305 __forceinline__ __device__ void load_array(float *smem, int ig, int jg, int w, int h, int p)  in load_array()  argument
309     load_array_element<tex>(smem, i, j, ig, jg, w, h, p);//load current pixel  in load_array()
314         load_array_element<tex>(smem, i, j-2, ig, jg-2, w, h, p);  in load_array()
318 …          load_array_element<tex>(smem, i+PSOR_TILE_WIDTH, j-2, ig+PSOR_TILE_WIDTH, jg-2, w, h, p);  in load_array()
320             load_array_element<tex>(smem, i+PSOR_TILE_WIDTH, j, ig+PSOR_TILE_WIDTH, jg, w, h, p);  in load_array()
325 …          load_array_element<tex>(smem, i-PSOR_TILE_WIDTH, j-2, ig-PSOR_TILE_WIDTH, jg-2, w, h, p);  in load_array()
327             load_array_element<tex>(smem, i-PSOR_TILE_WIDTH, j, ig-PSOR_TILE_WIDTH, jg, w, h, p);  in load_array()
333         load_array_element<tex>(smem, i, j+2, ig, jg+2, w, h, p);  in load_array()
337 …          load_array_element<tex>(smem, i+PSOR_TILE_WIDTH, j+2, ig+PSOR_TILE_WIDTH, jg+2, w, h, p);  in load_array()
339             load_array_element<tex>(smem, i+PSOR_TILE_WIDTH, j, ig+PSOR_TILE_WIDTH, jg, w, h, p);  in load_array()
344 …          load_array_element<tex>(smem, i-PSOR_TILE_WIDTH, j+2, ig-PSOR_TILE_WIDTH, jg+2, w, h, p);  in load_array()
346             load_array_element<tex>(smem, i-PSOR_TILE_WIDTH, j, ig-PSOR_TILE_WIDTH, jg, w, h, p);  in load_array()
355             load_array_element<tex>(smem, i+PSOR_TILE_WIDTH, j, ig+PSOR_TILE_WIDTH, jg, w, h, p);  in load_array()
360             load_array_element<tex>(smem, i-PSOR_TILE_WIDTH, j, ig-PSOR_TILE_WIDTH, jg, w, h, p);  in load_array()
404     const int ig  = blockIdx.x * blockDim.x + threadIdx.x;  in prepare_sor_stage_1_tex()  local
406     const int ijg = jg * s + ig;  in prepare_sor_stage_1_tex()
408     float x = (float)ig + 0.5f;  in prepare_sor_stage_1_tex()
411     load_array<0>(u, ig, jg, w, h, s);  in prepare_sor_stage_1_tex()
412     load_array<1>(v, ig, jg, w, h, s);  in prepare_sor_stage_1_tex()
413     load_array<2>(du, ig, jg, w, h, s);  in prepare_sor_stage_1_tex()
414     load_array<3>(dv, ig, jg, w, h, s);  in prepare_sor_stage_1_tex()
440     if(ig >= w || jg >= h) return;  in prepare_sor_stage_1_tex()
445     if(ig == 0) sx = 0.0f;  in prepare_sor_stage_1_tex()
476     const int ig  = blockIdx.x * blockDim.x + threadIdx.x;  in prepare_sor_stage_2()  local
478     const int ijg = jg*s + ig;  in prepare_sor_stage_2()
479     int inside = ig < w && jg < h;  in prepare_sor_stage_2()
512         if(ig < w-1 && inside)  in prepare_sor_stage_2()