Catch2 Test Fixes (#2961)

* test fixes

* address PR comment

* PR comment fixing Nvidia pass

* add additional kernel launch checks

* pr comments

tests/src/cudaRegister.cu

@@ -69,8 +69,8 @@ int main(){
 	check("Registering A",status);
 	cudaHostGetDevicePointer(&Ad, A, 0);
 	cudaMalloc((void**)&Bd, SIZE);
-	dim3 dimGrid(LEN/512,1,1);
-	dim3 dimBlock(512,1,1);
+	dim3 dimGrid(LEN/256,1,1);
+	dim3 dimBlock(256,1,1);
 	Inc1<<<dimGrid, dimBlock>>>(Ad, Bd);
 	sleep(3);
 	A[0] = -(ITER*1.0f);

tests/src/runtimeApi/memory/hipHostGetFlags.cpp

@@ -71,8 +71,8 @@ int main() {
         B[i] = 2.0f;
     }
 
-    dim3 dimGrid(LEN / 512, 1, 1);
-    dim3 dimBlock(512, 1, 1);
+    dim3 dimGrid(LEN / 256, 1, 1);
+    dim3 dimBlock(256, 1, 1);
 
     hipLaunchKernelGGL(HIP_KERNEL_NAME(Add), dimGrid, dimBlock, 0, 0, Ad, Bd, Cd);
 

tests/src/runtimeApi/memory/hipHostMalloc.cpp

@@ -134,8 +134,8 @@ int main() {
         HIPCHECK(hipMalloc((void**)&Bd, SIZE));
         HIPCHECK(hipMemcpy(Bd, B, SIZE, hipMemcpyHostToDevice));
 
-        dim3 dimGrid(LEN / 512, 1, 1);
-        dim3 dimBlock(512, 1, 1);
+        dim3 dimGrid(LEN / 256, 1, 1);
+        dim3 dimBlock(256, 1, 1);
 
         hipLaunchKernelGGL(Add, dimGrid, dimBlock, 0, 0, Ad, Bd, Cd);