Baunsgaard commented on PR #2064: URL: https://github.com/apache/systemds/pull/2064#issuecomment-2289864260
after: ```txt MCSR to MCSR: --- faster especially with more threads. Input Size: 10000 x 10000 , sparsity: 0.100041 same, 0.123+- 0.020 ms, replace_div 2 Parallel: 16, 19.914+- 7.455 ms, replace_div 4 Parallel: 16, 14.524+- 2.185 ms, replace_div 5 Parallel: 16, 14.212+- 2.474 ms, replace_div 8 Parallel: 16, 15.831+- 2.863 ms, replace_div 10 Parallel: 16, 15.686+- 2.918 ms, replace_div 16 Parallel: 16, 15.785+- 1.338 ms, replace_div 20 Parallel: 16, 13.273+- 3.128 ms, replace_div 25 Parallel: 16, 15.810+- 1.173 ms, replace_div 40 Parallel: 16, 19.310+- 1.450 ms, replace_div 50 Parallel: 16, 20.257+- 0.976 ms, replace_div 80 Parallel: 16, 20.275+- 1.237 ms, // slowing because the GHz reduce on my laptop. ``` ```txt from MCSR to CSR -- not parallelized --- no change Input Size: 10000 x 10000 , sparsity: 0.100041 same, 0.001+- 0.000 ms, replace_div 2 Parallel: 1, 42.385+- 14.940 ms, replace_div 4 Parallel: 1, 21.943+- 0.597 ms, replace_div 5 Parallel: 1, 22.301+- 0.589 ms, replace_div 8 Parallel: 1, 22.262+- 0.616 ms, replace_div 10 Parallel: 1, 22.215+- 0.471 ms, replace_div 16 Parallel: 1, 21.738+- 0.254 ms, replace_div 20 Parallel: 1, 25.983+- 2.419 ms, replace_div 25 Parallel: 1, 25.399+- 1.891 ms, replace_div 40 Parallel: 1, 25.229+- 1.450 ms, replace_div 50 Parallel: 1, 27.832+- 2.815 ms, ``` ```txt CSR to CSR Input Size: 10000 x 10000 , sparsity: 0.100041 same, 0.001+- 0.000 ms, replace_div 2 Parallel: 1, 18.675+- 2.125 ms, replace_div 4 Parallel: 1, 18.011+- 1.650 ms, //not warmed up replace_div 5 Parallel: 1, 8.341+- 0.397 ms, replace_div 8 Parallel: 1, 8.596+- 0.318 ms, // actual replace_div 10 Parallel: 1, 8.031+- 0.107 ms, replace_div 16 Parallel: 1, 8.068+- 0.105 ms, replace_div 20 Parallel: 1, 8.094+- 0.112 ms, replace_div 25 Parallel: 1, 7.991+- 0.113 ms, replace_div 40 Parallel: 1, 8.095+- 0.237 ms, replace_div 50 Parallel: 1, 8.370+- 0.229 ms, ``` Before: ```txt MCSR to CSR Input Size: 10000 x 10000 , sparsity: 0.100041 same, 0.001+- 0.000 ms, replace_div 2 Parallel: 1, 26.289+- 2.820 ms, replace_div 4 Parallel: 1, 24.581+- 1.699 ms, replace_div 5 Parallel: 1, 24.536+- 1.951 ms, replace_div 8 Parallel: 1, 26.499+- 2.565 ms, replace_div 10 Parallel: 1, 31.506+- 5.144 ms, replace_div 16 Parallel: 1, 26.546+- 3.405 ms, replace_div 20 Parallel: 1, 28.454+- 4.591 ms, replace_div 25 Parallel: 1, 25.383+- 1.888 ms, replace_div 40 Parallel: 1, 25.796+- 2.262 ms, replace_div 50 Parallel: 1, 25.450+- 2.557 ms, MCSR to MCSR parallel Input Size: 10000 x 10000 , sparsity: 0.100041 same, 0.060+- 0.012 ms, replace_div 2 Parallel: 16, 30.955+- 11.512 ms, replace_div 4 Parallel: 16, 24.607+- 1.407 ms, replace_div 5 Parallel: 16, 26.062+- 1.623 ms, replace_div 8 Parallel: 16, 24.866+- 1.149 ms, replace_div 10 Parallel: 16, 24.846+- 2.307 ms, replace_div 16 Parallel: 16, 27.716+- 2.727 ms, replace_div 20 Parallel: 16, 25.680+- 2.053 ms, replace_div 25 Parallel: 16, 26.855+- 1.671 ms, replace_div 40 Parallel: 16, 25.532+- 1.742 ms, replace_div 50 Parallel: 16, 24.511+- 1.492 ms, CSR to CSR Input Size: 10000 x 10000 , sparsity: 0.100041 same, 0.001+- 0.000 ms, replace_div 2 Parallel: 1, 8.910+- 0.660 ms, replace_div 4 Parallel: 1, 8.683+- 0.391 ms, replace_div 5 Parallel: 1, 8.260+- 0.354 ms, replace_div 8 Parallel: 1, 8.217+- 0.443 ms, replace_div 10 Parallel: 1, 8.652+- 0.455 ms, replace_div 16 Parallel: 1, 8.246+- 0.366 ms, replace_div 20 Parallel: 1, 8.888+- 1.142 ms, replace_div 25 Parallel: 1, 8.067+- 0.348 ms, replace_div 40 Parallel: 1, 8.175+- 0.344 ms, replace_div 50 Parallel: 1, 8.173+- 0.355 ms, ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: dev-unsubscr...@systemds.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org