[mlir][gpu] Add support for lowering math.erf to __nv_erf

[majnemer]

No description provided.

[llvmbot]

@llvm/pr-subscribers-mlir

@llvm/pr-subscribers-mlir-gpu

Author: David Majnemer (majnemer)

Changes

---

Full diff: https://github.com/llvm/llvm-project/pull/79848.diff

1 Files Affected:

• (modified) mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp (+1)

diff --git a/mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp b/mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp
index a7ac2332961ae21..d6a5d8cd74d5f20 100644
--- a/mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp
+++ b/mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp
@@ -367,6 +367,7 @@ void mlir::populateGpuToNVVMConversionPatterns(LLVMTypeConverter &converter,
   populateOpPatterns<math::CeilOp>(converter, patterns, "__nv_ceilf",
                                    "__nv_ceil");
   populateOpPatterns<math::CosOp>(converter, patterns, "__nv_cosf", "__nv_cos");
+  populateOpPatterns<math::ErfOp>(converter, patterns, "__nv_erff", "__nv_erf");
   populateOpPatterns<math::ExpOp>(converter, patterns, "__nv_expf", "__nv_exp");
   populateOpPatterns<math::Exp2Op>(converter, patterns, "__nv_exp2f",
                                    "__nv_exp2");

[cheshire]

What about tests?

[majnemer]

What about tests?

Done.