projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 5c6b3fb) | patch
[X86][AVX512DQ] Use packed instructions for scalar FP<->i64 conversions on 32-bit...
authorCraig Topper <craig.topper@intel.com>
Wed, 16 May 2018 17:40:07 +0000 (17:40 +0000)
committerCraig Topper <craig.topper@intel.com>
Wed, 16 May 2018 17:40:07 +0000 (17:40 +0000)
commit67aa726f8c521200151b2bd3c8eb897bc4c739f0
tree7e11185dddcce9cb2b6e66108c7d85119c66a657tree | snapshot
parent5c6b3fb8fbe414bb930fbdf00dde70510fcef488commit | diff
[X86][AVX512DQ] Use packed instructions for scalar FP<->i64 conversions on 32-bit targets

As i64 types are not legal on 32-bit targets, insert these into a suitable zero vector and use the packed vXi64<->FP conversion instructions instead.

Fixes PR3163.

Differential Revision: https://reviews.llvm.org/D43441

llvm-svn: 332498
llvm/lib/Target/X86/X86ISelLowering.cpp diff | blob | history
llvm/test/CodeGen/X86/avx512-regcall-NoMask.ll diff | blob | history
llvm/test/CodeGen/X86/scalar-fp-to-i64.ll diff | blob | history
llvm/test/CodeGen/X86/scalar-int-to-fp.ll diff | blob | history
Domain: System / Toolchain;