From 6bd19c272fe37e8b3f157afc33a4eb0870347927 Mon Sep 17 00:00:00 2001 From: Ruipeng Li Date: Thu, 1 Sep 2022 08:25:03 -0700 Subject: [PATCH] new benchmark results on tioga (#727) Improved benchmark performance on tioga. --- .../TEST_bench/benchmark_ij.perf.saved.tioga | 108 +++++++++--------- .../benchmark_spgemm.perf.saved.tioga | 72 ++++++------ .../benchmark_struct.perf.saved.tioga | 32 +++--- 3 files changed, 106 insertions(+), 106 deletions(-) diff --git a/src/test/TEST_bench/benchmark_ij.perf.saved.tioga b/src/test/TEST_bench/benchmark_ij.perf.saved.tioga index 50c3f9fae..619f22b45 100644 --- a/src/test/TEST_bench/benchmark_ij.perf.saved.tioga +++ b/src/test/TEST_bench/benchmark_ij.perf.saved.tioga @@ -1,81 +1,81 @@ # Output file: benchmark_ij.out.1 -PCG Setup wall clock time = 0.359271 seconds -PCG Solve wall clock time = 0.504610 seconds +PCG Setup wall clock time = 0.194717 seconds +PCG Solve wall clock time = 0.484421 seconds # Output file: benchmark_ij.out.2 -PCG Setup wall clock time = 0.312138 seconds -PCG Solve wall clock time = 0.416958 seconds +PCG Setup wall clock time = 0.181343 seconds +PCG Solve wall clock time = 0.378145 seconds # Output file: benchmark_ij.out.3 -PCG Setup wall clock time = 0.448648 seconds -PCG Solve wall clock time = 1.005090 seconds +PCG Setup wall clock time = 0.227025 seconds +PCG Solve wall clock time = 0.938486 seconds # Output file: benchmark_ij.out.4 -PCG Setup wall clock time = 0.477832 seconds -PCG Solve wall clock time = 0.215731 seconds +PCG Setup wall clock time = 0.258373 seconds +PCG Solve wall clock time = 0.204451 seconds # Output file: benchmark_ij.out.5 -PCG Setup wall clock time = 0.433146 seconds -PCG Solve wall clock time = 0.182600 seconds +PCG Setup wall clock time = 0.278769 seconds +PCG Solve wall clock time = 0.181327 seconds # Output file: benchmark_ij.out.6 -PCG Setup wall clock time = 0.419559 seconds -PCG Solve wall clock time = 0.929381 seconds +PCG Setup wall clock time = 0.219248 seconds +PCG Solve wall clock time = 0.822896 seconds # Output file: benchmark_ij.out.7 -PCG Setup wall clock time = 0.498759 seconds -PCG Solve wall clock time = 0.217208 seconds +PCG Setup wall clock time = 0.313280 seconds +PCG Solve wall clock time = 0.206185 seconds # Output file: benchmark_ij.out.8 -PCG Setup wall clock time = 0.336192 seconds -PCG Solve wall clock time = 0.492642 seconds +PCG Setup wall clock time = 0.177377 seconds +PCG Solve wall clock time = 0.473177 seconds # Output file: benchmark_ij.out.9 -PCG Setup wall clock time = 0.317082 seconds -PCG Solve wall clock time = 0.408582 seconds +PCG Setup wall clock time = 0.169903 seconds +PCG Solve wall clock time = 0.383708 seconds # Output file: benchmark_ij.out.10 -PCG Setup wall clock time = 0.415598 seconds -PCG Solve wall clock time = 0.956484 seconds +PCG Setup wall clock time = 0.209733 seconds +PCG Solve wall clock time = 0.888684 seconds # Output file: benchmark_ij.out.11 -PCG Setup wall clock time = 0.464038 seconds -PCG Solve wall clock time = 0.228662 seconds +PCG Setup wall clock time = 0.246183 seconds +PCG Solve wall clock time = 0.204804 seconds # Output file: benchmark_ij.out.12 -PCG Setup wall clock time = 0.400749 seconds -PCG Solve wall clock time = 0.179209 seconds +PCG Setup wall clock time = 0.266770 seconds +PCG Solve wall clock time = 0.174594 seconds # Output file: benchmark_ij.out.13 -PCG Setup wall clock time = 0.459325 seconds -PCG Solve wall clock time = 0.950444 seconds +PCG Setup wall clock time = 0.204547 seconds +PCG Solve wall clock time = 0.822958 seconds # Output file: benchmark_ij.out.14 -PCG Setup wall clock time = 0.969487 seconds -PCG Solve wall clock time = 0.908802 seconds +PCG Setup wall clock time = 0.975108 seconds +PCG Solve wall clock time = 0.908121 seconds # Output file: benchmark_ij.out.15 -PCG Setup wall clock time = 0.883830 seconds -PCG Solve wall clock time = 0.983369 seconds +PCG Setup wall clock time = 0.881669 seconds +PCG Solve wall clock time = 0.981050 seconds # Output file: benchmark_ij.out.16 -PCG Setup wall clock time = 0.901717 seconds -PCG Solve wall clock time = 0.867845 seconds +PCG Setup wall clock time = 0.903953 seconds +PCG Solve wall clock time = 0.866769 seconds # Output file: benchmark_ij.out.17 -PCG Setup wall clock time = 0.466032 seconds -PCG Solve wall clock time = 0.188925 seconds +PCG Setup wall clock time = 0.265580 seconds +PCG Solve wall clock time = 0.184383 seconds # Output file: benchmark_ij.out.18 -PCG Setup wall clock time = 0.514625 seconds -PCG Solve wall clock time = 0.164335 seconds +PCG Setup wall clock time = 0.262278 seconds +PCG Solve wall clock time = 0.152684 seconds # Output file: benchmark_ij.out.19 -PCG Setup wall clock time = 0.510175 seconds -PCG Solve wall clock time = 0.466027 seconds +PCG Setup wall clock time = 0.304177 seconds +PCG Solve wall clock time = 0.451599 seconds # Output file: benchmark_ij.out.20 -PCG Setup wall clock time = 0.303927 seconds -PCG Solve wall clock time = 0.249723 seconds +PCG Setup wall clock time = 0.209579 seconds +PCG Solve wall clock time = 0.246713 seconds # Output file: benchmark_ij.out.21 -PCG Setup wall clock time = 0.477654 seconds -PCG Solve wall clock time = 0.243726 seconds +PCG Setup wall clock time = 0.270481 seconds +PCG Solve wall clock time = 0.237917 seconds # Output file: benchmark_ij.out.22 -PCG Setup wall clock time = 0.510603 seconds -PCG Solve wall clock time = 0.187041 seconds +PCG Setup wall clock time = 0.273456 seconds +PCG Solve wall clock time = 0.170919 seconds # Output file: benchmark_ij.out.23 -PCG Setup wall clock time = 0.538188 seconds -PCG Solve wall clock time = 0.192323 seconds +PCG Setup wall clock time = 0.268130 seconds +PCG Solve wall clock time = 0.175079 seconds # Output file: benchmark_ij.out.24 -PCG Setup wall clock time = 0.484426 seconds -PCG Solve wall clock time = 0.203396 seconds +PCG Setup wall clock time = 0.244931 seconds +PCG Solve wall clock time = 0.188812 seconds # Output file: benchmark_ij.out.25 -PCG Setup wall clock time = 0.504628 seconds -PCG Solve wall clock time = 0.211952 seconds +PCG Setup wall clock time = 0.236667 seconds +PCG Solve wall clock time = 0.205809 seconds # Output file: benchmark_ij.out.26 -PCG Setup wall clock time = 0.479777 seconds -PCG Solve wall clock time = 0.542081 seconds +PCG Setup wall clock time = 0.289616 seconds +PCG Solve wall clock time = 0.536969 seconds # Output file: benchmark_ij.out.27 -PCG Setup wall clock time = 0.539575 seconds -PCG Solve wall clock time = 0.392673 seconds +PCG Setup wall clock time = 0.310268 seconds +PCG Solve wall clock time = 0.384598 seconds diff --git a/src/test/TEST_bench/benchmark_spgemm.perf.saved.tioga b/src/test/TEST_bench/benchmark_spgemm.perf.saved.tioga index f89b12122..baace3103 100644 --- a/src/test/TEST_bench/benchmark_spgemm.perf.saved.tioga +++ b/src/test/TEST_bench/benchmark_spgemm.perf.saved.tioga @@ -1,72 +1,72 @@ # Output file: benchmark_spgemm.out.1 -Device Parcsr Matrix-by-Matrix wall clock time = 0.019733 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.019889 seconds # Output file: benchmark_spgemm.out.2 -Device Parcsr Matrix-by-Matrix wall clock time = 0.077451 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.076819 seconds # Output file: benchmark_spgemm.out.3 -Device Parcsr Matrix-by-Matrix wall clock time = 0.008892 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.008884 seconds # Output file: benchmark_spgemm.out.4 -Device Parcsr Matrix-by-Matrix wall clock time = 0.009887 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.010011 seconds # Output file: benchmark_spgemm.out.5 -Device Parcsr Matrix-by-Matrix wall clock time = 0.008617 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.008854 seconds # Output file: benchmark_spgemm.out.6 -Device Parcsr Matrix-by-Matrix wall clock time = 0.082292 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.081874 seconds # Output file: benchmark_spgemm.out.7 -Device Parcsr Matrix-by-Matrix wall clock time = 0.004192 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.004065 seconds # Output file: benchmark_spgemm.out.8 -Device Parcsr Matrix-by-Matrix wall clock time = 0.006853 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.006895 seconds # Output file: benchmark_spgemm.out.9 -Device Parcsr Matrix-by-Matrix wall clock time = 0.039470 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.039068 seconds # Output file: benchmark_spgemm.out.10 -Device Parcsr Matrix-by-Matrix wall clock time = 0.115897 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.116622 seconds # Output file: benchmark_spgemm.out.11 -Device Parcsr Matrix-by-Matrix wall clock time = 0.016865 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.016892 seconds # Output file: benchmark_spgemm.out.12 -Device Parcsr Matrix-by-Matrix wall clock time = 0.020121 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.019927 seconds # Output file: benchmark_spgemm.out.13 -Device Parcsr Matrix-by-Matrix wall clock time = 0.009073 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.009113 seconds # Output file: benchmark_spgemm.out.14 -Device Parcsr Matrix-by-Matrix wall clock time = 0.131827 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.132173 seconds # Output file: benchmark_spgemm.out.15 -Device Parcsr Matrix-by-Matrix wall clock time = 0.002368 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.002346 seconds # Output file: benchmark_spgemm.out.16 -Device Parcsr Matrix-by-Matrix wall clock time = 0.008839 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.008833 seconds # Output file: benchmark_spgemm.out.17 -Device Parcsr Matrix-by-Matrix wall clock time = 0.002929 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.002913 seconds # Output file: benchmark_spgemm.out.18 -Device Parcsr Matrix-by-Matrix wall clock time = 0.005855 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.005856 seconds # Output file: benchmark_spgemm.out.19 -Device Parcsr Matrix-by-Matrix wall clock time = 0.004649 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.002930 seconds # Output file: benchmark_spgemm.out.20 -Device Parcsr Matrix-by-Matrix wall clock time = 0.002893 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.003172 seconds # Output file: benchmark_spgemm.out.21 -Device Parcsr Matrix-by-Matrix wall clock time = 0.033856 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.034395 seconds # Output file: benchmark_spgemm.out.22 -Device Parcsr Matrix-by-Matrix wall clock time = 0.115937 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.116523 seconds # Output file: benchmark_spgemm.out.23 -Device Parcsr Matrix-by-Matrix wall clock time = 0.014854 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.014950 seconds # Output file: benchmark_spgemm.out.24 -Device Parcsr Matrix-by-Matrix wall clock time = 0.021270 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.017896 seconds # Output file: benchmark_spgemm.out.25 -Device Parcsr Matrix-by-Matrix wall clock time = 0.059886 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.060106 seconds # Output file: benchmark_spgemm.out.26 -Device Parcsr Matrix-by-Matrix wall clock time = 0.588026 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.588868 seconds # Output file: benchmark_spgemm.out.27 -Device Parcsr Matrix-by-Matrix wall clock time = 0.026862 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.024888 seconds # Output file: benchmark_spgemm.out.28 -Device Parcsr Matrix-by-Matrix wall clock time = 0.032868 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.030994 seconds # Output file: benchmark_spgemm.out.29 -Device Parcsr Matrix-by-Matrix wall clock time = 0.065964 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.060870 seconds # Output file: benchmark_spgemm.out.30 -Device Parcsr Matrix-by-Matrix wall clock time = 0.081763 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.062836 seconds # Output file: benchmark_spgemm.out.31 -Device Parcsr Matrix-by-Matrix wall clock time = 0.022984 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.022955 seconds # Output file: benchmark_spgemm.out.32 -Device Parcsr Matrix-by-Matrix wall clock time = 0.021927 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.021522 seconds # Output file: benchmark_spgemm.out.33 -Device Parcsr Matrix-by-Matrix wall clock time = 0.067044 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.069183 seconds # Output file: benchmark_spgemm.out.34 -Device Parcsr Matrix-by-Matrix wall clock time = 0.071896 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.071740 seconds # Output file: benchmark_spgemm.out.35 -Device Parcsr Matrix-by-Matrix wall clock time = 0.029163 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.026649 seconds # Output file: benchmark_spgemm.out.36 -Device Parcsr Matrix-by-Matrix wall clock time = 0.028319 seconds +Device Parcsr Matrix-by-Matrix wall clock time = 0.024854 seconds diff --git a/src/test/TEST_bench/benchmark_struct.perf.saved.tioga b/src/test/TEST_bench/benchmark_struct.perf.saved.tioga index 740451765..9f6abb8f5 100644 --- a/src/test/TEST_bench/benchmark_struct.perf.saved.tioga +++ b/src/test/TEST_bench/benchmark_struct.perf.saved.tioga @@ -1,24 +1,24 @@ # Output file: benchmark_struct.out.1 -PCG Setup wall clock time = 0.901832 seconds -PCG Solve wall clock time = 0.700184 seconds +PCG Setup wall clock time = 0.120527 seconds +PCG Solve wall clock time = 0.524466 seconds # Output file: benchmark_struct.out.2 -PCG Setup wall clock time = 1.046741 seconds -PCG Solve wall clock time = 0.867303 seconds +PCG Setup wall clock time = 0.153928 seconds +PCG Solve wall clock time = 0.703765 seconds # Output file: benchmark_struct.out.3 -PCG Setup wall clock time = 8.931370 seconds -PCG Solve wall clock time = 6.930335 seconds +PCG Setup wall clock time = 1.099122 seconds +PCG Solve wall clock time = 5.576967 seconds # Output file: benchmark_struct.out.4 -PCG Setup wall clock time = 12.306432 seconds -PCG Solve wall clock time = 11.384939 seconds +PCG Setup wall clock time = 1.768987 seconds +PCG Solve wall clock time = 9.280957 seconds # Output file: benchmark_struct.out.5 -PCG Setup wall clock time = 0.174218 seconds -PCG Solve wall clock time = 0.190781 seconds +PCG Setup wall clock time = 0.013666 seconds +PCG Solve wall clock time = 0.106329 seconds # Output file: benchmark_struct.out.6 -PCG Setup wall clock time = 0.200355 seconds -PCG Solve wall clock time = 0.377829 seconds +PCG Setup wall clock time = 0.020694 seconds +PCG Solve wall clock time = 0.259598 seconds # Output file: benchmark_struct.out.7 -PCG Setup wall clock time = 0.250921 seconds -PCG Solve wall clock time = 0.444261 seconds +PCG Setup wall clock time = 0.056668 seconds +PCG Solve wall clock time = 0.355209 seconds # Output file: benchmark_struct.out.8 -PCG Setup wall clock time = 0.262308 seconds -PCG Solve wall clock time = 0.687628 seconds +PCG Setup wall clock time = 0.074631 seconds +PCG Solve wall clock time = 0.548870 seconds