less bad test

TransformationGrid.cu

@@ -7,6 +7,24 @@
 
 #include <TransformationGrid.h>
 
+__device__
+float identity(float value)
+{
+   return value;
+}
+
+__device__
+float entropy(float value)
+{
+   return value;
+}
+
+__device__ pointFloatToFloat_t p_identity = identity;
+//the host-side function pointer to your __device__ function
+
+
+//in host code: copy the function pointers to their host equivalent
+
 void threeaxisrot(double r11, double r12, double r21, double r31, double r32, float3* const& res){
   res->z = (float) atan2( r31, r32 ) + M_PI;
   res->y = (float) asin ( r21 ) + M_PI_2;
@@ -173,6 +191,9 @@ TransformationGrid::TransformationGrid(const std::pair<float3,float3> bounding_b
 	cudaMalloc((void **) &d_scores,h_num_transformations[0]*sizeof(*d_scores));
 	cudaMalloc((void **) &d_tmp_rot_quadrature_scores,h_num_translations[0]*sizeof(*d_scores));
 	CudaCheckError();
+
+	cudaMemcpyFromSymbol(&h_identity, p_identity, sizeof(pointFloatToFloat_t));
+	CudaCheckError();
 }
 
 __host__ void TransformationGrid::quadrature_test_function(void){
@@ -181,13 +202,14 @@ __host__ void TransformationGrid::quadrature_test_function(void){
 	//being 8*pi^2 for the euler parametrisation of SO(3)
 	float rotation_normalization = 2/M_PI;
 	//normalization for 6D space
-	float six_d_normalization = translation_normalization*rotation_normalization;
-//	float six_d_normalization_inv = rotation_normalization;
+//	float six_d_normalization = translation_normalization*rotation_normalization;
+	float six_d_normalization = translation_normalization;
 	int rot_index;
 	int trans_index;
 	for (uint i=0;i<h_num_transformations[0];i++){
 		rot_index = i/h_num_translations[0];
-		h_scores[i] = six_d_normalization/sinf(h_rotations_euler[rot_index].y);
+//		h_scores[i] = six_d_normalization/sinf(h_rotations_euler[rot_index].y);
+		h_scores[i] = six_d_normalization;
 	}
 	cudaMemcpy(d_scores,h_scores,h_num_transformations[0]*sizeof(*h_scores),cudaMemcpyHostToDevice);
 	CudaCheckError();
@@ -392,7 +414,7 @@ void translation_quadrature(TransformationGrid grid, float * d_tmp_quadrature){
 
 template <uint blockSize>
 __global__
-void rotation_quadrature(TransformationGrid grid){
+void rotation_quadrature(TransformationGrid grid,pointFloatToFloat_t function){
 	__shared__ float reduce[blockSize];
 	uint num_translations = grid.d_num_translations[0];
 	uint num_rotations = grid.d_num_rotations[0];
@@ -413,7 +435,8 @@ void rotation_quadrature(TransformationGrid grid){
 			//						.								.
 			//						.								.
 			//					  (r_{N_r-1},t0);(r_{N_r-1},t1);(r_{N_r-1},t2); ... ;(r_{N_r-1},t_{N_t-1});
-			reduce[threadIdx.x] += d_scores[t*num_translations+b];
+			//t-Design! - weights implicit at the end
+			reduce[threadIdx.x] += function(d_scores[t*num_translations+b]);
 //			if (b ==1 && threadIdx.x ==0){
 //				printf("%i,%i,%f\n",t,t*num_translations,d_scores[t*num_translations]);
 //			}
@@ -445,7 +468,7 @@ void TransformationGrid::quadrature(float* const &result){
 //	template <uint blockSize>
 //	__global__
 //	void rotation_quadrature(TransformationGrid grid){
-	rotation_quadrature<block_size><<<rot_quat_grid_dim,block_size>>>(*this);
+	rotation_quadrature<block_size><<<rot_quat_grid_dim,block_size>>>(*this,h_identity);
 	cudaDeviceSynchronize();
 
 	uint secondary_quadrature_grid_dim = h_num_translations[0]/block_size + 1;
@@ -770,10 +793,10 @@ void TransformationGrid::shift_and_scale_chamfer_score(float * d_score, float *
 	cudaMalloc((void **)&d_sum,sizeof(*d_sum));
 	cudaMemcpy(d_sum,h_sum,sizeof(*h_sum),cudaMemcpyHostToDevice);
 	CudaCheckError();
-	thrust::for_each(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]), thrust::placeholders::_1 /= sum);
-	thrust::for_each(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]), entropy());
-	float entr = -1.0*thrust::reduce(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]));
-	printf("Entropy for envelope score: %f\n",entr);
+//	thrust::for_each(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]), thrust::placeholders::_1 /= sum);
+//	thrust::for_each(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]), entropy());
+//	float entr = -1.0*thrust::reduce(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]));
+//	printf("Entropy for envelope score: %f\n",entr);
 }
 
 
@@ -792,10 +815,10 @@ void TransformationGrid::shift_and_scale_envelope_score(float * d_score, float *
 	cudaMalloc((void **)&d_sum,sizeof(*d_sum));
 	cudaMemcpy(d_sum,h_sum,sizeof(*h_sum),cudaMemcpyHostToDevice);
 	CudaCheckError();
-	thrust::for_each(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]), thrust::placeholders::_1 /= sum);
-	thrust::for_each(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]), entropy());
-	float entr = -1.0*thrust::reduce(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]));
-	printf("Entropy for envelope score: %f\n",entr);
+//	thrust::for_each(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]), thrust::placeholders::_1 /= sum);
+//	thrust::for_each(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]), entropy());
+//	float entr = -1.0*thrust::reduce(thrust::device_pointer_cast(&d_normalized_score[0]),thrust::device_pointer_cast(&d_normalized_score[num_scores-1]));
+//	printf("Entropy for envelope score: %f\n",entr);
 }
 
 TransformationGrid::~TransformationGrid() {

TransformationGrid.h

@@ -47,6 +47,7 @@
 //#include "DelaunayChecker.h"
 //
 
+typedef float(*pointFloatToFloat_t)(float);
 
 class TransformationGrid{
 public:
@@ -85,6 +86,7 @@ public:
 	float * d_scores;
 	uint * h_num_transformations;
 	uint * d_num_transformations;
+	pointFloatToFloat_t h_identity;
 	__device__ int4 d_transformation_index_from_linear(const uint& linear_index);
 	__host__ int4 h_transformation_index_from_linear(const uint& linear_index);
 	__device__ void d_transformation_to_memory(const uint& i, float4 * const& rotation, float4 * const& translation);