Add CUDA macro, add Makefile and syncfile

.sync.toml

+# This file steers synchronization of the current folder with different target machines
+# See https://github.com/AndiH/simpler-rsync for more information
+
+[juhydra]
+	hostname = "juhydra"
+	target_folder = "~/TMP/CUDA-Hello-World/"
+
+[juron]
+	hostname = "juron"
+	target_folder = "~/TMP/JURON/CUDA-Hello-World"
+	default = true

Makefile

+NVCC=nvcc
+
+hello-world: hello-world.cu Makefile
+	$(NVCC) -o hello-world hello-world.cu
+
+.PHONY: clean run
+clean:
+	rm hello-world
+run: hello-world
+	bsub -R "rusage[ngpus_shared=20]" -I ./hello-world

hello-world.cu

@@ -6,6 +6,23 @@
 
 // nvcc hello-world.cu -L /usr/local/cuda/lib -lcudart -o hello-world
 
+//Macro for checking cuda errors
+// #define CUDA_CALL(call) {                                          \
+//  cudaError_t e = call;                                 \
+//  if(e != cudaSuccess) {                                              \
+//    printf("Cuda failure %s:%d: '%s' in %s\n", __FILE__, __LINE__, cudaGetErrorString(e), #call);           \
+//    exit(0); \
+//  }                                                                 \
+// }
+#include <iostream>
+
+#define CUDA_CALL( call )				\
+{                                       \
+cudaError_t result = call;              \
+if ( cudaSuccess != result )            \
+	std::cerr << "CUDA error " << result << " in " << __FILE__ << ":" << __LINE__ << ": " << cudaGetErrorString( result ) << " (" << #call << ")" << std::endl;  \
+}
+
 #include <stdio.h>
 
 const int N = 16; 
@@ -29,17 +46,17 @@ int main()
 
 	printf("%s", a);
 
-	cudaMalloc( (void**)&ad, csize ); 
-	cudaMalloc( (void**)&bd, isize ); 
-	cudaMemcpy( ad, a, csize, cudaMemcpyHostToDevice ); 
-	cudaMemcpy( bd, b, isize, cudaMemcpyHostToDevice ); 
+	CUDA_CALL(cudaMalloc( (void**)&ad, csize );)
+	CUDA_CALL(cudaMalloc( (void**)&bd, isize );)
+	CUDA_CALL(cudaMemcpy( ad, a, csize, cudaMemcpyHostToDevice );)
+	CUDA_CALL(cudaMemcpy( bd, b, isize, cudaMemcpyHostToDevice );)
 	
 	dim3 dimBlock( blocksize, 1 );
 	dim3 dimGrid( 1, 1 );
 	hello<<<dimGrid, dimBlock>>>(ad, bd);
-	cudaMemcpy( a, ad, csize, cudaMemcpyDeviceToHost ); 
-	cudaFree( ad );
-	cudaFree( bd );
+	CUDA_CALL(cudaMemcpy( a, ad, csize, cudaMemcpyDeviceToHost );)
+	CUDA_CALL(cudaFree( ad );)
+	CUDA_CALL(cudaFree( bd );)
 	
 	printf("%s\n", a);
 //	sleep(1);