summaryrefslogtreecommitdiff
path: root/silx/resources/opencl/backproj.cl
diff options
context:
space:
mode:
Diffstat (limited to 'silx/resources/opencl/backproj.cl')
-rw-r--r--silx/resources/opencl/backproj.cl485
1 files changed, 485 insertions, 0 deletions
diff --git a/silx/resources/opencl/backproj.cl b/silx/resources/opencl/backproj.cl
new file mode 100644
index 0000000..6fadc2c
--- /dev/null
+++ b/silx/resources/opencl/backproj.cl
@@ -0,0 +1,485 @@
+/*
+ * Project: silx: filtered backprojection
+ *
+ * Copyright (C) 2016-2017 European Synchrotron Radiation Facility
+ * Grenoble, France
+ *
+ * Principal authors: A. Mirone
+ * P. Paleo
+ *
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use,
+ * copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following
+ * conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
+ * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
+ * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+/*******************************************************************************/
+/************************ GPU VERSION (with textures) **************************/
+/*******************************************************************************/
+
+
+kernel void backproj_kernel(
+ int num_proj,
+ int num_bins,
+ float axis_position,
+ global float *d_SLICE,
+ read_only image2d_t d_sino,
+ float gpu_offset_x,
+ float gpu_offset_y,
+ global float * d_cos_s, // precalculated cos(theta[i])
+ global float* d_sin_s, // precalculated sin(theta[i])
+ global float* d_axis_s, // array of axis positions (n_projs)
+ local float* shared2) // 768B of local mem
+{
+ const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE | CLK_ADDRESS_CLAMP_TO_EDGE | CLK_FILTER_LINEAR;
+ const int tidx = get_local_id(0); //threadIdx.x;
+ const int bidx = get_group_id(0); //blockIdx.x;
+ const int tidy = get_local_id(1); //threadIdx.y;
+ const int bidy = get_group_id(1); //blockIdx.y;
+
+ //~ local float shared[768];
+ //~ float * sh_sin = shared;
+ //~ float * sh_cos = shared+256;
+ //~ float * sh_axis = sh_cos+256;
+
+ local float sh_cos[256];
+ local float sh_sin[256];
+ local float sh_axis[256];
+
+ float pcos, psin;
+ float h0, h1, h2, h3;
+ const float apos_off_x= gpu_offset_x - axis_position ;
+ const float apos_off_y= gpu_offset_y - axis_position ;
+ float acorr05;
+ float res0 = 0, res1 = 0, res2 = 0, res3 = 0;
+
+ const float bx00 = (32 * bidx + 2 * tidx + 0 + apos_off_x ) ;
+ const float by00 = (32 * bidy + 2 * tidy + 0 + apos_off_y ) ;
+
+ int read=0;
+ for(int proj=0; proj<num_proj; proj++) {
+ if(proj>=read) {
+ barrier(CLK_LOCAL_MEM_FENCE);
+ int ip = tidy*16+tidx;
+ if( read+ip < num_proj) {
+ sh_cos [ip] = d_cos_s[read+ip] ;
+ sh_sin [ip] = d_sin_s[read+ip] ;
+ sh_axis[ip] = d_axis_s[read+ip] ;
+ }
+ read=read+256; // 256=16*16 block size
+ barrier(CLK_LOCAL_MEM_FENCE);
+ }
+ pcos = sh_cos[256-read + proj] ;
+ psin = sh_sin[256-read + proj] ;
+
+ acorr05 = sh_axis[256 - read + proj] ;
+
+ h0 = (acorr05 + bx00*pcos - by00*psin);
+ h1 = (acorr05 + (bx00+0)*pcos - (by00+1)*psin);
+ h2 = (acorr05 + (bx00+1)*pcos - (by00+0)*psin);
+ h3 = (acorr05 + (bx00+1)*pcos - (by00+1)*psin);
+
+ if(h0>=0 && h0<num_bins) res0 += read_imagef(d_sino, sampler, (float2) (h0 +0.5f,proj +0.5f)).x; // tex2D(texprojs,h0 +0.5f,proj +0.5f);
+ if(h1>=0 && h1<num_bins) res1 += read_imagef(d_sino, sampler, (float2) (h1 +0.5f,proj +0.5f)).x; // tex2D(texprojs,h1 +0.5f,proj +0.5f);
+ if(h2>=0 && h2<num_bins) res2 += read_imagef(d_sino, sampler, (float2) (h2 +0.5f,proj +0.5f)).x; // tex2D(texprojs,h2 +0.5f,proj +0.5f);
+ if(h3>=0 && h3<num_bins) res3 += read_imagef(d_sino, sampler, (float2) (h3 +0.5f,proj +0.5f)).x; // tex2D(texprojs,h3 +0.5f,proj +0.5f);
+ }
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+0) + bidx*32 + tidx*2 + 0] = res0;
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+1) + bidx*32 + tidx*2 + 0] = res1;
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+0) + bidx*32 + tidx*2 + 1] = res2;
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+1) + bidx*32 + tidx*2 + 1] = res3;
+}
+
+
+
+
+
+/*******************************************************************************/
+/********************* CPU VERSION (without textures) **************************/
+/*******************************************************************************/
+
+
+#define CLIP_MAX(x, N) (fmin(fmax(x, 0.0f), (N - 1.0f)))
+
+#define FLOORCEIL_x(x) {\
+ xm = (int) floor(x);\
+ xp = (int) ceil(x);\
+}
+
+#define ADJACENT_PIXELS_VALS(arr, Nx, y, xm, xp) ((float2) (arr[y*Nx+xm], arr[y*Nx+xp]))
+
+//Simple linear interpolator for working on the GPU
+static float linear_interpolation(float2 vals,
+ float x,
+ int xm,
+ int xp)
+{
+ if (xm == xp)
+ return vals.s0;
+ else
+ return (vals.s0 * (xp - x)) + (vals.s1 * (x - xm));
+}
+
+/**
+ *
+ * Same kernel as backproj_kernel, but targets the CPU (no texture)
+ *
+**/
+kernel void backproj_cpu_kernel(
+ int num_proj,
+ int num_bins,
+ float axis_position,
+ global float *d_SLICE,
+ global float* d_sino,
+ float gpu_offset_x,
+ float gpu_offset_y,
+ global float * d_cos_s, // precalculated cos(theta[i])
+ global float * d_sin_s, // precalculated sin(theta[i])
+ global float * d_axis_s, // array of axis positions (n_projs)
+ local float* shared2) // 768B of local mem
+{
+ const int tidx = get_local_id(0); //threadIdx.x;
+ const int bidx = get_group_id(0); //blockIdx.x;
+ const int tidy = get_local_id(1); //threadIdx.y;
+ const int bidy = get_group_id(1); //blockIdx.y;
+
+ local float sh_cos[256];
+ local float sh_sin[256];
+ local float sh_axis[256];
+
+ float pcos, psin;
+ float h0, h1, h2, h3;
+ const float apos_off_x= gpu_offset_x - axis_position ;
+ const float apos_off_y= gpu_offset_y - axis_position ;
+ float acorr05;
+ float res0 = 0, res1 = 0, res2 = 0, res3 = 0;
+
+ const float bx00 = (32 * bidx + 2 * tidx + 0 + apos_off_x ) ;
+ const float by00 = (32 * bidy + 2 * tidy + 0 + apos_off_y ) ;
+
+ int read=0;
+ for(int proj=0; proj<num_proj; proj++) {
+ if(proj>=read) {
+ barrier(CLK_LOCAL_MEM_FENCE);
+ int ip = tidy*16+tidx;
+ if( read+ip < num_proj) {
+ sh_cos [ip] = d_cos_s[read+ip] ;
+ sh_sin [ip] = d_sin_s[read+ip] ;
+ sh_axis[ip] = d_axis_s[read+ip] ;
+ }
+ read=read+256; // 256=16*16 block size
+ barrier(CLK_LOCAL_MEM_FENCE);
+ }
+ pcos = sh_cos[256-read + proj] ;
+ psin = sh_sin[256-read + proj] ;
+
+ acorr05 = sh_axis[256 - read + proj] ;
+
+ h0 = (acorr05 + bx00*pcos - by00*psin);
+ h1 = (acorr05 + (bx00+0)*pcos - (by00+1)*psin);
+ h2 = (acorr05 + (bx00+1)*pcos - (by00+0)*psin);
+ h3 = (acorr05 + (bx00+1)*pcos - (by00+1)*psin);
+
+
+ float x;
+ int ym, xm, xp;
+ ym = proj;
+ float2 vals;
+
+ if(h0>=0 && h0<num_bins) {
+ x = CLIP_MAX(h0, num_bins);
+ FLOORCEIL_x(x);
+ vals = ADJACENT_PIXELS_VALS(d_sino, num_bins, ym, xm, xp);
+ res0 += linear_interpolation(vals, x, xm, xp);
+ }
+ if(h1>=0 && h1<num_bins) {
+ x = CLIP_MAX(h1, num_bins);
+ FLOORCEIL_x(x);
+ vals = ADJACENT_PIXELS_VALS(d_sino, num_bins, ym, xm, xp);
+ res1 += linear_interpolation(vals, x, xm, xp);
+ }
+ if(h2>=0 && h2<num_bins) {
+ x = CLIP_MAX(h2, num_bins);
+ FLOORCEIL_x(x);
+ vals = ADJACENT_PIXELS_VALS(d_sino, num_bins, ym, xm, xp);
+ res2 += linear_interpolation(vals, x, xm, xp);
+ }
+ if(h3>=0 && h3<num_bins) {
+ x = CLIP_MAX(h3, num_bins);
+ FLOORCEIL_x(x);
+ vals = ADJACENT_PIXELS_VALS(d_sino, num_bins, ym, xm, xp);
+ res3 += linear_interpolation(vals, x, xm, xp);
+ }
+ }
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+0) + bidx*32 + tidx*2 + 0] = res0;
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+1) + bidx*32 + tidx*2 + 0] = res1;
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+0) + bidx*32 + tidx*2 + 1] = res2;
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+1) + bidx*32 + tidx*2 + 1] = res3;
+}
+
+
+
+
+
+
+
+/*******************************************************************************/
+/************************** OLD STUFF, for tinkering **************************/
+/*******************************************************************************/
+
+
+
+
+/// arr(xm, ym), arr(xm, yp), arr(xp, yp), arr(xp, ym)
+//~ #define ADJACENT_PIXELS_VALS2(arr, Nx, xm, xp, ym, yp) ((float4) (arr[ym*Nx + xm], arr[yp*Nx + xm], arr[yp*Nx + xp], arr[ym*Nx + xp]))
+
+
+/** xm, xp, ym, yp **/
+//~ #define ADJACENT_PIXELS_COORDS(x, y) ((int4)((int) floor(x), (int) ceil(x), (int) floor(y), (int) ceil(y)))
+
+/**
+ (xm, ym) (xp, ym)
+ (x, y)
+ (xm, yp) (xp, yp)
+**/
+/// arr(xm, ym), arr(xm, yp), arr(xp, yp), arr(xp, ym)
+//~ #define ADJACENT_PIXELS_VALS(arr, Nx, coords) ((float4) (arr[coords.s2*Nx + coords.s0], arr[coords.s3*Nx + coords.s0], arr[coords.s3*Nx + coords.s1], arr[coords.s2*Nx + coords.s1]))
+
+
+/** xm, xp **/
+//~ #define ADJACENT_PIXELS_COORDS2(x) ((int2)((int) floor(x), (int) ceil(x)))
+
+
+
+/*
+float bilinear_interpolation(
+ float x, // x position in the image
+ float y, // y position in the image
+ int Nx, // image width
+ int Ny, // image height
+ int4 adj_coords,
+ float4 adj_vals
+) {
+ float val;
+ float tol = 0.001f; // CHECKME
+ val = y - adj_coords.s2;
+ if ((x - adj_coords.s0) < tol && (y - adj_coords.s2) < tol) val = adj_vals.s0;
+ else if ((adj_coords.s1 - x) < tol && (adj_coords.s3 - y) < tol) val = adj_vals.s2;
+ else {
+ // Mirror - TODO: clamp ?
+ if (adj_coords.s0 < 0) adj_coords.s0 = 0;
+ if (adj_coords.s1 >= Nx) adj_coords.s1 = Nx - 1;
+ if (adj_coords.s2 < 0) adj_coords.s2 = 0;
+ if (adj_coords.s3 >= Ny) adj_coords.s3 = Ny -1;
+ if (adj_coords.s0 >= Nx) adj_coords.s0 = Nx - 1;
+ if (adj_coords.s2 >= Ny) adj_coords.s2 = Ny -1;
+ // Interp
+ val = adj_vals.s1*(adj_coords.s1-x)*(y-adj_coords.s2)
+ + adj_vals.s2 *(x-adj_coords.s0)*(y-adj_coords.s2)
+ + adj_vals.s0 *(adj_coords.s1-x)*(adj_coords.s3-y)
+ + adj_vals.s3 *(x-adj_coords.s0)*(adj_coords.s3-y);
+
+ }
+ return val;
+}
+*/
+
+
+/*
+__kernel void backproj_cpu_kernel_good(
+ int num_proj,
+ int num_bins,
+ float axis_position,
+ __global float *d_SLICE,
+ __global float* d_sino,
+ float gpu_offset_x,
+ float gpu_offset_y,
+ __global float * d_cos_s, // precalculated cos(theta[i])
+ __global float * d_sin_s, // precalculated sin(theta[i])
+ __global float * d_axis_s, // array of axis positions (n_projs)
+ __local float* shared2) // 768B of local mem
+{
+ const int tidx = get_local_id(0); //threadIdx.x;
+ const int bidx = get_group_id(0); //blockIdx.x;
+ const int tidy = get_local_id(1); //threadIdx.y;
+ const int bidy = get_group_id(1); //blockIdx.y;
+
+ //~ __local float shared[768];
+ //~ float * sh_sin = shared;
+ //~ float * sh_cos = shared+256;
+ //~ float * sh_axis = sh_cos+256;
+
+ __local float sh_cos[256];
+ __local float sh_sin[256];
+ __local float sh_axis[256];
+
+ float pcos, psin;
+ float h0, h1, h2, h3;
+ const float apos_off_x= gpu_offset_x - axis_position ;
+ const float apos_off_y= gpu_offset_y - axis_position ;
+ float acorr05;
+ float res0 = 0, res1 = 0, res2 = 0, res3 = 0;
+
+ const float bx00 = (32 * bidx + 2 * tidx + 0 + apos_off_x ) ;
+ const float by00 = (32 * bidy + 2 * tidy + 0 + apos_off_y ) ;
+
+ int read=0;
+ for(int proj=0; proj<num_proj; proj++) {
+ if(proj>=read) {
+ barrier(CLK_LOCAL_MEM_FENCE);
+ int ip = tidy*16+tidx;
+ if( read+ip < num_proj) {
+ sh_cos [ip] = d_cos_s[read+ip] ;
+ sh_sin [ip] = d_sin_s[read+ip] ;
+ sh_axis[ip] = d_axis_s[read+ip] ;
+ }
+ read=read+256; // 256=16*16 block size
+ barrier(CLK_LOCAL_MEM_FENCE);
+ }
+ pcos = sh_cos[256-read + proj] ;
+ psin = sh_sin[256-read + proj] ;
+
+ acorr05 = sh_axis[256 - read + proj] ;
+
+ h0 = (acorr05 + bx00*pcos - by00*psin);
+ h1 = (acorr05 + (bx00+0)*pcos - (by00+1)*psin);
+ h2 = (acorr05 + (bx00+1)*pcos - (by00+0)*psin);
+ h3 = (acorr05 + (bx00+1)*pcos - (by00+1)*psin);
+
+
+ float x, val;
+ float tol = 0.001f; // CHECKME
+ float y = proj + 0.5f;
+ int ym = (int) floor(y);
+ int yp = (int) ceil(y);
+ int xm, xp;
+
+ //
+ int i0, i1, j0, j1;
+ float d0, d1, x0, x1, y0, y1;
+ d0 = fmin(fmax(proj+0*0.5f, 0.0f), (num_proj - 1.0f));
+ x0 = floor(d0);
+ x1 = ceil(d0);
+ i0 = (int) x0;
+ i1 = (int) x1;
+
+ if(h0>=0 && h0<num_bins) {
+ d1 = fmin(fmax(h0+0*0.5f, 0.0f), (num_bins - 1.0f));
+ y0 = floor(d1);
+ y1 = ceil(d1);
+ j0 = (int) y0;
+ j1 = (int) y1;
+
+ if ((i0 == i1) && (j0 == j1))
+ val = d_sino[i0*num_bins + j0]; //self.data[i0, j0]
+ else if (i0 == i1)
+ val = (d_sino[i0*num_bins + j0] * (y1 - d1)) + (d_sino[i0*num_bins + j1] * (d1 - y0)); // self.data[i0, j0], self.data[i0, j1]
+ else if (j0 == j1)
+ val = (d_sino[i0*num_bins + j0] * (x1 - d0)) + (d_sino[i1*num_bins + j0] * (d0 - x0)); // i0, j0 ; i1, j0
+ else
+ val = (d_sino[i0*num_bins + j0] * (x1 - d0) * (y1 - d1)) // i0, j0
+ + (d_sino[i1*num_bins + j0] * (d0 - x0) * (y1 - d1)) // i1, j0
+ + (d_sino[i0*num_bins + j1] * (x1 - d0) * (d1 - y0)) // i0, j1
+ + (d_sino[i1*num_bins + j1] * (d0 - x0) * (d1 - y0)); // i1, j1
+
+ res0 += val;
+ }
+ if(h1>=0 && h1<num_bins) {
+ //~ int4 coords = ADJACENT_PIXELS_COORDS(h1 +0.5f, proj +0.5f);
+ //~ res1 += bilinear_interpolation(h1 +0.5f, proj +0.5f, num_bins, num_proj, coords, ADJACENT_PIXELS_VALS(d_sino, num_bins, coords)); //tex2D(texProjes,h1 +0.5f,proj +0.5f);
+ d1 = fmin(fmax(h1+0*0.5f, 0.0f), (num_bins - 1.0f));
+ y0 = floor(d1);
+ y1 = ceil(d1);
+ j0 = (int) y0;
+ j1 = (int) y1;
+
+ if ((i0 == i1) && (j0 == j1))
+ val = d_sino[i0*num_bins + j0]; //self.data[i0, j0]
+ else if (i0 == i1)
+ val = (d_sino[i0*num_bins + j0] * (y1 - d1)) + (d_sino[i0*num_bins + j1] * (d1 - y0)); // self.data[i0, j0], self.data[i0, j1]
+ else if (j0 == j1)
+ val = (d_sino[i0*num_bins + j0] * (x1 - d0)) + (d_sino[i1*num_bins + j0] * (d0 - x0)); // i0, j0 ; i1, j0
+ else
+ val = (d_sino[i0*num_bins + j0] * (x1 - d0) * (y1 - d1)) // i0, j0
+ + (d_sino[i1*num_bins + j0] * (d0 - x0) * (y1 - d1)) // i1, j0
+ + (d_sino[i0*num_bins + j1] * (x1 - d0) * (d1 - y0)) // i0, j1
+ + (d_sino[i1*num_bins + j1] * (d0 - x0) * (d1 - y0)); // i1, j1
+
+
+ res1 += val;
+ }
+ if(h2>=0 && h2<num_bins) {
+ //~ int4 coords = ADJACENT_PIXELS_COORDS(h2 +0.5f, proj +0.5f);
+ //~ res2 += 0; //bilinear_interpolation(h2 +0.5f, proj +0.5f, num_bins, num_proj, coords, ADJACENT_PIXELS_VALS(d_sino, num_bins, coords)); //tex2D(texProjes,h2 +0.5f,proj +0.5f);
+ d1 = fmin(fmax(h2+0*0.5f, 0.0f), (num_bins - 1.0f));
+ y0 = floor(d1);
+ y1 = ceil(d1);
+ j0 = (int) y0;
+ j1 = (int) y1;
+
+ if ((i0 == i1) && (j0 == j1))
+ val = d_sino[i0*num_bins + j0]; //self.data[i0, j0]
+ else if (i0 == i1)
+ val = (d_sino[i0*num_bins + j0] * (y1 - d1)) + (d_sino[i0*num_bins + j1] * (d1 - y0)); // self.data[i0, j0], self.data[i0, j1]
+ else if (j0 == j1)
+ val = (d_sino[i0*num_bins + j0] * (x1 - d0)) + (d_sino[i1*num_bins + j0] * (d0 - x0)); // i0, j0 ; i1, j0
+ else
+ val = (d_sino[i0*num_bins + j0] * (x1 - d0) * (y1 - d1)) // i0, j0
+ + (d_sino[i1*num_bins + j0] * (d0 - x0) * (y1 - d1)) // i1, j0
+ + (d_sino[i0*num_bins + j1] * (x1 - d0) * (d1 - y0)) // i0, j1
+ + (d_sino[i1*num_bins + j1] * (d0 - x0) * (d1 - y0)); // i1, j1
+
+ res2+= val;
+ }
+ if(h3>=0 && h3<num_bins) {
+ //~ int4 coords = ADJACENT_PIXELS_COORDS(h3 +0.5f, proj +0.5f);
+ //~ res3 += 0; //bilinear_interpolation(h3 +0.5f, proj +0.5f, num_bins, num_proj, coords, ADJACENT_PIXELS_VALS(d_sino, num_bins, coords)); //tex2D(texProjes,h3 +0.5f,proj +0.5f);
+ d1 = fmin(fmax(h3+0*0.5f, 0.0f), (num_bins - 1.0f));
+ y0 = floor(d1);
+ y1 = ceil(d1);
+ j0 = (int) y0;
+ j1 = (int) y1;
+
+ if ((i0 == i1) && (j0 == j1))
+ val = d_sino[i0*num_bins + j0]; //self.data[i0, j0]
+ else if (i0 == i1)
+ val = (d_sino[i0*num_bins + j0] * (y1 - d1)) + (d_sino[i0*num_bins + j1] * (d1 - y0)); // self.data[i0, j0], self.data[i0, j1]
+ else if (j0 == j1)
+ val = (d_sino[i0*num_bins + j0] * (x1 - d0)) + (d_sino[i1*num_bins + j0] * (d0 - x0)); // i0, j0 ; i1, j0
+ else
+ val = (d_sino[i0*num_bins + j0] * (x1 - d0) * (y1 - d1)) // i0, j0
+ + (d_sino[i1*num_bins + j0] * (d0 - x0) * (y1 - d1)) // i1, j0
+ + (d_sino[i0*num_bins + j1] * (x1 - d0) * (d1 - y0)) // i0, j1
+ + (d_sino[i1*num_bins + j1] * (d0 - x0) * (d1 - y0)); // i1, j1
+
+ res3 += val;
+ }
+ }
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+0) + bidx*32 + tidx*2 + 0] = res0;
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+1) + bidx*32 + tidx*2 + 0] = res1;
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+0) + bidx*32 + tidx*2 + 1] = res2;
+ d_SLICE[ 32*get_num_groups(0)*(bidy*32+tidy*2+1) + bidx*32 + tidx*2 + 1] = res3;
+}
+*/
+
+
+
+