[PATCH] D107187: [amdgpu] Add an enhanced conversion from i64 to f32.

[Michael Liao via Phabricator via llvm-commits]

hliao added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:2106
+  } else {
+    auto LZ = B.buildCTLZ_ZERO_UNDEF(S32, Unmerge.getReg(1));
+    auto NotAllZeros =
----------------
foad wrote:
> Can't you buildCTLZ here and let the legalizer worry about how to handle the ZERO case? In particular it would be better to do v_ffbh followed by v_min_u32 with 32, instead of a compare and select.
I reused part of the original code and didn't try whether we support CLTZ without undefined zero behavior. We may try that later.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D107187/new/

https://reviews.llvm.org/D107187