[PATCH] D80032: [AMDGPU] Always expand ext/insertelement with divergent idx

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:102-106
+static cl::opt<bool> ExpandDynVectors(
+  "amdgpu-expand-dynamic-vectors",
+  cl::Hidden,
+  cl::desc("Expand extract and insert vector elements with cmd/cndmask"),
+  cl::init(true));
----------------
I would invert this and rename it. How about -amdgpu-use-divergent-register-indexing, default false?


================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:9549
   // instructions.
-  if (VecSize <= 256 && (VecSize > 64 || EltSize >= 32) &&
+  // Always do this if var-idx is divergent, otherwise it will become a loop.
+  if (ExpandDynVectors &&
----------------
GlobalISel needs the compare and select path implemented


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D80032/new/

https://reviews.llvm.org/D80032