[PATCH] D140208: [AMDGPU] Improved wide multiplies

[Thomas Symalla via Phabricator via llvm-commits]

tsymalla added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:2930
 
+  auto length = Src0.size();
+  SmallVector<KnownBits, 2> Src0KB, Src1KB;
----------------
Can inline this into the loop


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:3049
+            if (!AtLeastOneArgIsZero) {
+              auto Mad = B.buildInstr(AMDGPU::G_AMDGPU_MAD_U64_U32, {S64, S1},
+                                      {Src0[j0], Src1[j1], Tmp});
----------------
Intendation issue?


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:3182
 
+  const bool Src0IsZero = Helper.getKnownBits()->getKnownBits(Src0).isZero();
+  const bool Src1IsZero = Helper.getKnownBits()->getKnownBits(Src1).isZero();
----------------
Naming is a bit unfortunate, as the function returning the pointer to the KBAnalysis is named the same as the function getting the known bits.


Repository:
  rG LLVM Github Monorepo

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

https://reviews.llvm.org/D140208