[llvm] [WIP][AMDGPU] Enable "amdgpu-uniform-intrinsic-combine" pass in pipeline. (PR #128687)

[via llvm-commits]

github-actions[bot] wrote:

<!--LLVM CODE FORMAT COMMENT: {clang-format}-->


:warning: C/C++ code formatter, clang-format found issues in your code. :warning:

<details>
<summary>
You can test this locally with the following command:
</summary>

``````````bash
git-clang-format --diff origin/main HEAD --extensions cpp,h -- llvm/lib/Target/AMDGPU/AMDGPU.h llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp llvm/lib/Target/AMDGPU/AMDGPUUniformIntrinsicCombine.cpp
``````````

:warning:
The reproduction instructions above might return results for more than one PR
in a stack if you are using a stacked PR workflow. You can limit the results by
changing `origin/main` to the base branch/commit you want to compare against.
:warning:

</details>

<details>
<summary>
View the diff from clang-format here.
</summary>

``````````diff
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUUniformIntrinsicCombine.cpp b/llvm/lib/Target/AMDGPU/AMDGPUUniformIntrinsicCombine.cpp
index cdd6df940..45466542b 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUUniformIntrinsicCombine.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUUniformIntrinsicCombine.cpp
@@ -99,48 +99,48 @@ static bool optimizeUniformIntrinsic(IntrinsicInst &II,
       return false;
     LLVM_DEBUG(dbgs() << "Found uniform ballot intrinsic: " << II << '\n');
 
-        ICmpInst::Predicate Pred = ICmp->getPredicate();
-        Value *OtherOp = Op0 == &II ? Op1 : Op0;
+    ICmpInst::Predicate Pred = ICmp->getPredicate();
+    Value *OtherOp = Op0 == &II ? Op1 : Op0;
 
-        if (Pred == ICmpInst::ICMP_EQ && match(OtherOp, m_Zero())) {
+    if (Pred == ICmpInst::ICMP_EQ && match(OtherOp, m_Zero())) {
 <<<<<<< HEAD
 <<<<<<< HEAD
-          // Case: (icmp eq %ballot, 0) -> xor %ballot_arg, 1
-          Instruction *NotOp =
-              BinaryOperator::CreateNot(Src, "", ICmp->getIterator());
-          Tracker[NotOp] = true; // NOT preserves uniformity
+      // Case: (icmp eq %ballot, 0) -> xor %ballot_arg, 1
+      Instruction *NotOp =
+          BinaryOperator::CreateNot(Src, "", ICmp->getIterator());
+      Tracker[NotOp] = true; // NOT preserves uniformity
 =======
-          // Case (icmp eq %ballot, 0) -->  xor %ballot_arg, 1
-          Instruction *NotOp =
-              BinaryOperator::CreateNot(Src, "", ICmp->getIterator());
+      // Case (icmp eq %ballot, 0) -->  xor %ballot_arg, 1
+      Instruction *NotOp =
+          BinaryOperator::CreateNot(Src, "", ICmp->getIterator());
 >>>>>>> ee370a5a77ca ([AMDGPU] combine uniform AMDGPU lane Intrinsics)
 =======
-          // Case: (icmp eq %ballot, 0) -> xor %ballot_arg, 1
-          Instruction *NotOp =
-              BinaryOperator::CreateNot(Src, "", ICmp->getIterator());
-          // Record uniformity: Src is uniform, and NOT preserves uniformity.
-          NewUMap[NotOp] = true;
+      // Case: (icmp eq %ballot, 0) -> xor %ballot_arg, 1
+      Instruction *NotOp =
+          BinaryOperator::CreateNot(Src, "", ICmp->getIterator());
+      // Record uniformity: Src is uniform, and NOT preserves uniformity.
+      NewUMap[NotOp] = true;
 >>>>>>> f6cc7aa1522d (store newly inserted inst and its uniformity)
-          LLVM_DEBUG(dbgs() << "Replacing ICMP_EQ: " << *NotOp << '\n');
-          ICmp->replaceAllUsesWith(NotOp);
-          ICmp->eraseFromParent();
-          Changed = true;
-        } else if (Pred == ICmpInst::ICMP_NE && match(OtherOp, m_Zero())) {
+      LLVM_DEBUG(dbgs() << "Replacing ICMP_EQ: " << *NotOp << '\n');
+      ICmp->replaceAllUsesWith(NotOp);
+      ICmp->eraseFromParent();
+      Changed = true;
+    } else if (Pred == ICmpInst::ICMP_NE && match(OtherOp, m_Zero())) {
 <<<<<<< HEAD
 <<<<<<< HEAD
-          // Case: (icmp ne %ballot, 0) -> %ballot_arg
+      // Case: (icmp ne %ballot, 0) -> %ballot_arg
 =======
-          // (icmp ne %ballot, 0)  -->  %ballot_arg
+      // (icmp ne %ballot, 0)  -->  %ballot_arg
 >>>>>>> ee370a5a77ca ([AMDGPU] combine uniform AMDGPU lane Intrinsics)
 =======
-          // Case: (icmp ne %ballot, 0) -> %ballot_arg
+      // Case: (icmp ne %ballot, 0) -> %ballot_arg
 >>>>>>> f6cc7aa1522d (store newly inserted inst and its uniformity)
-          LLVM_DEBUG(dbgs() << "Replacing ICMP_NE with ballot argument: "
-                            << *Src << '\n');
-          ICmp->replaceAllUsesWith(Src);
-          ICmp->eraseFromParent();
-          Changed = true;
-        }
+      LLVM_DEBUG(dbgs() << "Replacing ICMP_NE with ballot argument: " << *Src
+                        << '\n');
+      ICmp->replaceAllUsesWith(Src);
+      ICmp->eraseFromParent();
+      Changed = true;
+    }
       }
     }
     // Erase the intrinsic if it has no remaining uses.
@@ -204,7 +204,7 @@ static bool runUniformIntrinsicCombine(Module &M, ModuleAnalysisManager &AM) {
       IsChanged |= optimizeUniformIntrinsic(*II, UI);
 >>>>>>> ee370a5a77ca ([AMDGPU] combine uniform AMDGPU lane Intrinsics)
 =======
-      IsChanged |= optimizeUniformIntrinsic(*II, UI, NewUMap);
+    IsChanged |= optimizeUniformIntrinsic(*II, UI, NewUMap);
 >>>>>>> f6cc7aa1522d (store newly inserted inst and its uniformity)
     }
   }

``````````

</details>


https://github.com/llvm/llvm-project/pull/128687