[PATCH] D86079: [X86] Improved lowering for saturating float to int.

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: llvm/lib/Target/X86/X86ISelLowering.cpp:281
 
+  if (Subtarget.hasSSE1() || Subtarget.hasSSE2()) {
+    // Custom lowering for saturating float to int conversions.
----------------
hasSSE2() implies hasSSE1() so you don't need to check both.

If you want this to work for f32 SSE1 only targets. You'll need to explicitly check hasSSE2 in the Lower function. The default CPUs on Linux, Windows, and Mac all have SSE2 enabled. So I don't know if you want to support SSE1 only or not. Our test coverage for SSE1 only isn't great.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D86079/new/

https://reviews.llvm.org/D86079