[PATCH] D11316: [X86] -- Fix fptoui i64 conversions for IA32 (performance and correctness)

[Michael Kuperstein]

mkuper added inline comments.

================
Comment at: lib/Target/X86/X86ISelLowering.cpp:26320
@@ -26204,2 +26319,3 @@
 bool X86TargetLowering::isTargetFTOL() const {
-  return Subtarget->isTargetKnownWindowsMSVC() && !Subtarget->is64Bit();
+  // FTOL usage is currently incorrect, as it computes conversion
+  // to *signed* i64, so it is disabled here.  Given the presence
----------------
I'm ok with just deleting the FTOL code immediately, unless anyone objects.
(Can be a follow-up patch)

================
Comment at: test/CodeGen/X86/scalar-fp-to-i64.ll:17
@@ +16,3 @@
+;
+; RUN: llc < %s -mtriple=i386-pc-windows-msvc     -mcpu=skx      | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512_32
+; RUN: llc < %s -mtriple=i386-unknown-linux-gnu   -mcpu=skx      | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512_32
----------------
Can you use -mattr here instead of an explicit -mcpu?

================
Comment at: test/CodeGen/X86/win_ftol2.ll:4
@@ -3,3 +3,3 @@
 ; RUN: llc < %s -mtriple=i686-pc-linux | FileCheck %s -check-prefix=COMPILERRT
-; RUN: llc < %s -mtriple=x86_64-pc-win32 | FileCheck %s -check-prefix=COMPILERRT
+; R U N: llc < %s -mtriple=x86_64-pc-win32 | FileCheck %s -check-prefix=COMPILERRT
 ; RUN: llc < %s -mtriple=x86_64-pc-mingw32 | FileCheck %s -check-prefix=COMPILERRT
----------------
What happened to this RUN line?


http://reviews.llvm.org/D11316