AMDGPU: Add statistic for number of MFMAs moved to AGPR form

[arsenm]

No description provided.

[arsenm]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• AMDGPU: Add tests for every mfma intrinsic v-to-a mapping #153026
• AMDGPU: Add test for mfma rewrite pass respecting optnone #153025
• AMDGPU: Add statistic for number of MFMAs moved to AGPR form #153024 👈 (View in Graphite)
• AMDGPU: Handle rewriting VGPR MFMA fed from AGPR copy #153022 : 2 other dependent PRs (#153023 , #154322 )
• AMDGPU: Add baseline test for mfma rewrite with phi #153021
• AMDGPU: Add baseline test for vgpr mfma with copied-from AGPR #153020
• AMDGPU: Handle rewriting VGPR MFMA to AGPR with subregister copies #153019
• AMDGPU: Add some baseline test for mfma rewrite with subregister copies #153018
• AMDGPU: Handle rewriting VGPR MFMAs with immediate src2 #153016
• AMDGPU: Handle rewriting non-tied MFMA to AGPR form #153015
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[llvmbot]

@llvm/pr-subscribers-backend-amdgpu

Author: Matt Arsenault (arsenm)

Changes

---

Full diff: https://github.com/llvm/llvm-project/pull/153024.diff

1 Files Affected:

• (modified) llvm/lib/Target/AMDGPU/AMDGPURewriteAGPRCopyMFMA.cpp (+5)

diff --git a/llvm/lib/Target/AMDGPU/AMDGPURewriteAGPRCopyMFMA.cpp b/llvm/lib/Target/AMDGPU/AMDGPURewriteAGPRCopyMFMA.cpp
index b71c70db5e6b3..c4a370225b606 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPURewriteAGPRCopyMFMA.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPURewriteAGPRCopyMFMA.cpp
@@ -26,6 +26,7 @@
 #include "GCNSubtarget.h"
 #include "SIMachineFunctionInfo.h"
 #include "SIRegisterInfo.h"
+#include "llvm/ADT/Statistic.h"
 #include "llvm/CodeGen/LiveIntervals.h"
 #include "llvm/CodeGen/LiveRegMatrix.h"
 #include "llvm/CodeGen/MachineFunctionPass.h"
@@ -38,6 +39,9 @@ using namespace llvm;
 
 namespace {
 
+STATISTIC(NumMFMAsRewrittenToAGPR,
+          "Number of MFMA instructions rewritten to use AGPR form");
+
 class AMDGPURewriteAGPRCopyMFMAImpl {
   MachineFunction &MF;
   const GCNSubtarget &ST;
@@ -261,6 +265,7 @@ bool AMDGPURewriteAGPRCopyMFMAImpl::tryReassigningMFMAChain(
     int NewMFMAOp =
         AMDGPU::getMFMASrcCVDstAGPROp(RewriteCandidate->getOpcode());
     RewriteCandidate->setDesc(TII.get(NewMFMAOp));
+    ++NumMFMAsRewrittenToAGPR;
   }
 
   return true;