[Bug tree-optimization/91934] [8 Regression] Performance regression on 8.3.0 with -O3 and avx

2019-10-01 Thread tochansky at tochlab dot net
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91934 --- Comment #10 from Dmitrii Tochanskii --- Anyway thanks for your work. Now we know where problem is and users can make their own decision about patch. RedHat 8 uses gcc 8.2 but debian 10 - gcc 8.3...

[Bug tree-optimization/91934] Performance regression on 8.3.0 with -O3 and avx

2019-09-30 Thread tochansky at tochlab dot net
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91934 --- Comment #4 from Dmitrii Tochanskii --- I'm not a good specialist in avx, so I just see something like loop unroll or may be very log data preparation. For example: = vmovups ymm3, YMMWORD PTR [r8+r9] vmovups ymm5, YMM

[Bug tree-optimization/91934] Performance regression on 8.3.0 with -O3 and avx

2019-09-30 Thread tochansky at tochlab dot net
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91934 --- Comment #3 from Dmitrii Tochanskii --- Yep, -fno-loop-unroll-and-jam helps me! Interesting.

[Bug c/91934] New: Performance regression on 8.3.0 with -O3 and avx

2019-09-30 Thread tochansky at tochlab dot net
: c Assignee: unassigned at gcc dot gnu.org Reporter: tochansky at tochlab dot net Target Milestone: --- Created attachment 46980 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=46980&action=edit Function where problem occurs GCC version is 8.3.0 building on