https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/ en-us PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench Changelog PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench Changelog chgd _brch__perf_tuning_paper back to _brch__default.. in default branch https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/rev/efc1ee85f1ee Sean Halle <seanhalle@yahoo.com> Sun, 09 Sep 2012 20:50:02 -0700 Added tag lots of cores version for changeset 0d7177551bff https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/rev/3477d7443620 Nina Engelhardt <nengel@mailbox.tu-berlin.de> Wed, 18 Apr 2012 15:51:01 +0200 optimization for large number of cores: dedicate core 0 to create/receive https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/rev/0d7177551bff Nina Engelhardt <nengel@mailbox.tu-berlin.de> Wed, 18 Apr 2012 15:50:14 +0200 Merge with perf_tuning_paper branch, which grabs fixes and best performing version https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/rev/8972c00c00dd Sean Halle <seanhalle@yahoo.com> Sun, 15 Jul 2012 01:27:39 -0700 slicing factor 0.5 actually gives better results in this ver https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/rev/c35cb1f48f89 Nina Engelhardt <nengel@mailbox.tu-berlin.de> Tue, 17 Apr 2012 20:13:26 +0200 slicing factor 0.4, which gives good results for dedicated ver, gives bad result here https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/rev/e348e7f58034 Nina Engelhardt <nengel@mailbox.tu-berlin.de> Tue, 17 Apr 2012 19:58:33 +0200 try to adjust slicing, pick too small task size (creation becomes bottleneck) https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/rev/30910c0b615a Nina Engelhardt <nengel@mailbox.tu-berlin.de> Tue, 17 Apr 2012 19:52:50 +0200 move receivePr to core 0 https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/rev/1d2e1cc75356 Nina Engelhardt <nengel@mailbox.tu-berlin.de> Tue, 17 Apr 2012 19:46:38 +0200 Added tag portable version for changeset b7d8cf6680a9 https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/rev/1f1603c15b61 Nina Engelhardt <nengel@mailbox.tu-berlin.de> Tue, 17 Apr 2012 19:45:47 +0200 portable ver: start division at core 1 https://hg.opensourceresearchinstitute.org/cgi-bin/hgwebdir.cgi/PR/Applications/SSR/SSR__Blocked_Matrix_Mult__Bench/rev/b7d8cf6680a9 Nina Engelhardt <nengel@mailbox.tu-berlin.de> Tue, 17 Apr 2012 19:39:18 +0200