[PATCH] D123878: [AMDGPU] Add remarks to output some resource usage

Matt Arsenault via Phabricator via cfe-commits cfe-commits at lists.llvm.org
Fri Apr 29 14:45:45 PDT 2022


arsenm added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp:1222
+                                             &MF.front())
+           << "ScratchSize [bytes/thread]: "
+           << ore::NV("ScratchSize", CurrentProgramInfo.ScratchSize);
----------------
Probably should not use the word thread. lane?


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp:1243
+  ORE->emit([&]() {
+    return MachineOptimizationRemarkAnalysis(Name, "VGPRSpill",
+                                             MF.getFunction().getSubprogram(),
----------------
Unrelated but we should reimplement VGPRSpill. It's now reporting number of lowered spill instructions rather than number of spilled values


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D123878/new/

https://reviews.llvm.org/D123878



More information about the cfe-commits mailing list