about done

Author: jenzy

NBody OpenCL/Main.cpp

@@ -30,7 +30,7 @@ void printHelp() {
 int main( int argc, char **argv ) {
 	info_t info;
 	info.n = 1000; 				
-	info.steps = 100;	
+	info.steps = 1000;	
 	info.sphereRadius = 10; //10
 	info.kappa = 1; 			
 	info.mass = 1; 			
@@ -42,11 +42,11 @@ int main( int argc, char **argv ) {
 	info.randFunc = SPHERE_2_POLES;
 
 	bool doMPI		= false;
-	bool doCPU		= true;
-	bool doCPUOpt	= true;
-	bool doGPU1		= false;
-	bool doGPU2		= false;
-	bool doGPU3		= false;
+	bool doCPU		= false;
+	bool doCPUOpt	= false;
+	bool doGPU1		= true;
+	bool doGPU2		= true;
+	bool doGPU3		= true;
 	bool doCombo	= false;
 	bool doGL		= false;
 

NBody OpenCL/WOCL.cpp

@@ -169,4 +169,7 @@ void WOCL::AcquireObjectsFromGLAndFinish( cl_uint num, cl_mem *objects ) {
 void WOCL::ReleaseObjectsToGLAndFinish( cl_uint num, cl_mem *objects ) {
 	ret = clEnqueueReleaseGLObjects( m_queue, num, objects, 0, NULL, NULL );		CheckForError( ret, "clEnqueueReleaseGLObjects" );
 	ret = clFinish( m_queue );														CheckForError( ret, "clFinish @ ReleaseObjectsToGL" );
+}
+void WOCL::Finish() {
+	ret = clFinish( m_queue );														CheckForError( ret, "clFinish" );
 }

NBody OpenCL/WOCL.h

@@ -103,6 +103,7 @@ class WOCL {
 	void ExecuteKernel();
 	void AcquireObjectsFromGLAndFinish( cl_uint num, cl_mem *objects );
 	void ReleaseObjectsToGLAndFinish( cl_uint num, cl_mem *objects );
+	void Finish();
 
 	void SetAndAllocKernelArgument( int idx, size_t size );
 	template<typename T> void SetKernelArgument( int idx, T *parameter ) {

NBody OpenCL/gpu.cpp

@@ -25,7 +25,7 @@ void gpu( info_t *info ) {
 	float *V = (float *) calloc( info->n, sizeof(float) );
 	generateCoordinates( X, Y, Z, M, info );
 
-	t.Tic();
+	t.TicSimple();
 	gpu.CreateAndBuildKernel( "res/kernelFloat1.cl", "kernelFloat1" );
 
 	// Device alokacija in kopiranje podatkov
@@ -70,7 +70,8 @@ void gpu( info_t *info ) {
 	gpu.CopyDeviceToHost( &devY, Y, info->n*sizeof(float) );
 	gpu.CopyDeviceToHost( &devZ, Z, info->n*sizeof(float) );
 
-	printf( "Cas izvajanja: %lf\n", t.Toc() );
+	gpu.Finish();
+	printf( "Time: %.3lf\n", t.TocSimple() );
 	checkResults( X, Y, Z, info->n );
 
 #pragma region Cleanup
@@ -93,7 +94,7 @@ void gpuVec( info_t *info ) {
 	float *V = (float *) calloc( info->n, 4 * sizeof(float) );
 	generateCoordinatesFloat4( Coord, info );
 
-	t.Tic( );
+	t.TicSimple( );
 	gpu.CreateAndBuildKernel( "res/kernelVec.cl", "kernelVec" );
 
 	// Device alokacija in kopiranje podatkov
@@ -118,7 +119,8 @@ void gpuVec( info_t *info ) {
 	// Prenos rezultatov na gostitelja
 	gpu.CopyDeviceToHost( &devCoord, Coord, info->n*sizeof(cl_float4) );
 
-	printf( "Cas izvajanja: %lf\n", t.Toc() );
+	gpu.Finish( );
+	printf( "Time: %.3lf\n", t.TocSimple() );
 	checkResultsFloat4( Coord, info->n );
 
 	free( V );
@@ -137,7 +139,7 @@ void gpuVecLocal( info_t *info ) {
 	float *V = (float *) calloc( info->n, 4 * sizeof(float) );
 	generateCoordinatesFloat4( Coord, info );
 
-	t.Tic( );
+	t.TicSimple( );
 	gpu.CreateAndBuildKernel( "res/kernelVecLocal.cl", "kernelVecLocal" );
 
 	// Device alokacija in kopiranje podatkov
@@ -163,7 +165,8 @@ void gpuVecLocal( info_t *info ) {
 	// Prenos rezultatov na gostitelja
 	gpu.CopyDeviceToHost( &devCoord, Coord, info->n*sizeof(cl_float4) );
 
-	printf( "Cas izvajanja: %lf\n", t.Toc( ) );
+	gpu.Finish();
+	printf( "Time: %.3lf\n", t.TocSimple( ) );
 	checkResultsFloat4( Coord, info->n );
 
 	free( V );