From fb9f77575a08918a682d7eb0781c6e5f1d4f0bf3 Mon Sep 17 00:00:00 2001
From: Tailing Yuan <yuantailing@gmail.com>
Date: Sun, 26 May 2024 18:30:30 +0800
Subject: [PATCH] Fixing correctness of bf16TensorCoreGemm

---
 .../3_CUDA_Features/bf16TensorCoreGemm/bf16TensorCoreGemm.cu  | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/Samples/3_CUDA_Features/bf16TensorCoreGemm/bf16TensorCoreGemm.cu b/Samples/3_CUDA_Features/bf16TensorCoreGemm/bf16TensorCoreGemm.cu
index a89e4096..dc42ecbb 100644
--- a/Samples/3_CUDA_Features/bf16TensorCoreGemm/bf16TensorCoreGemm.cu
+++ b/Samples/3_CUDA_Features/bf16TensorCoreGemm/bf16TensorCoreGemm.cu
@@ -592,8 +592,8 @@ __global__ void simple_wmma_bf16gemm(__nv_bfloat16 *a, __nv_bfloat16 *b, float *
       int aCol = i; 
       int aRow = warpM * M;
 
-      int bCol = i;
-      int bRow = warpN * N;
+      int bCol = warpN * N;
+      int bRow = i;
 
       // Bounds checking
       if (aRow < m_ld && aCol < k_ld && bRow < k_ld && bCol < n_ld) {