xref: /aosp_15_r20/external/pytorch/benchmarks/sparse/dlmc/README.md (revision da0073e96a02ea20f0ac840b70461e3646d07c45)
1*da0073e9SAndroid Build Coastguard Worker# Sparse benchmarks
2*da0073e9SAndroid Build Coastguard Worker
3*da0073e9SAndroid Build Coastguard WorkerThese sets of benchmarks are for the sparse matrix functionality using a popular real dataset collection called the Deep Learning Matrix Collection (DLMC), which were used in recent studies [1, 2].
4*da0073e9SAndroid Build Coastguard Worker
5*da0073e9SAndroid Build Coastguard WorkerPerformance benchmarks scripts for matrix-matrix and matrix-vector ops (dense-sparse, sparse-sparse, and compare to dense-dense) are implemented here.
6*da0073e9SAndroid Build Coastguard Worker
7*da0073e9SAndroid Build Coastguard Worker- `matmul_bench.py` with `--operation sparse@sparse|sparse@dense` is for Sparse matrix-matrix multiplication (SPMM) performance test. It can run in forward and backward mode with `--backward-test`, on CPU or CUDA with `--with-cuda`, using different datasets from the dataset collection DLMC. For more details see `test.sh` file.
8*da0073e9SAndroid Build Coastguard Worker
9*da0073e9SAndroid Build Coastguard Worker- `matmul_bench.py` with `--operation sparse@vector` is for Sparse matrix-vector multiplication (SPMV) performance test.
10*da0073e9SAndroid Build Coastguard Worker
11*da0073e9SAndroid Build Coastguard WorkerReferences:
12*da0073e9SAndroid Build Coastguard Worker
13*da0073e9SAndroid Build Coastguard Worker1. Trevor Gale, Matei Zaharia, Cliff Young, Erich Elsen. Sparse GPU Kernels for Deep Learning. Proceedings of the International Conference for High Performance Computing, 2020. https://github.com/google-research/google-research/tree/master/sgk
14*da0073e9SAndroid Build Coastguard Worker
15*da0073e9SAndroid Build Coastguard Worker2. Trevor Gale, Erich Elsen, Sara Hooker. The State of Sparsity in Deep Neural Networks. https://github.com/google-research/google-research/tree/master/state_of_sparsity
16