srun: Job time limit was unset; set to partition default of 60 minutes srun: ################################################################################ srun: # Please note that the oasis compute nodes have aarch64 architecture CPUs. # srun: # All submission nodes and all other compute nodes have x86_64 architecture # srun: # CPUs. Programs, environments, or other software that was built on x86_64 # srun: # nodes may need to be rebuilt to properly execute on these nodes. # srun: ################################################################################ srun: job 3394988 queued and waiting for resources srun: job 3394988 has been allocated resources /nfshomes/vut/ampere_research/pytorch/spmv.py:20: UserWarning: Sparse CSR tensor support is in beta state. If 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.) ).to_sparse_csr().type(torch.float) tensor(crow_indices=tensor([ 0, 4, 7, ..., 155588, 155592, 155598]), col_indices=tensor([ 131, 561, 996, ..., 32237, 32238, 32570]), values=tensor([79040., 7820., 15136., ..., 2828., 17986., 2482.]), size=(32580, 32580), nnz=155598, layout=torch.sparse_csr) tensor([0.2022, 0.3400, 0.2561, ..., 0.8370, 0.0285, 0.6506]) Matrix: vt2010 Shape: torch.Size([32580, 32580]) NNZ: 155598 Density: 0.00014658915806621921 Time: 3.74875545501709 seconds Performance counter stats for 'apptainer run pytorch-altra.sif -c numactl --cpunodebind=0 --membind=0 python spmv.py matrices/vt2010.mtx 1000': 48.59 msec task-clock:u # 0.007 CPUs utilized 0 context-switches:u # 0.000 /sec 0 cpu-migrations:u # 0.000 /sec 3,274 page-faults:u # 67.376 K/sec 55,030,923 cycles:u # 1.132 GHz (65.54%) 78,222,423 instructions:u # 1.42 insn per cycle (83.60%) branches:u 369,917 branch-misses:u 32,435,815 L1-dcache-loads:u # 667.500 M/sec 467,963 L1-dcache-load-misses:u # 1.44% of all L1-dcache accesses LLC-loads:u LLC-load-misses:u 31,013,287 L1-icache-loads:u # 638.226 M/sec 289,982 L1-icache-load-misses:u # 0.94% of all L1-icache accesses 60,644,978 dTLB-loads:u # 1.248 G/sec (17.29%) dTLB-load-misses:u (0.00%) iTLB-loads:u (0.00%) iTLB-load-misses:u (0.00%) 6.978143797 seconds time elapsed 18.401752000 seconds user 28.060858000 seconds sys /nfshomes/vut/ampere_research/pytorch/spmv.py:20: UserWarning: Sparse CSR tensor support is in beta state. If 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.) ).to_sparse_csr().type(torch.float) tensor(crow_indices=tensor([ 0, 4, 7, ..., 155588, 155592, 155598]), col_indices=tensor([ 131, 561, 996, ..., 32237, 32238, 32570]), values=tensor([79040., 7820., 15136., ..., 2828., 17986., 2482.]), size=(32580, 32580), nnz=155598, layout=torch.sparse_csr) tensor([0.3381, 0.0423, 0.5363, ..., 0.0429, 0.4077, 0.4744]) Matrix: vt2010 Shape: torch.Size([32580, 32580]) NNZ: 155598 Density: 0.00014658915806621921 Time: 3.7925527095794678 seconds Performance counter stats for 'apptainer run pytorch-altra.sif -c numactl --cpunodebind=0 --membind=0 python spmv.py matrices/vt2010.mtx 1000': 323,004 BR_MIS_PRED_RETIRED:u # 0.0 per branch branch_misprediction_ratio 19,091,130 BR_RETIRED:u 7.233250772 seconds time elapsed 19.111768000 seconds user 32.178633000 seconds sys /nfshomes/vut/ampere_research/pytorch/spmv.py:20: UserWarning: Sparse CSR tensor support is in beta state. If 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.) ).to_sparse_csr().type(torch.float) tensor(crow_indices=tensor([ 0, 4, 7, ..., 155588, 155592, 155598]), col_indices=tensor([ 131, 561, 996, ..., 32237, 32238, 32570]), values=tensor([79040., 7820., 15136., ..., 2828., 17986., 2482.]), size=(32580, 32580), nnz=155598, layout=torch.sparse_csr) tensor([0.7962, 0.6492, 0.2778, ..., 0.5407, 0.1159, 0.3587]) Matrix: vt2010 Shape: torch.Size([32580, 32580]) NNZ: 155598 Density: 0.00014658915806621921 Time: 3.668635129928589 seconds Performance counter stats for 'apptainer run pytorch-altra.sif -c numactl --cpunodebind=0 --membind=0 python spmv.py matrices/vt2010.mtx 1000': 27,178,617 L1I_TLB:u # 0.0 per TLB access itlb_walk_ratio 6,398 ITLB_WALK:u 19,770 DTLB_WALK:u # 0.0 per TLB access dtlb_walk_ratio 36,355,567 L1D_TLB:u 6.925944164 seconds time elapsed 18.970654000 seconds user 30.786317000 seconds sys /nfshomes/vut/ampere_research/pytorch/spmv.py:20: UserWarning: Sparse CSR tensor support is in beta state. If 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.) ).to_sparse_csr().type(torch.float) tensor(crow_indices=tensor([ 0, 4, 7, ..., 155588, 155592, 155598]), col_indices=tensor([ 131, 561, 996, ..., 32237, 32238, 32570]), values=tensor([79040., 7820., 15136., ..., 2828., 17986., 2482.]), size=(32580, 32580), nnz=155598, layout=torch.sparse_csr) tensor([0.8340, 0.3434, 0.3449, ..., 0.9828, 0.6683, 0.0312]) Matrix: vt2010 Shape: torch.Size([32580, 32580]) NNZ: 155598 Density: 0.00014658915806621921 Time: 3.623232126235962 seconds Performance counter stats for 'apptainer run pytorch-altra.sif -c numactl --cpunodebind=0 --membind=0 python spmv.py matrices/vt2010.mtx 1000': 31,341,858 L1I_CACHE:u # 0.0 per cache access l1i_cache_miss_ratio 291,951 L1I_CACHE_REFILL:u 468,242 L1D_CACHE_REFILL:u # 0.0 per cache access l1d_cache_miss_ratio 32,805,413 L1D_CACHE:u 6.941260499 seconds time elapsed 18.410270000 seconds user 27.908787000 seconds sys /nfshomes/vut/ampere_research/pytorch/spmv.py:20: UserWarning: Sparse CSR tensor support is in beta state. If 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.) ).to_sparse_csr().type(torch.float) tensor(crow_indices=tensor([ 0, 4, 7, ..., 155588, 155592, 155598]), col_indices=tensor([ 131, 561, 996, ..., 32237, 32238, 32570]), values=tensor([79040., 7820., 15136., ..., 2828., 17986., 2482.]), size=(32580, 32580), nnz=155598, layout=torch.sparse_csr) tensor([0.2754, 0.3661, 0.9484, ..., 0.7285, 0.5354, 0.4116]) Matrix: vt2010 Shape: torch.Size([32580, 32580]) NNZ: 155598 Density: 0.00014658915806621921 Time: 3.7337992191314697 seconds Performance counter stats for 'apptainer run pytorch-altra.sif -c numactl --cpunodebind=0 --membind=0 python spmv.py matrices/vt2010.mtx 1000': 520,057 LL_CACHE_MISS_RD:u # 1.0 per cache access ll_cache_read_miss_ratio 541,186 LL_CACHE_RD:u 191,068 L2D_TLB:u # 0.1 per TLB access l2_tlb_miss_ratio 22,725 L2D_TLB_REFILL:u 288,895 L2D_CACHE_REFILL:u # 0.2 per cache access l2_cache_miss_ratio 1,728,320 L2D_CACHE:u 7.164825085 seconds time elapsed 18.193885000 seconds user 30.023194000 seconds sys