[PATCH] D150347: [AMDGPU] Improve abs modifier usage

[Jay Foad via Phabricator via llvm-commits]

foad added a comment.

LGTM, thanks, but please try to simplify the code - I've added some ideas inline.



================
Comment at: llvm/lib/CodeGen/CodeGenPrepare.cpp:7122
       // Now we clone an instruction, its operands' defs may sink to this BB
-      // now. So we put the operands defs' BBs into FreshBBs to do optmization.
+      // now. So we put the operands defs' BBs into FreshBBs to do optimization.
       for (unsigned I = 0; I < NI->getNumOperands(); ++I) {
----------------
Please commit this separately (no review required!)


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp:5151
+
+    if (IntrinsicInst *UI = dyn_cast<IntrinsicInst>(Op)) {
+      switch (UI->getIntrinsicID()) {
----------------
Nit: if you're using PatternMatch already, you can use `m_FAbs` here.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp:5166
+
+  return ShouldSink;
+}
----------------
Nit: might be simpler to return `!Ops.empty()`?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D150347/new/

https://reviews.llvm.org/D150347