Bisect

iree-org · Feb 5, 2025 · b6a41ab · b6a41ab
1 parent 5368df4
commit b6a41ab
Show file tree

Hide file tree

Showing 3 changed files with 5 additions and 4 deletions.
diff --git a/compiler/src/iree/compiler/Codegen/LLVMGPU/CMakeLists.txt b/compiler/src/iree/compiler/Codegen/LLVMGPU/CMakeLists.txt
@@ -139,6 +139,7 @@ iree_cc_library(
     MLIRLinalgTransforms
     MLIRMathDialect
     MLIRMathToLLVM
+    MLIRMathTransforms
     MLIRMemRefDialect
     MLIRMemRefToLLVM
     MLIRMemRefTransforms

diff --git a/compiler/src/iree/compiler/Codegen/LLVMGPU/ConvertToLLVM.cpp b/compiler/src/iree/compiler/Codegen/LLVMGPU/ConvertToLLVM.cpp
@@ -566,7 +566,8 @@ void populateScalarizeMathOps(RewritePatternSet &patterns) {
                ScalarizeMathOp<math::Log10Op>, ScalarizeMathOp<math::Log2Op>,
                ScalarizeMathOp<math::PowFOp>, ScalarizeMathOp<math::RsqrtOp>,
                ScalarizeMathOp<math::SinOp>, ScalarizeMathOp<math::SqrtOp>,
-               ScalarizeMathOp<math::TanhOp>>(patterns.getContext());
+               ScalarizeMathOp<math::TanhOp>, ScalarizeMathOp<math::ErfOp>>(
+      patterns.getContext());
 }
 
 void populateConvertSharedMemoryAllocOps(RewritePatternSet &patterns) {

diff --git a/compiler/src/iree/compiler/Codegen/LLVMGPU/Passes.cpp b/compiler/src/iree/compiler/Codegen/LLVMGPU/Passes.cpp
@@ -1110,10 +1110,9 @@ static void addLowerToLLVMGPUPasses(OpPassManager &modulePassManager,
       .addPass(createConvertBf16ArithToF32Pass)
       .addPass(createConvertBf16ToUInt16BuffersPass)
       // Convert math dialect elementry functions to polynomial form.
-      .addPass([&]() {
+      .addPass([]() {
         return createPolynomialApproximationPass(
-            {/*noApproxOps =*/{"tan", "sinh", "cosh", "asinh", "acosh", "atanh",
-                               "powf", "fpowi"}});
+            PolynomialApproximationPassOptions{/*noApproxOps =*/{"tan"}});
       })
       .addPass(memref::createExpandOpsPass)
       .addPass(memref::createFoldMemRefAliasOpsPass)