[PATCH] D140208: [AMDGPU] Improved wide multiplies

[Thomas Symalla via Phabricator via llvm-commits]

tsymalla added a comment.

Can you re-apply clang-format on `AMDGPULegalizerInfo.cpp` please?



================
Comment at: llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:2924
+  SmallVector<bool, 2> Src0KnownZeros, Src1KnownZeros;
+  for (Register Src0Reg : Src0) {
+    Src0KnownZeros.push_back(KB.getKnownBits(Src0Reg).isZero());
----------------
I think you wanted to iterate with a classical for loop and access Src0 and Src via the subscript operator?


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:2998
             auto Mul = B.buildMul(S32, Src0[j0], Src1[j1]);
-            if (!LocalAccum[0]) {
+            if (!LocalAccum[0] || /*KB.getKnownBits(LocalAccum[0]).isZero()*/ LocalAccum[0] == 0) {
               LocalAccum[0] = Mul.getReg(0);
----------------
Is that check correct? Can you remove the comment please?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D140208/new/

https://reviews.llvm.org/D140208