[Bug tree-optimization/79534] [7 Regression] tree-ifcombine aarch64 performance regression with trunk@245151

2017-04-20 Thread jgreenhalgh at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79534 --- Comment #10 from James Greenhalgh --- The most striking improvement was in libquantum, for which we saw a 15% performance improvement on Cortex-A72 (3% on cortex-A57) directly attributable to basic block ordering after this patch.

[Bug tree-optimization/79534] [7 Regression] tree-ifcombine aarch64 performance regression with trunk@245151

2017-04-19 Thread brzycki at gmail dot com
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79534 --- Comment #9 from Brian Rzycki --- Hello James, If this is working as designed this may just be a case of having to live with the regression. Do we know if anyone analyzed the net-benefit of Honza's patch? If more benchmarks/tests win then I

[Bug tree-optimization/79534] [7 Regression] tree-ifcombine aarch64 performance regression with trunk@245151

2017-04-19 Thread jgreenhalgh at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79534 --- Comment #8 from James Greenhalgh --- In the case before Honza's patch, corrupt profile information leads to a branch being marked as 100% taken. After Honza's patch, the branch is instead seen with 95.6% taken: (jump_insn 1916 1915 1922 309

[Bug tree-optimization/79534] [7 Regression] tree-ifcombine aarch64 performance regression with trunk@245151

2017-03-31 Thread law at redhat dot com
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79534 Jeffrey A. Law changed: What|Removed |Added Priority|P3 |P2 CC|

[Bug tree-optimization/79534] [7 Regression] tree-ifcombine aarch64 performance regression with trunk@245151

2017-03-30 Thread jgreenhalgh at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79534 --- Comment #7 from James Greenhalgh --- I'm not sure there are any bugs here to fix, though I can still reproduce the performance differences. First up, basic block reordering causes an issue across all microarchitectures on which I've looked

[Bug tree-optimization/79534] [7 Regression] tree-ifcombine aarch64 performance regression with trunk@245151

2017-03-29 Thread brzycki at gmail dot com
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79534 --- Comment #6 from Brian Rzycki --- James, my apologies if it wasn't clear enough what the compile options were. The test platform in this case is a Juno A57 running Ubuntu. I actually never turned off -mcpu=cortex-a57 during my testing. I'll

[Bug tree-optimization/79534] [7 Regression] tree-ifcombine aarch64 performance regression with trunk@245151

2017-03-29 Thread jgreenhalgh at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79534 James Greenhalgh changed: What|Removed |Added Status|UNCONFIRMED |NEW Last reconfirmed|

[Bug tree-optimization/79534] [7 Regression] tree-ifcombine aarch64 performance regression with trunk@245151

2017-03-28 Thread brzycki at gmail dot com
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79534 --- Comment #4 from Brian Rzycki --- Hi James, thank you for taking the time to test this. Unfortunately I can't post the original code because of its license. The problem is with the weighting changing enough that at least one pair of if

[Bug tree-optimization/79534] [7 Regression] tree-ifcombine aarch64 performance regression with trunk@245151

2017-03-28 Thread jgreenhalgh at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79534 James Greenhalgh changed: What|Removed |Added CC||jgreenhalgh at gcc dot gnu.org ---