projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 535a566) | patch
[X86] Use a more realisitic cost for truncate v16i64->v16i8 with avx512f.
authorCraig Topper <craig.topper@intel.com>
Tue, 14 Apr 2020 03:43:43 +0000 (20:43 -0700)
committerCraig Topper <craig.topper@intel.com>
Tue, 14 Apr 2020 04:09:43 +0000 (21:09 -0700)
commit2f60fbce6cdac9ef72339dde32946b883b6a96a8
tree0806bd182d7580be553b5c845a843f5bfe93d975tree | snapshot
parent535a566a01e5aa3808de5af7e3d7fc2888eed0dacommit | diff
[X86] Use a more realisitic cost for truncate v16i64->v16i8 with avx512f.

Still not great and we could probably codegen this better, but
11 was clearly ridiculous.
llvm/lib/Target/X86/X86TargetTransformInfo.cpp diff | blob | history
llvm/test/Analysis/CostModel/X86/trunc.ll diff | blob | history
Domain: System / Toolchain;