projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 9f4f012) | patch
[X86] AMD Zen 3: cap LoopMicroOpBufferSize to workaround PR50384 (quadratic IndVars...
authorRoman Lebedev <lebedev.ri@gmail.com>
Tue, 18 May 2021 12:55:02 +0000 (15:55 +0300)
committerRoman Lebedev <lebedev.ri@gmail.com>
Tue, 18 May 2021 12:56:57 +0000 (15:56 +0300)
commit3cc39607668d6f942458a37e7a9111294d9373bd
treee33f7f24fb56f774e484668a486171dbf0c94c0atree | snapshot
parent9f4f012c10266c2117f80b7c7eb699aadef0589dcommit | diff
[X86] AMD Zen 3: cap LoopMicroOpBufferSize to workaround PR50384 (quadratic IndVars runtime)

While i would like to keep the right value here,
i would also like to be able to actually compile
e.g. vanilla test-suite.

256 is a pretty random guess, it should be pretty good enough
for serious loops, but small enough to result in tolerant
compile times for certain edge cases.

https://bugs.llvm.org/show_bug.cgi?id=50384
llvm/lib/Target/X86/X86ScheduleZnver3.td diff | blob | history
Domain: System / Toolchain;