pytorch-epyc_7313p.sif', 'python3', 'spmv.py', 'synthetic', 'csr', '3442', '-ss', '5000', '-sd', '0.5', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 12500000, "MATRIX_DENSITY": 0.5, "TIME_S": 10.042348861694336}
/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, 2398, 4864, ..., 12494963,
12497460, 12500000]),
col_indices=tensor([ 0, 1, 3, ..., 4991, 4993, 4998]),
values=tensor([0.1337, 0.4343, 0.0128, ..., 0.8248, 0.7529, 0.8640]),
size=(5000, 5000), nnz=12500000, layout=torch.sparse_csr)
tensor([0.5366, 0.6388, 0.1814, ..., 0.4104, 0.8844, 0.8048])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 12500000
Density: 0.5
Time: 10.042348861694336 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, 2398, 4864, ..., 12494963,
12497460, 12500000]),
col_indices=tensor([ 0, 1, 3, ..., 4991, 4993, 4998]),
values=tensor([0.1337, 0.4343, 0.0128, ..., 0.8248, 0.7529, 0.8640]),
size=(5000, 5000), nnz=12500000, layout=torch.sparse_csr)
tensor([0.5366, 0.6388, 0.1814, ..., 0.4104, 0.8844, 0.8048])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 12500000
Density: 0.5
Time: 10.042348861694336 seconds
[40.84, 39.12, 39.49, 39.13, 39.27, 39.35, 39.11, 39.03, 39.19, 39.06]
[121.03]
14.196188926696777
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 3442, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 12500000, 'MATRIX_DENSITY': 0.5, 'TIME_S': 10.042348861694336, 'TIME_S_1KI': 2.9175911858496035, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1718.164745798111, 'W': 121.03}
[40.84, 39.12, 39.49, 39.13, 39.27, 39.35, 39.11, 39.03, 39.19, 39.06, 42.71, 40.11, 39.63, 39.49, 40.35, 39.8, 39.81, 39.58, 40.7, 39.1]
714.0150000000001
35.700750000000006
{'CPU': 'Epyc 7313P', 'CORES': 16, 'ITERATIONS': 3442, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 12500000, 'MATRIX_DENSITY': 0.5, 'TIME_S': 10.042348861694336, 'TIME_S_1KI': 2.9175911858496035, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1718.164745798111, 'W': 121.03, 'J_1KI': 499.1762771057847, 'W_1KI': 35.16269610691458, 'W_D': 85.32925, 'J_D': 1211.350153973341, 'W_D_1KI': 24.79060139453806, 'J_D_1KI': 7.202382741004667}

2
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.1.json
View File

@ -1 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 2843, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.2704017162323, "TIME_S_1KI": 3.6125225874893774, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1409.3875760412218, "W": 82.18, "J_1KI": 495.73956244854793, "W_1KI": 28.90608512135069, "W_D": 66.09875000000001, "J_D": 1133.5940258196, "W_D_1KI": 23.249648258881464, "J_D_1KI": 8.177857284165128} {"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 2733, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.100283861160278, "TIME_S_1KI": 3.695676495119019, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1379.3402201175688, "W": 80.38, "J_1KI": 504.69821445941045, "W_1KI": 29.410903768752288, "W_D": 64.47149999999999, "J_D": 1106.3465165627001, "W_D_1KI": 23.590010976948406, "J_D_1KI": 8.631544448206515}

58
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.1.output
View File

@ -1,14 +1,14 @@
['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', '0.1', '-c', '16'] ['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', '0.1', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 3.6929545402526855} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 3.8413217067718506}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 984, 2024, ..., 9998063, tensor(crow_indices=tensor([ 0, 1005, 2046, ..., 9997977,
9998995, 10000000]), 9998998, 10000000]),
col_indices=tensor([ 8, 13, 17, ..., 9976, 9985, 9991]), col_indices=tensor([ 2, 12, 23, ..., 9983, 9984, 9993]),
values=tensor([0.9364, 0.6574, 0.1385, ..., 0.6834, 0.0920, 0.4928]), values=tensor([0.7359, 0.8841, 0.1080, ..., 0.9122, 0.7253, 0.6265]),
size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.2645, 0.6514, 0.1258, ..., 0.8959, 0.1836, 0.1827]) tensor([0.8932, 0.3673, 0.9234, ..., 0.2176, 0.0275, 0.3760])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -16,19 +16,19 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 10000000 NNZ: 10000000
Density: 0.1 Density: 0.1
Time: 3.6929545402526855 seconds Time: 3.8413217067718506 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', '2843', '-ss', '10000', '-sd', '0.1', '-c', '16'] ['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', '2733', '-ss', '10000', '-sd', '0.1', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.2704017162323} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.100283861160278}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 979, 2005, ..., 9997991, tensor(crow_indices=tensor([ 0, 961, 1970, ..., 9997917,
9998986, 10000000]), 9998991, 10000000]),
col_indices=tensor([ 33, 43, 63, ..., 9975, 9988, 9994]), col_indices=tensor([ 3, 10, 20, ..., 9978, 9982, 9998]),
values=tensor([0.7459, 0.7397, 0.9950, ..., 0.6626, 0.6614, 0.8057]), values=tensor([0.4857, 0.9168, 0.4028, ..., 0.0388, 0.1577, 0.6588]),
size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.0776, 0.4434, 0.3294, ..., 0.9636, 0.8443, 0.5700]) tensor([0.1504, 0.7286, 0.9298, ..., 0.2641, 0.6031, 0.0488])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -36,16 +36,16 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 10000000 NNZ: 10000000
Density: 0.1 Density: 0.1
Time: 10.2704017162323 seconds Time: 10.100283861160278 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 979, 2005, ..., 9997991, tensor(crow_indices=tensor([ 0, 961, 1970, ..., 9997917,
9998986, 10000000]), 9998991, 10000000]),
col_indices=tensor([ 33, 43, 63, ..., 9975, 9988, 9994]), col_indices=tensor([ 3, 10, 20, ..., 9978, 9982, 9998]),
values=tensor([0.7459, 0.7397, 0.9950, ..., 0.6626, 0.6614, 0.8057]), values=tensor([0.4857, 0.9168, 0.4028, ..., 0.0388, 0.1577, 0.6588]),
size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.0776, 0.4434, 0.3294, ..., 0.9636, 0.8443, 0.5700]) tensor([0.1504, 0.7286, 0.9298, ..., 0.2641, 0.6031, 0.0488])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -53,13 +53,13 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 10000000 NNZ: 10000000
Density: 0.1 Density: 0.1
Time: 10.2704017162323 seconds Time: 10.100283861160278 seconds
[18.3, 17.96, 17.98, 17.52, 17.65, 17.88, 17.7, 17.55, 17.72, 17.81] [18.01, 17.45, 17.58, 17.83, 17.38, 17.63, 17.79, 17.67, 17.39, 17.59]
[82.18] [80.38]
17.150007009506226 17.16024160385132
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 2843, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.2704017162323, 'TIME_S_1KI': 3.6125225874893774, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1409.3875760412218, 'W': 82.18} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 2733, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.100283861160278, 'TIME_S_1KI': 3.695676495119019, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1379.3402201175688, 'W': 80.38}
[18.3, 17.96, 17.98, 17.52, 17.65, 17.88, 17.7, 17.55, 17.72, 17.81, 18.29, 17.77, 17.64, 17.69, 18.52, 18.77, 17.66, 17.68, 17.94, 17.59] [18.01, 17.45, 17.58, 17.83, 17.38, 17.63, 17.79, 17.67, 17.39, 17.59, 18.23, 17.44, 17.92, 17.53, 18.07, 17.84, 17.81, 17.59, 17.47, 17.73]
321.625 318.1700000000001
16.08125 15.908500000000004
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 2843, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.2704017162323, 'TIME_S_1KI': 3.6125225874893774, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1409.3875760412218, 'W': 82.18, 'J_1KI': 495.73956244854793, 'W_1KI': 28.90608512135069, 'W_D': 66.09875000000001, 'J_D': 1133.5940258196, 'W_D_1KI': 23.249648258881464, 'J_D_1KI': 8.177857284165128} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 2733, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.100283861160278, 'TIME_S_1KI': 3.695676495119019, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1379.3402201175688, 'W': 80.38, 'J_1KI': 504.69821445941045, 'W_1KI': 29.410903768752288, 'W_D': 64.47149999999999, 'J_D': 1106.3465165627001, 'W_D_1KI': 23.590010976948406, 'J_D_1KI': 8.631544448206515}

2
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.2.json
View File

@ -1 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 1414, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.28559684753418, "TIME_S_1KI": 7.27411375356024, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2220.7405538129806, "W": 61.419999999999995, "J_1KI": 1570.5378739837204, "W_1KI": 43.437057991513434, "W_D": 45.185249999999996, "J_D": 1633.7466152585148, "W_D_1KI": 31.955622347949074, "J_D_1KI": 22.599450033910237} {"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 1466, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.305012702941895, "TIME_S_1KI": 7.0293401793600925, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 2200.8211415290834, "W": 63.38, "J_1KI": 1501.2422520662235, "W_1KI": 43.23328785811733, "W_D": 47.25450000000001, "J_D": 1640.875712091923, "W_D_1KI": 32.233628922237386, "J_D_1KI": 21.98746856905688}

58
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.2.output
View File

@ -1,14 +1,14 @@
['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', '0.2', '-c', '16'] ['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', '0.2', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 7.425642490386963} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 7.160099029541016}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 2015, 4097, ..., 19995972, tensor(crow_indices=tensor([ 0, 2043, 4140, ..., 19995945,
19998068, 20000000]), 19998003, 20000000]),
col_indices=tensor([ 2, 3, 5, ..., 9990, 9995, 9998]), col_indices=tensor([ 9, 18, 21, ..., 9988, 9992, 9993]),
values=tensor([0.5695, 0.2550, 0.6356, ..., 0.8876, 0.2921, 0.6555]), values=tensor([0.2892, 0.5577, 0.6311, ..., 0.1921, 0.3093, 0.8733]),
size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr)
tensor([0.8689, 0.2102, 0.7438, ..., 0.4178, 0.9795, 0.1690]) tensor([0.1991, 0.2737, 0.2500, ..., 0.3139, 0.4046, 0.9981])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -16,19 +16,19 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 20000000 NNZ: 20000000
Density: 0.2 Density: 0.2
Time: 7.425642490386963 seconds Time: 7.160099029541016 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', '1414', '-ss', '10000', '-sd', '0.2', '-c', '16'] ['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', '1466', '-ss', '10000', '-sd', '0.2', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.28559684753418} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 20000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.305012702941895}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1956, 4016, ..., 19996011, tensor(crow_indices=tensor([ 0, 1974, 3955, ..., 19996112,
19998026, 20000000]), 19998028, 20000000]),
col_indices=tensor([ 0, 3, 4, ..., 9992, 9993, 9997]), col_indices=tensor([ 4, 6, 10, ..., 9992, 9994, 9998]),
values=tensor([0.4860, 0.9038, 0.0641, ..., 0.5614, 0.3748, 0.7950]), values=tensor([0.7033, 0.3961, 0.2301, ..., 0.8377, 0.1653, 0.2698]),
size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr)
tensor([0.5984, 0.1665, 0.8658, ..., 0.8665, 0.0874, 0.1467]) tensor([0.6095, 0.7552, 0.1154, ..., 0.2753, 0.8409, 0.9940])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -36,16 +36,16 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 20000000 NNZ: 20000000
Density: 0.2 Density: 0.2
Time: 10.28559684753418 seconds Time: 10.305012702941895 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1956, 4016, ..., 19996011, tensor(crow_indices=tensor([ 0, 1974, 3955, ..., 19996112,
19998026, 20000000]), 19998028, 20000000]),
col_indices=tensor([ 0, 3, 4, ..., 9992, 9993, 9997]), col_indices=tensor([ 4, 6, 10, ..., 9992, 9994, 9998]),
values=tensor([0.4860, 0.9038, 0.0641, ..., 0.5614, 0.3748, 0.7950]), values=tensor([0.7033, 0.3961, 0.2301, ..., 0.8377, 0.1653, 0.2698]),
size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=20000000, layout=torch.sparse_csr)
tensor([0.5984, 0.1665, 0.8658, ..., 0.8665, 0.0874, 0.1467]) tensor([0.6095, 0.7552, 0.1154, ..., 0.2753, 0.8409, 0.9940])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -53,13 +53,13 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 20000000 NNZ: 20000000
Density: 0.2 Density: 0.2
Time: 10.28559684753418 seconds Time: 10.305012702941895 seconds
[18.83, 17.69, 17.79, 17.74, 18.11, 18.01, 17.84, 17.99, 20.88, 19.1] [18.39, 17.53, 17.66, 17.9, 17.98, 17.61, 17.98, 18.92, 18.01, 17.69]
[61.42] [63.38]
36.15663552284241 34.72422122955322
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 1414, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.28559684753418, 'TIME_S_1KI': 7.27411375356024, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2220.7405538129806, 'W': 61.419999999999995} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 1466, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.305012702941895, 'TIME_S_1KI': 7.0293401793600925, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2200.8211415290834, 'W': 63.38}
[18.83, 17.69, 17.79, 17.74, 18.11, 18.01, 17.84, 17.99, 20.88, 19.1, 17.89, 17.55, 18.69, 17.55, 17.64, 17.53, 17.47, 17.68, 17.69, 17.87] [18.39, 17.53, 17.66, 17.9, 17.98, 17.61, 17.98, 18.92, 18.01, 17.69, 18.14, 18.53, 17.76, 17.48, 17.87, 17.7, 17.84, 17.65, 18.06, 17.84]
324.69500000000005 322.51
16.234750000000002 16.1255
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 1414, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.28559684753418, 'TIME_S_1KI': 7.27411375356024, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2220.7405538129806, 'W': 61.419999999999995, 'J_1KI': 1570.5378739837204, 'W_1KI': 43.437057991513434, 'W_D': 45.185249999999996, 'J_D': 1633.7466152585148, 'W_D_1KI': 31.955622347949074, 'J_D_1KI': 22.599450033910237} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 1466, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 20000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.305012702941895, 'TIME_S_1KI': 7.0293401793600925, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 2200.8211415290834, 'W': 63.38, 'J_1KI': 1501.2422520662235, 'W_1KI': 43.23328785811733, 'W_D': 47.25450000000001, 'J_D': 1640.875712091923, 'W_D_1KI': 32.233628922237386, 'J_D_1KI': 21.98746856905688}

2
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.3.json
View File

@ -1 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 12.322984457015991, "TIME_S_1KI": 12.322984457015991, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 4274.3693664526945, "W": 53.17000000000001, "J_1KI": 4274.3693664526945, "W_1KI": 53.17000000000001, "W_D": 36.76275000000001, "J_D": 2955.380335274757, "W_D_1KI": 36.76275000000001, "J_D_1KI": 36.76275000000001} {"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 1000, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 11.716556549072266, "TIME_S_1KI": 11.716556549072266, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 4241.666789231301, "W": 53.31, "J_1KI": 4241.666789231301, "W_1KI": 53.31, "W_D": 37.065, "J_D": 2949.116104724407, "W_D_1KI": 37.065, "J_D_1KI": 37.065}

42
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_10000_0.3.output
View File

@ -1,14 +1,14 @@
['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', '0.3', '-c', '16'] ['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', '0.3', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 12.322984457015991} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [10000, 10000], "MATRIX_ROWS": 10000, "MATRIX_SIZE": 100000000, "MATRIX_NNZ": 30000000, "MATRIX_DENSITY": 0.3, "TIME_S": 11.716556549072266}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3028, 6045, ..., 29993985, tensor(crow_indices=tensor([ 0, 3043, 6100, ..., 29993952,
29996995, 30000000]), 29997033, 30000000]),
col_indices=tensor([ 0, 1, 2, ..., 9992, 9993, 9997]), col_indices=tensor([ 1, 6, 7, ..., 9991, 9996, 9999]),
values=tensor([0.0070, 0.5345, 0.8585, ..., 0.9349, 0.6772, 0.9628]), values=tensor([0.2505, 0.5332, 0.4314, ..., 0.9186, 0.8523, 0.0373]),
size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr)
tensor([0.3885, 0.1371, 0.6444, ..., 0.5368, 0.0739, 0.9250]) tensor([0.5706, 0.2901, 0.6021, ..., 0.8123, 0.3136, 0.2413])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -16,16 +16,16 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 30000000 NNZ: 30000000
Density: 0.3 Density: 0.3
Time: 12.322984457015991 seconds Time: 11.716556549072266 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 3028, 6045, ..., 29993985, tensor(crow_indices=tensor([ 0, 3043, 6100, ..., 29993952,
29996995, 30000000]), 29997033, 30000000]),
col_indices=tensor([ 0, 1, 2, ..., 9992, 9993, 9997]), col_indices=tensor([ 1, 6, 7, ..., 9991, 9996, 9999]),
values=tensor([0.0070, 0.5345, 0.8585, ..., 0.9349, 0.6772, 0.9628]), values=tensor([0.2505, 0.5332, 0.4314, ..., 0.9186, 0.8523, 0.0373]),
size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr) size=(10000, 10000), nnz=30000000, layout=torch.sparse_csr)
tensor([0.3885, 0.1371, 0.6444, ..., 0.5368, 0.0739, 0.9250]) tensor([0.5706, 0.2901, 0.6021, ..., 0.8123, 0.3136, 0.2413])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([10000, 10000]) Shape: torch.Size([10000, 10000])
@ -33,13 +33,13 @@ Rows: 10000
Size: 100000000 Size: 100000000
NNZ: 30000000 NNZ: 30000000
Density: 0.3 Density: 0.3
Time: 12.322984457015991 seconds Time: 11.716556549072266 seconds
[18.78, 17.72, 18.05, 18.08, 21.63, 19.29, 17.99, 17.89, 17.96, 17.7] [18.37, 17.73, 18.07, 17.59, 17.73, 17.59, 17.79, 17.61, 17.6, 17.94]
[53.17] [53.31]
80.39062190055847 79.56606245040894
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 12.322984457015991, 'TIME_S_1KI': 12.322984457015991, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 4274.3693664526945, 'W': 53.17000000000001} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 11.716556549072266, 'TIME_S_1KI': 11.716556549072266, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 4241.666789231301, 'W': 53.31}
[18.78, 17.72, 18.05, 18.08, 21.63, 19.29, 17.99, 17.89, 17.96, 17.7, 18.37, 17.7, 18.04, 17.87, 17.81, 17.74, 17.89, 18.06, 17.94, 18.12] [18.37, 17.73, 18.07, 17.59, 17.73, 17.59, 17.79, 17.61, 17.6, 17.94, 18.37, 17.63, 18.42, 21.66, 17.64, 17.42, 18.81, 17.54, 17.75, 17.96]
328.145 324.90000000000003
16.407249999999998 16.245
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 12.322984457015991, 'TIME_S_1KI': 12.322984457015991, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 4274.3693664526945, 'W': 53.17000000000001, 'J_1KI': 4274.3693664526945, 'W_1KI': 53.17000000000001, 'W_D': 36.76275000000001, 'J_D': 2955.380335274757, 'W_D_1KI': 36.76275000000001, 'J_D_1KI': 36.76275000000001} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 1000, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [10000, 10000], 'MATRIX_ROWS': 10000, 'MATRIX_SIZE': 100000000, 'MATRIX_NNZ': 30000000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 11.716556549072266, 'TIME_S_1KI': 11.716556549072266, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 4241.666789231301, 'W': 53.31, 'J_1KI': 4241.666789231301, 'W_1KI': 53.31, 'W_D': 37.065, 'J_D': 2949.116104724407, 'W_D_1KI': 37.065, 'J_D_1KI': 37.065}

2
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_5000_0.1.json
View File

@ -1 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 19209, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.589914798736572, "TIME_S_1KI": 0.5512996407276054, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1296.501946372986, "W": 88.72, "J_1KI": 67.49450499104513, "W_1KI": 4.618668332552449, "W_D": 72.5775, "J_D": 1060.6049370253086, "W_D_1KI": 3.778307043573325, "J_D_1KI": 0.19669462458083842} {"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 18622, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.317775011062622, "TIME_S_1KI": 0.5540637424048234, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1264.8904221296311, "W": 88.65, "J_1KI": 67.9245205740324, "W_1KI": 4.760498335302331, "W_D": 72.60650000000001, "J_D": 1035.9759327056408, "W_D_1KI": 3.8989635914509724, "J_D_1KI": 0.2093740517372448}

58
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_5000_0.1.output
View File

@ -1,14 +1,14 @@
['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', '5000', '-sd', '0.1', '-c', '16'] ['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', '5000', '-sd', '0.1', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 0.5466129779815674} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 0.5638444423675537}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 514, 1026, ..., 2498984, tensor(crow_indices=tensor([ 0, 491, 1004, ..., 2498969,
2499497, 2500000]), 2499500, 2500000]),
col_indices=tensor([ 4, 26, 34, ..., 4975, 4994, 4997]), col_indices=tensor([ 22, 38, 49, ..., 4962, 4970, 4976]),
values=tensor([0.5421, 0.0550, 0.0297, ..., 0.2626, 0.0439, 0.1648]), values=tensor([0.5529, 0.3874, 0.4848, ..., 0.5629, 0.9931, 0.9487]),
size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.7995, 0.3197, 0.3485, ..., 0.5295, 0.0131, 0.9723]) tensor([0.1713, 0.2154, 0.4105, ..., 0.0933, 0.5075, 0.3878])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -16,19 +16,19 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 2500000 NNZ: 2500000
Density: 0.1 Density: 0.1
Time: 0.5466129779815674 seconds Time: 0.5638444423675537 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', '19209', '-ss', '5000', '-sd', '0.1', '-c', '16'] ['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', '18622', '-ss', '5000', '-sd', '0.1', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.589914798736572} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 2500000, "MATRIX_DENSITY": 0.1, "TIME_S": 10.317775011062622}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 503, 1003, ..., 2498928, tensor(crow_indices=tensor([ 0, 525, 987, ..., 2498981,
2499468, 2500000]), 2499483, 2500000]),
col_indices=tensor([ 6, 8, 21, ..., 4933, 4958, 4973]), col_indices=tensor([ 8, 19, 22, ..., 4987, 4993, 4999]),
values=tensor([0.5143, 0.8442, 0.2205, ..., 0.0567, 0.9724, 0.7726]), values=tensor([0.0390, 0.1537, 0.3143, ..., 0.0698, 0.3942, 0.2310]),
size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.8237, 0.2559, 0.0746, ..., 0.2976, 0.1284, 0.3075]) tensor([0.3040, 0.7679, 0.7012, ..., 0.6231, 0.1253, 0.7846])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -36,16 +36,16 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 2500000 NNZ: 2500000
Density: 0.1 Density: 0.1
Time: 10.589914798736572 seconds Time: 10.317775011062622 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 503, 1003, ..., 2498928, tensor(crow_indices=tensor([ 0, 525, 987, ..., 2498981,
2499468, 2500000]), 2499483, 2500000]),
col_indices=tensor([ 6, 8, 21, ..., 4933, 4958, 4973]), col_indices=tensor([ 8, 19, 22, ..., 4987, 4993, 4999]),
values=tensor([0.5143, 0.8442, 0.2205, ..., 0.0567, 0.9724, 0.7726]), values=tensor([0.0390, 0.1537, 0.3143, ..., 0.0698, 0.3942, 0.2310]),
size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=2500000, layout=torch.sparse_csr)
tensor([0.8237, 0.2559, 0.0746, ..., 0.2976, 0.1284, 0.3075]) tensor([0.3040, 0.7679, 0.7012, ..., 0.6231, 0.1253, 0.7846])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -53,13 +53,13 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 2500000 NNZ: 2500000
Density: 0.1 Density: 0.1
Time: 10.589914798736572 seconds Time: 10.317775011062622 seconds
[18.45, 18.0, 17.83, 17.78, 17.91, 17.57, 18.07, 17.82, 17.87, 17.88] [18.14, 17.77, 17.8, 18.17, 17.99, 17.63, 17.94, 18.68, 17.91, 17.77]
[88.72] [88.65]
14.613412380218506 14.26836347579956
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 19209, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.589914798736572, 'TIME_S_1KI': 0.5512996407276054, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1296.501946372986, 'W': 88.72} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 18622, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.317775011062622, 'TIME_S_1KI': 0.5540637424048234, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1264.8904221296311, 'W': 88.65}
[18.45, 18.0, 17.83, 17.78, 17.91, 17.57, 18.07, 17.82, 17.87, 17.88, 18.16, 17.97, 18.01, 17.95, 17.55, 18.45, 18.51, 17.65, 17.73, 17.87] [18.14, 17.77, 17.8, 18.17, 17.99, 17.63, 17.94, 18.68, 17.91, 17.77, 17.91, 17.46, 17.97, 17.49, 17.74, 17.66, 17.77, 17.61, 17.59, 17.56]
322.85 320.87
16.142500000000002 16.0435
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 19209, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.589914798736572, 'TIME_S_1KI': 0.5512996407276054, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1296.501946372986, 'W': 88.72, 'J_1KI': 67.49450499104513, 'W_1KI': 4.618668332552449, 'W_D': 72.5775, 'J_D': 1060.6049370253086, 'W_D_1KI': 3.778307043573325, 'J_D_1KI': 0.19669462458083842} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 18622, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 2500000, 'MATRIX_DENSITY': 0.1, 'TIME_S': 10.317775011062622, 'TIME_S_1KI': 0.5540637424048234, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1264.8904221296311, 'W': 88.65, 'J_1KI': 67.9245205740324, 'W_1KI': 4.760498335302331, 'W_D': 72.60650000000001, 'J_D': 1035.9759327056408, 'W_D_1KI': 3.8989635914509724, 'J_D_1KI': 0.2093740517372448}

2
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_5000_0.2.json
View File

@ -1 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 8907, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.383960962295532, "TIME_S_1KI": 1.1658202494998913, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1297.8042968559266, "W": 87.12, "J_1KI": 145.7061072028659, "W_1KI": 9.781071067699562, "W_D": 70.95125, "J_D": 1056.9425748082995, "W_D_1KI": 7.965785337375098, "J_D_1KI": 0.8943286558184684} {"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 8928, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.436826467514038, "TIME_S_1KI": 1.1689993803219128, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1316.4306947374344, "W": 86.99, "J_1KI": 147.4496745897664, "W_1KI": 9.743503584229389, "W_D": 70.63274999999999, "J_D": 1068.894357440114, "W_D_1KI": 7.911374327956988, "J_D_1KI": 0.8861306370919565}

58
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_5000_0.2.output
View File

@ -1,14 +1,14 @@
['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', '5000', '-sd', '0.2', '-c', '16'] ['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', '5000', '-sd', '0.2', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 1.1788151264190674} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 1.1760613918304443}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1020, 2047, ..., 4997993, tensor(crow_indices=tensor([ 0, 978, 2025, ..., 4997968,
4999046, 5000000]), 4998985, 5000000]),
col_indices=tensor([ 0, 5, 11, ..., 4968, 4973, 4981]), col_indices=tensor([ 2, 4, 11, ..., 4986, 4990, 4994]),
values=tensor([0.5537, 0.3394, 0.2735, ..., 0.5110, 0.5845, 0.1364]), values=tensor([0.0122, 0.4949, 0.8008, ..., 0.1011, 0.5261, 0.6800]),
size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr) size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.2008, 0.9176, 0.6130, ..., 0.2299, 0.3337, 0.1107]) tensor([0.4016, 0.7392, 0.9091, ..., 0.8685, 0.7503, 0.4791])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -16,19 +16,19 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 5000000 NNZ: 5000000
Density: 0.2 Density: 0.2
Time: 1.1788151264190674 seconds Time: 1.1760613918304443 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', '8907', '-ss', '5000', '-sd', '0.2', '-c', '16'] ['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', '8928', '-ss', '5000', '-sd', '0.2', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.383960962295532} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 5000000, "MATRIX_DENSITY": 0.2, "TIME_S": 10.436826467514038}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1043, 2083, ..., 4998055, tensor(crow_indices=tensor([ 0, 990, 1955, ..., 4998009,
4999016, 5000000]), 4998983, 5000000]),
col_indices=tensor([ 11, 19, 21, ..., 4992, 4995, 4997]), col_indices=tensor([ 11, 14, 16, ..., 4997, 4998, 4999]),
values=tensor([0.8807, 0.7903, 0.9771, ..., 0.4816, 0.8305, 0.1823]), values=tensor([0.2669, 0.2498, 0.4510, ..., 0.5817, 0.0466, 0.6772]),
size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr) size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.8731, 0.0343, 0.1598, ..., 0.1067, 0.9050, 0.6111]) tensor([0.9945, 0.9638, 0.1643, ..., 0.2698, 0.5025, 0.0642])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -36,16 +36,16 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 5000000 NNZ: 5000000
Density: 0.2 Density: 0.2
Time: 10.383960962295532 seconds Time: 10.436826467514038 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1043, 2083, ..., 4998055, tensor(crow_indices=tensor([ 0, 990, 1955, ..., 4998009,
4999016, 5000000]), 4998983, 5000000]),
col_indices=tensor([ 11, 19, 21, ..., 4992, 4995, 4997]), col_indices=tensor([ 11, 14, 16, ..., 4997, 4998, 4999]),
values=tensor([0.8807, 0.7903, 0.9771, ..., 0.4816, 0.8305, 0.1823]), values=tensor([0.2669, 0.2498, 0.4510, ..., 0.5817, 0.0466, 0.6772]),
size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr) size=(5000, 5000), nnz=5000000, layout=torch.sparse_csr)
tensor([0.8731, 0.0343, 0.1598, ..., 0.1067, 0.9050, 0.6111]) tensor([0.9945, 0.9638, 0.1643, ..., 0.2698, 0.5025, 0.0642])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -53,13 +53,13 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 5000000 NNZ: 5000000
Density: 0.2 Density: 0.2
Time: 10.383960962295532 seconds Time: 10.436826467514038 seconds
[17.98, 21.91, 18.08, 17.67, 17.68, 17.88, 18.1, 17.78, 17.61, 17.52] [18.18, 17.97, 22.4, 18.19, 17.69, 18.6, 17.87, 17.63, 17.75, 17.93]
[87.12] [86.99]
14.896743535995483 15.133126735687256
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 8907, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.383960962295532, 'TIME_S_1KI': 1.1658202494998913, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1297.8042968559266, 'W': 87.12} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 8928, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.436826467514038, 'TIME_S_1KI': 1.1689993803219128, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1316.4306947374344, 'W': 86.99}
[17.98, 21.91, 18.08, 17.67, 17.68, 17.88, 18.1, 17.78, 17.61, 17.52, 18.28, 17.97, 17.57, 17.52, 17.69, 17.59, 17.67, 17.39, 17.74, 17.27] [18.18, 17.97, 22.4, 18.19, 17.69, 18.6, 17.87, 17.63, 17.75, 17.93, 18.01, 17.97, 18.36, 17.9, 17.82, 17.52, 18.18, 17.58, 17.7, 17.91]
323.375 327.145
16.16875 16.35725
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 8907, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.383960962295532, 'TIME_S_1KI': 1.1658202494998913, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1297.8042968559266, 'W': 87.12, 'J_1KI': 145.7061072028659, 'W_1KI': 9.781071067699562, 'W_D': 70.95125, 'J_D': 1056.9425748082995, 'W_D_1KI': 7.965785337375098, 'J_D_1KI': 0.8943286558184684} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 8928, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 5000000, 'MATRIX_DENSITY': 0.2, 'TIME_S': 10.436826467514038, 'TIME_S_1KI': 1.1689993803219128, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1316.4306947374344, 'W': 86.99, 'J_1KI': 147.4496745897664, 'W_1KI': 9.743503584229389, 'W_D': 70.63274999999999, 'J_D': 1068.894357440114, 'W_D_1KI': 7.911374327956988, 'J_D_1KI': 0.8861306370919565}

2
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_5000_0.3.json
View File

@ -1 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 5572, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.464208841323853, "TIME_S_1KI": 1.8779987152411795, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1351.3364907836915, "W": 85.76, "J_1KI": 242.5227011456733, "W_1KI": 15.391241923905243, "W_D": 69.66475, "J_D": 1097.7206016362309, "W_D_1KI": 12.502647164393395, "J_D_1KI": 2.2438347387640696} {"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 5478, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.325071811676025, "TIME_S_1KI": 1.8848250842782084, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1339.3279536104203, "W": 85.87, "J_1KI": 244.49214195151885, "W_1KI": 15.675428988682002, "W_D": 69.74000000000001, "J_D": 1087.7457957935335, "W_D_1KI": 12.730923694779118, "J_D_1KI": 2.3240094367979403}

58
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_5000_0.3.output
View File

@ -1,14 +1,14 @@
['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', '5000', '-sd', '0.3', '-c', '16'] ['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', '5000', '-sd', '0.3', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 1.8840982913970947} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 1.9164557456970215}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1512, 3045, ..., 7497001, tensor(crow_indices=tensor([ 0, 1507, 2972, ..., 7496967,
7498469, 7500000]), 7498514, 7500000]),
col_indices=tensor([ 1, 3, 6, ..., 4985, 4987, 4999]), col_indices=tensor([ 0, 1, 7, ..., 4979, 4980, 4982]),
values=tensor([0.2926, 0.1712, 0.0979, ..., 0.0306, 0.4197, 0.0742]), values=tensor([0.5806, 0.6588, 0.4500, ..., 0.8030, 0.5443, 0.1304]),
size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr)
tensor([0.8071, 0.3432, 0.7096, ..., 0.6319, 0.9579, 0.4287]) tensor([0.5421, 0.7085, 0.6995, ..., 0.1955, 0.6605, 0.4332])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -16,19 +16,19 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 7500000 NNZ: 7500000
Density: 0.3 Density: 0.3
Time: 1.8840982913970947 seconds Time: 1.9164557456970215 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', '5572', '-ss', '5000', '-sd', '0.3', '-c', '16'] ['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', '5478', '-ss', '5000', '-sd', '0.3', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.464208841323853} {"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 7500000, "MATRIX_DENSITY": 0.3, "TIME_S": 10.325071811676025}
/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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1440, 2946, ..., 7496997, tensor(crow_indices=tensor([ 0, 1469, 2909, ..., 7496983,
7498526, 7500000]), 7498509, 7500000]),
col_indices=tensor([ 4, 11, 17, ..., 4977, 4981, 4999]), col_indices=tensor([ 2, 8, 9, ..., 4995, 4997, 4998]),
values=tensor([0.5270, 0.2118, 0.1968, ..., 0.5430, 0.9062, 0.9328]), values=tensor([0.1494, 0.3197, 0.4986, ..., 0.6801, 0.8207, 0.3086]),
size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr)
tensor([0.6230, 0.9791, 0.5460, ..., 0.3523, 0.6316, 0.0066]) tensor([0.0051, 0.9261, 0.2746, ..., 0.2239, 0.8025, 0.6653])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -36,16 +36,16 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 7500000 NNZ: 7500000
Density: 0.3 Density: 0.3
Time: 10.464208841323853 seconds Time: 10.325071811676025 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.) /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) matrix = matrix.to_sparse_csr().type(torch.float32)
tensor(crow_indices=tensor([ 0, 1440, 2946, ..., 7496997, tensor(crow_indices=tensor([ 0, 1469, 2909, ..., 7496983,
7498526, 7500000]), 7498509, 7500000]),
col_indices=tensor([ 4, 11, 17, ..., 4977, 4981, 4999]), col_indices=tensor([ 2, 8, 9, ..., 4995, 4997, 4998]),
values=tensor([0.5270, 0.2118, 0.1968, ..., 0.5430, 0.9062, 0.9328]), values=tensor([0.1494, 0.3197, 0.4986, ..., 0.6801, 0.8207, 0.3086]),
size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr) size=(5000, 5000), nnz=7500000, layout=torch.sparse_csr)
tensor([0.6230, 0.9791, 0.5460, ..., 0.3523, 0.6316, 0.0066]) tensor([0.0051, 0.9261, 0.2746, ..., 0.2239, 0.8025, 0.6653])
Matrix Type: synthetic Matrix Type: synthetic
Matrix Format: csr Matrix Format: csr
Shape: torch.Size([5000, 5000]) Shape: torch.Size([5000, 5000])
@ -53,13 +53,13 @@ Rows: 5000
Size: 25000000 Size: 25000000
NNZ: 7500000 NNZ: 7500000
Density: 0.3 Density: 0.3
Time: 10.464208841323853 seconds Time: 10.325071811676025 seconds
[18.1, 18.05, 18.13, 17.56, 18.21, 17.81, 17.97, 17.66, 18.07, 17.92] [18.13, 17.6, 17.78, 17.86, 17.7, 17.85, 17.51, 17.77, 17.69, 17.5]
[85.76] [85.87]
15.757188558578491 15.597157955169678
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 5572, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.464208841323853, 'TIME_S_1KI': 1.8779987152411795, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1351.3364907836915, 'W': 85.76} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 5478, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.325071811676025, 'TIME_S_1KI': 1.8848250842782084, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1339.3279536104203, 'W': 85.87}
[18.1, 18.05, 18.13, 17.56, 18.21, 17.81, 17.97, 17.66, 18.07, 17.92, 17.91, 17.43, 18.3, 17.79, 17.88, 17.63, 18.23, 17.64, 17.78, 17.6] [18.13, 17.6, 17.78, 17.86, 17.7, 17.85, 17.51, 17.77, 17.69, 17.5, 18.03, 17.39, 17.78, 17.59, 17.86, 17.49, 17.59, 17.65, 21.57, 18.18]
321.905 322.6
16.09525 16.130000000000003
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 5572, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.464208841323853, 'TIME_S_1KI': 1.8779987152411795, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1351.3364907836915, 'W': 85.76, 'J_1KI': 242.5227011456733, 'W_1KI': 15.391241923905243, 'W_D': 69.66475, 'J_D': 1097.7206016362309, 'W_D_1KI': 12.502647164393395, 'J_D_1KI': 2.2438347387640696} {'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 5478, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 7500000, 'MATRIX_DENSITY': 0.3, 'TIME_S': 10.325071811676025, 'TIME_S_1KI': 1.8848250842782084, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1339.3279536104203, 'W': 85.87, 'J_1KI': 244.49214195151885, 'W_1KI': 15.675428988682002, 'W_D': 69.74000000000001, 'J_D': 1087.7457957935335, 'W_D_1KI': 12.730923694779118, 'J_D_1KI': 2.3240094367979403}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_5000_0.4.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 2943, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 11.241667032241821, "TIME_S_1KI": 3.819798515882372, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1393.030681221485, "W": 83.67, "J_1KI": 473.3369626984319, "W_1KI": 28.430173292558617, "W_D": 67.793, "J_D": 1128.6928286368848, "W_D_1KI": 23.035338090383963, "J_D_1KI": 7.82716211022221}

85
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_5000_0.4.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', '100', '-ss', '5000', '-sd', '0.4', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 0.38892388343811035}
/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, 2020, 4067, ..., 9996060,
9998056, 10000000]),
col_indices=tensor([ 1, 4, 6, ..., 4992, 4998, 4999]),
values=tensor([0.0057, 0.0348, 0.0719, ..., 0.9587, 0.3044, 0.9978]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.1784, 0.7215, 0.7937, ..., 0.4492, 0.0838, 0.9744])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 0.38892388343811035 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', '2699', '-ss', '5000', '-sd', '0.4', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 9.626921653747559}
/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, 1992, 4044, ..., 9995930,
9997986, 10000000]),
col_indices=tensor([ 0, 5, 10, ..., 4990, 4997, 4999]),
values=tensor([0.5930, 0.1714, 0.3969, ..., 0.4489, 0.5630, 0.4403]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.5546, 0.5609, 0.4280, ..., 0.4447, 0.9246, 0.2629])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 9.626921653747559 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', '2943', '-ss', '5000', '-sd', '0.4', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 10000000, "MATRIX_DENSITY": 0.4, "TIME_S": 11.241667032241821}
/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, 1928, 3951, ..., 9996029,
9997995, 10000000]),
col_indices=tensor([ 1, 3, 6, ..., 4994, 4996, 4997]),
values=tensor([0.0792, 0.4043, 0.0506, ..., 0.1635, 0.6350, 0.0332]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.4326, 0.5030, 0.1276, ..., 0.3674, 0.6623, 0.2020])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 11.241667032241821 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, 1928, 3951, ..., 9996029,
9997995, 10000000]),
col_indices=tensor([ 1, 3, 6, ..., 4994, 4996, 4997]),
values=tensor([0.0792, 0.4043, 0.0506, ..., 0.1635, 0.6350, 0.0332]),
size=(5000, 5000), nnz=10000000, layout=torch.sparse_csr)
tensor([0.4326, 0.5030, 0.1276, ..., 0.3674, 0.6623, 0.2020])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 10000000
Density: 0.4
Time: 11.241667032241821 seconds
[18.06, 17.64, 17.57, 17.56, 18.11, 17.49, 17.52, 17.53, 17.66, 17.49]
[83.67]
16.64910578727722
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 2943, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.4, 'TIME_S': 11.241667032241821, 'TIME_S_1KI': 3.819798515882372, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1393.030681221485, 'W': 83.67}
[18.06, 17.64, 17.57, 17.56, 18.11, 17.49, 17.52, 17.53, 17.66, 17.49, 17.99, 17.7, 17.78, 17.54, 17.51, 17.38, 17.97, 17.59, 17.4, 17.64]
317.53999999999996
15.876999999999999
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 2943, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 10000000, 'MATRIX_DENSITY': 0.4, 'TIME_S': 11.241667032241821, 'TIME_S_1KI': 3.819798515882372, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1393.030681221485, 'W': 83.67, 'J_1KI': 473.3369626984319, 'W_1KI': 28.430173292558617, 'W_D': 67.793, 'J_D': 1128.6928286368848, 'W_D_1KI': 23.035338090383963, 'J_D_1KI': 7.82716211022221}

1
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_5000_0.5.json Normal file
View File

@ -0,0 +1 @@
{"CPU": "Xeon 4216", "CORES": 16, "ITERATIONS": 2293, "MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 12500000, "MATRIX_DENSITY": 0.5, "TIME_S": 10.488942384719849, "TIME_S_1KI": 4.5743316113039025, "BASELINE_TIME_S": 10, "BASELINE_DELAY_S": 10, "J": 1464.8316407203674, "W": 79.5, "J_1KI": 638.827579904216, "W_1KI": 34.67073702573049, "W_D": 63.46325, "J_D": 1169.3456178987026, "W_D_1KI": 27.676951591801135, "J_D_1KI": 12.0701925825561}

85
pytorch/output_synthetic_16core/xeon_4216_16_csr_10_10_10_synthetic_5000_0.5.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', '100', '-ss', '5000', '-sd', '0.5', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 12500000, "MATRIX_DENSITY": 0.5, "TIME_S": 0.49056053161621094}
/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, 2534, 5049, ..., 12494993,
12497525, 12500000]),
col_indices=tensor([ 2, 3, 7, ..., 4995, 4997, 4999]),
values=tensor([0.8256, 0.5002, 0.0945, ..., 0.8539, 0.0059, 0.0754]),
size=(5000, 5000), nnz=12500000, layout=torch.sparse_csr)
tensor([0.6389, 0.8699, 0.2648, ..., 0.6441, 0.5427, 0.0050])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 12500000
Density: 0.5
Time: 0.49056053161621094 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', '2140', '-ss', '5000', '-sd', '0.5', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 12500000, "MATRIX_DENSITY": 0.5, "TIME_S": 9.796798944473267}
/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, 2473, 4961, ..., 12495026,
12497501, 12500000]),
col_indices=tensor([ 0, 1, 2, ..., 4993, 4995, 4998]),
values=tensor([0.3292, 0.8142, 0.0983, ..., 0.8374, 0.1805, 0.7414]),
size=(5000, 5000), nnz=12500000, layout=torch.sparse_csr)
tensor([0.5256, 0.8051, 0.2947, ..., 0.7660, 0.3711, 0.2787])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 12500000
Density: 0.5
Time: 9.796798944473267 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', '2293', '-ss', '5000', '-sd', '0.5', '-c', '16']
{"MATRIX_TYPE": "synthetic", "MATRIX_FORMAT": "csr", "MATRIX_SHAPE": [5000, 5000], "MATRIX_ROWS": 5000, "MATRIX_SIZE": 25000000, "MATRIX_NNZ": 12500000, "MATRIX_DENSITY": 0.5, "TIME_S": 10.488942384719849}
/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, 2483, 4907, ..., 12494946,
12497460, 12500000]),
col_indices=tensor([ 3, 6, 8, ..., 4993, 4997, 4999]),
values=tensor([0.9265, 0.6407, 0.1426, ..., 0.0639, 0.3199, 0.5517]),
size=(5000, 5000), nnz=12500000, layout=torch.sparse_csr)
tensor([0.6629, 0.1553, 0.8462, ..., 0.7358, 0.1133, 0.5218])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 12500000
Density: 0.5
Time: 10.488942384719849 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, 2483, 4907, ..., 12494946,
12497460, 12500000]),
col_indices=tensor([ 3, 6, 8, ..., 4993, 4997, 4999]),
values=tensor([0.9265, 0.6407, 0.1426, ..., 0.0639, 0.3199, 0.5517]),
size=(5000, 5000), nnz=12500000, layout=torch.sparse_csr)
tensor([0.6629, 0.1553, 0.8462, ..., 0.7358, 0.1133, 0.5218])
Matrix Type: synthetic
Matrix Format: csr
Shape: torch.Size([5000, 5000])
Rows: 5000
Size: 25000000
NNZ: 12500000
Density: 0.5
Time: 10.488942384719849 seconds
[18.25, 17.84, 17.64, 17.31, 17.75, 17.54, 17.53, 17.59, 17.7, 17.45]
[79.5]
18.42555522918701
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 2293, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 12500000, 'MATRIX_DENSITY': 0.5, 'TIME_S': 10.488942384719849, 'TIME_S_1KI': 4.5743316113039025, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1464.8316407203674, 'W': 79.5}
[18.25, 17.84, 17.64, 17.31, 17.75, 17.54, 17.53, 17.59, 17.7, 17.45, 18.03, 17.47, 17.56, 17.58, 17.59, 17.37, 21.25, 17.8, 17.44, 17.82]
320.735
16.03675
{'CPU': 'Xeon 4216', 'CORES': 16, 'ITERATIONS': 2293, 'MATRIX_TYPE': 'synthetic', 'MATRIX_FORMAT': 'csr', 'MATRIX_SHAPE': [5000, 5000], 'MATRIX_ROWS': 5000, 'MATRIX_SIZE': 25000000, 'MATRIX_NNZ': 12500000, 'MATRIX_DENSITY': 0.5, 'TIME_S': 10.488942384719849, 'TIME_S_1KI': 4.5743316113039025, 'BASELINE_TIME_S': 10, 'BASELINE_DELAY_S': 10, 'J': 1464.8316407203674, 'W': 79.5, 'J_1KI': 638.827579904216, 'W_1KI': 34.67073702573049, 'W_D': 63.46325, 'J_D': 1169.3456178987026, 'W_D_1KI': 27.676951591801135, 'J_D_1KI': 12.0701925825561}

3
pytorch/synthetic_densities
View File

@ -6,3 +6,6 @@
0.05 0.05
0.1 0.1
0.2 0.2
0.3
0.4
0.5