From 8203533a8c9a1af5c273796a39de4a5a7f8e0626 Mon Sep 17 00:00:00 2001
From: James <78285353+linjames0@users.noreply.github.com>
Date: Thu, 21 Sep 2023 17:19:23 -0400
Subject: [PATCH] Update matScale.cu

---
 matScale.cu | 36 ++++++++++++++++++------------------
 1 file changed, 18 insertions(+), 18 deletions(-)

diff --git a/matScale.cu b/matScale.cu
index 1df698c..6990402 100644
--- a/matScale.cu
+++ b/matScale.cu
@@ -2,50 +2,50 @@
 #include <stdlib.h>
 
 // scale kernel
-__global__ void matScale(float *d_A, float *d_B, float scale, int N, int M) {
+__global__ void matScale(float *d_A, float *d_B, float scale, int M, int N) {
 	int row = blockIdx.y * blockDim.y + threadIdx.y;
 	int col = blockIdx.x * blockDim.x + threadIdx.x;
 
 	// scale matrix elements
-	if (row < N && col < M) {
-		d_B[row * M + col] = d_A[row * M + col] / scale;
+	if (row < M && col < N) {
+		d_B[row * N + col] = d_A[row * N + col] / scale;
 	}
 }
 
 int main() {
   	// var declaration
-	int N = 3;
 	int M = 3;
+	int N = 3;
 	float scale = 2.0f;
-	float *A, *B;
+	float *h_A, *h_B;
 	float *d_A, *d_B;
 
 	// allocate host memory
-	A = (float *)malloc(N * M * sizeof(float));
-	B = (float *)malloc(N * M * sizeof(float));
+	h_A = (float *)malloc(M * N * sizeof(float));
+	h_B = (float *)malloc(M * N * sizeof(float));
 
 	// allocate device memory
-	cudaMalloc(&d_A, N * M * sizeof(float));
-	cudaMalloc(&d_B, N * M * sizeof(float));
+	cudaMalloc(&d_A, M * N * sizeof(float));
+	cudaMalloc(&d_B, M * N * sizeof(float));
 
 	// initialize data
-	for (int i = 0; i < N * M; ++i) {
-		A[i] = i - 3;
-		B[i] = i;
+	for (int i = 0; i < M * N; ++i) {
+		h_A[i] = i - 3;
+		h_B[i] = i;
 	}
 	
 	// copy host data to device
-	cudaMemcpy(d_A, A, N * M * sizeof(float), cudaMemcpyHostToDevice);
-	cudaMemcpy(d_B, B, N * M * sizeof(float), cudaMemcpyHostToDevice);
+	cudaMemcpy(d_A, h_A, M * N * sizeof(float), cudaMemcpyHostToDevice);
+	cudaMemcpy(d_B, h_B, M * N * sizeof(float), cudaMemcpyHostToDevice);
         
 	// launch kernel instance
 	dim3 blockDim(16, 16);
-	dim3 gridDim((M + blockDim.x - 1)/blockDim.x, (N + blockDim.y - 1)/blockDim.y);
-	matScale<<<gridDim, blockDim>>>(d_A, d_B, scale, N, M);
+	dim3 gridDim((N + blockDim.x - 1)/blockDim.x, (M + blockDim.y - 1)/blockDim.y);
+	matScale<<<gridDim, blockDim>>>(d_A, d_B, scale, M, N);
 	
 	// copy result back to host
-	cudaMemcpy(A, d_A, N * M * sizeof(float), cudaMemcpyDeviceToHost);
-	cudaMemcpy(B, d_B, N * M * sizeof(float), cudaMemcpyDeviceToHost);
+	cudaMemcpy(A, d_A, M * N * sizeof(float), cudaMemcpyDeviceToHost);
+	cudaMemcpy(B, d_B, M * N * sizeof(float), cudaMemcpyDeviceToHost);
 
   	// display results
 	printf("Matrix A: \n");