Merge branch 'master' of https://github.com/gkernel/cuda_lab

gaurav/1_Assign/ex5/main.cu

@@ -38,14 +38,15 @@ __device__ void gradImage(float* imgIn, float* forwardDiffX, float* forwardDiffY
     int ix = threadIdx.x + blockDim.x * blockIdx.x;
     int iy = threadIdx.y + blockDim.y * blockIdx.y;
 
-    imgOut[ix + (iy * w)] = 0;
+    float value = 0.; 
     for(int i = 0; i < nc; i++)
     {
-        forwardDiffX[ix + (iy * w) + (i * w * h)] = (ix < (w-2)) ? (imgIn[ix + (iy * w) + (i * w * h) + 1] - imgIn[ix + (iy * w) + (i * w * h)]) : 0;
-        forwardDiffY[ix + (iy * w) + (i * w * h)] = (iy < (h-2)) ? (imgIn[ix + ((iy + 1) * w) + (i * w * h)] - imgIn[ix + (iy * w) + (i * w * h)]) : 0;
-        imgOut[ix + (iy * w)] += pow(forwardDiffX[ix + (iy * w) + (i * w * h)] , 2) +  pow(forwardDiffY[ix + (iy * w) + (i * w * h)] , 2);
+	int idx = ix + iy*w + i*w*h;
+        forwardDiffX[idx] = (ix < (w-2)) ? (imgIn[idx + 1] - imgIn[idx]) : 0;
+        forwardDiffY[idx] = (iy < (h-2)) ? (imgIn[idx + w] - imgIn[idx]) : 0;
+        value += pow(forwardDiffX[idx] , 2) +  pow(forwardDiffY[idx] , 2);
     }
-    imgOut[ix + (iy * w)] = sqrt(imgOut[ix + (iy * w)]);
+    imgOut[ix + (iy * w)] = sqrt(value);
 }
 
 __global__ void callKernel(float* imgIn, float* forwardDiffX, float* forwardDiffY, float* imgOut, int w, int h, int nc)

gaurav/1_Assign/ex6/main.cu

@@ -29,7 +29,7 @@
 using namespace std;
 
 // uncomment to use the camera
-#define CAMERA
+//#define CAMERA
 
 #define USING_GPU
 
@@ -122,7 +122,7 @@ int main(int argc, char **argv)
     getParam("gray", gray, argc, argv);
     cout << "gray: " << gray << endl;
 
-    float sigma = 0.5;
+    float sigma = 2.0;
     getParam("sigma", sigma, argc, argv);
     cout << "sigma = " << sigma << endl;
 
@@ -274,7 +274,7 @@ int main(int argc, char **argv)
 	size_t count = w * h * nc;        
 
         // Thread Dimensions
-        dim3 block = dim3(16, 4, nc);
+        dim3 block = dim3(16, 8, nc);
         dim3 grid = dim3((w + block.x - 1) / block.x, (h + block.y - 1) / block.y, 1);
 
         // Allocating memory on the device