[Mlir-commits] [mlir] [mlir][nvgpu] Fix `transposeB` in `nvgpu.warpgroup.mma` (PR #79271)

[Guray Ozen]

https://github.com/grypp created https://github.com/llvm/llvm-project/pull/79271

The #76150 fixed meaning of `transposeB` in NVVM dialect which was initially implemented with opposite meaning.

This PR fixes the lowering of `nvgpu.warpgroup.mma` to NVVM dialect.

This will fix two integration tests:
gemm_f32_f16_f16_128x128x128.mlir
gemm_pred_f32_f16_f16_128x128x128.mlir

>From 19865938501e42493ae8efbe11ef1ceba872c4b4 Mon Sep 17 00:00:00 2001
From: Guray Ozen <guray.ozen at gmail.com>
Date: Wed, 24 Jan 2024 11:44:28 +0100
Subject: [PATCH] [mlir][nvgpu] Fix `transposeB` in `nvgpu.warpgroup.mma`

The #76150 fixed meaning of `transposeB` in NVVM dialect which was initially implemented with opposite meaning.

This PR fixes the lowering of `nvgpu.warpgroup.mma` to NVVM dialect.

This will fix two integration tests:
gemm_f32_f16_f16_128x128x128.mlir
gemm_pred_f32_f16_f16_128x128x128.mlir
---
 mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp b/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp
index 43d05b872a4fbc8..5080956a4589828 100644
--- a/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp
+++ b/mlir/lib/Conversion/NVGPUToNVVM/NVGPUToNVVM.cpp
@@ -1407,7 +1407,7 @@ struct NVGPUWarpgroupMmaOpLowering
       NVVM::WGMMAScaleOutAttr scaleOut = generateScaleOut();
       NVVM::WGMMAScaleInAttr scaleIn = generateScaleIn();
       NVVM::MMALayoutAttr layoutA = generateWgmmaLayout(op.getTransposeA());
-      NVVM::MMALayoutAttr layoutB = generateWgmmaLayout(op.getTransposeB());
+      NVVM::MMALayoutAttr layoutB = generateWgmmaLayout(!op.getTransposeB());
 
       auto overflow = NVVM::MMAIntOverflowAttr::get(
           op->getContext(), NVVM::MMAIntOverflow::wrapped);