From 3f8d0e1ef8d6d86035f972c49c9cee79a3174802 Mon Sep 17 00:00:00 2001
From: Matthew Poremba <matthew.poremba@amd.com>
Date: Mon, 18 Mar 2024 19:02:10 -0500
Subject: [PATCH] arch-vega: Fix V_FMAC_F32 data type

The datatype is U32 but should be F32. This is causing an implicit cast
leading to incorrect results. This fixes nn.Dropout in PyTorch.

Change-Id: I546aa917fde1fd6bc832d9d0fa9ffe66505e87dd
---
 src/arch/amdgpu/vega/insts/vop2.cc | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/src/arch/amdgpu/vega/insts/vop2.cc b/src/arch/amdgpu/vega/insts/vop2.cc
index ddd77e27da..2672063d0c 100644
--- a/src/arch/amdgpu/vega/insts/vop2.cc
+++ b/src/arch/amdgpu/vega/insts/vop2.cc
@@ -2167,9 +2167,9 @@ namespace VegaISA
     Inst_VOP2__V_FMAC_F32::execute(GPUDynInstPtr gpuDynInst)
     {
         Wavefront *wf = gpuDynInst->wavefront();
-        ConstVecOperandU32 src0(gpuDynInst, instData.SRC0);
-        ConstVecOperandU32 src1(gpuDynInst, instData.VSRC1);
-        VecOperandU32 vdst(gpuDynInst, instData.VDST);
+        ConstVecOperandF32 src0(gpuDynInst, instData.SRC0);
+        ConstVecOperandF32 src1(gpuDynInst, instData.VSRC1);
+        VecOperandF32 vdst(gpuDynInst, instData.VDST);
 
         src0.readSrc();
         src1.read();