cleanup

src/TileProcessor.cuh

@@ -1218,22 +1218,22 @@ extern "C" __global__ void correlate2D_inner(
 #define USE_CDP
 #ifdef USE_CDP
 extern "C" __global__ void generate_RBGA(
-// Parameters to generate texture tasks
+		// Parameters to generate texture tasks
 		struct tp_task   * gpu_tasks,
 		int                num_tiles,          // number of tiles in task list
-// declare arrays in device code?
+		// declare arrays in device code?
 		int              * gpu_texture_indices,// packed tile + bits (now only (1 << 7)
 		int              * num_texture_tiles,  // number of texture tiles to process  (8 separate elements for accumulation)
 		int              * woi,                // x,y,width,height of the woi
 		int                width,  // <= TILESX, use for faster processing of LWIR images (should be actual + 1)
 		int                height, // <= TILESY, use for faster processing of LWIR images
-// Parameters for the texture generation
+		// Parameters for the texture generation
 		float          ** gpu_clt,            // [NUM_CAMS] ->[TILESY][TILESX][NUM_COLORS][DTT_SIZE*DTT_SIZE]
 		// TODO: use geometry_correction rXY !
 		struct gc       * gpu_geometry_correction,
 		int               colors,             // number of colors (3/1)
 		int               is_lwir,            // do not perform shot correction
-			float             params[5],
+		float             params[5],          // mitigating CUDA_ERROR_INVALID_PTX
 		/*
 			float             min_shot,           // 10.0
 			float             scale_shot,         // 3.0
@@ -1248,12 +1248,6 @@ extern "C" __global__ void generate_RBGA(
 		float           * gpu_texture_tiles)  // (number of colors +1 + ?)*16*16 rgba texture tiles
 //			float             aaaa)
 //			float           * gpu_diff_rgb_combo) // diff[NUM_CAMS], R[NUM_CAMS], B[NUM_CAMS],G[NUM_CAMS]
-/*
- 			const size_t      texture_rbga_stride,     // in floats
-			float           * gpu_texture_tiles,  // (number of colors +1 + ?)*16*16 rgba texture tiles
-			float             aaaa)
-
- */
 {
 	float             min_shot = params[0];           // 10.0
 	float             scale_shot = params[1];         // 3.0

src/TileProcessor.h

@@ -122,34 +122,33 @@ extern "C" __global__ void imclt_rbg(
 		int               woi_theight,
 		const size_t      dstride);            // in floats (pixels)
 
-/*
 extern "C" __global__ void generate_RBGA(
-// Parameters to generate texture tasks
+		// Parameters to generate texture tasks
 		struct tp_task   * gpu_tasks,
 		int                num_tiles,          // number of tiles in task list
-// declare arrays in device code?
+		// declare arrays in device code?
 		int              * gpu_texture_indices,// packed tile + bits (now only (1 << 7)
 		int              * num_texture_tiles,  // number of texture tiles to process  (8 separate elements for accumulation)
 		int              * woi,                // x,y,width,height of the woi
 		int                width,  // <= TILESX, use for faster processing of LWIR images (should be actual + 1)
 		int                height, // <= TILESY, use for faster processing of LWIR images
-// Parameters for the texture generation
+		// Parameters for the texture generation
 		float          ** gpu_clt,            // [NUM_CAMS] ->[TILESY][TILESX][NUM_COLORS][DTT_SIZE*DTT_SIZE]
 		// TODO: use geometry_correction rXY !
 		struct gc       * gpu_geometry_correction,
-//			float           * gpu_geometry_correction,
-//			float           * gpu_port_offsets,       // relative ports x,y offsets - just to scale differences, may be approximate
 		int               colors,             // number of colors (3/1)
 		int               is_lwir,            // do not perform shot correction
+		float             params[5],          // mitigating CUDA_ERROR_INVALID_PTX
+		/*
 			float             min_shot,           // 10.0
 			float             scale_shot,         // 3.0
 			float             diff_sigma,         // pixel value/pixel change
 			float             diff_threshold,     // pixel value/pixel change
 			float             min_agree,          // minimal number of channels to agree on a point (real number to work with fuzzy averages)
+		 */
 		float             weights[3],         // scale for R,B,G
 		int               dust_remove,        // Do not reduce average weight when only one image differs much from the average
 		int               keep_weights,       // return channel weights after A in RGBA (was removed)
 		const size_t      texture_rbga_stride,     // in floats
-			float           * gpu_texture_tiles,  // (number of colors +1 + ?)*16*16 rgba texture tiles
-			float           * gpu_diff_rgb_combo); // diff[NUM_CAMS], R[NUM_CAMS], B[NUM_CAMS],G[NUM_CAMS]
-*/
+		float           * gpu_texture_tiles);  // (number of colors +1 + ?)*16*16 rgba texture tiles
+