aboutsummaryrefslogtreecommitdiff
path: root/src/diffraction-gpu.c
diff options
context:
space:
mode:
Diffstat (limited to 'src/diffraction-gpu.c')
-rw-r--r--src/diffraction-gpu.c142
1 files changed, 63 insertions, 79 deletions
diff --git a/src/diffraction-gpu.c b/src/diffraction-gpu.c
index f41a45ec..ff4f2414 100644
--- a/src/diffraction-gpu.c
+++ b/src/diffraction-gpu.c
@@ -3,11 +3,13 @@
*
* Calculate diffraction patterns by Fourier methods (GPU version)
*
- * Copyright © 2012 Deutsches Elektronen-Synchrotron DESY,
- * a research centre of the Helmholtz Association.
+ * Copyright © 2012-2014 Deutsches Elektronen-Synchrotron DESY,
+ * a research centre of the Helmholtz Association.
*
* Authors:
- * 2010-2012 Thomas White <taw@physics.org>
+ * 2009-2014 Thomas White <taw@physics.org>
+ * 2013 Alexandra Tolstikova
+ * 2013-2014 Chun Hong Yoon <chun.hong.yoon@desy.de>
*
* This file is part of CrystFEL.
*
@@ -168,18 +170,20 @@ static int set_arg_mem(struct gpu_context *gctx, int idx, cl_mem val)
static void do_panels(struct gpu_context *gctx, struct image *image,
- int *n_inf, int *n_neg, int *n_nan, double k, double frac,
- double xo, double yo)
+ double k, double weight,
+ int *n_inf, int *n_neg, int *n_nan)
{
int i;
+ const int sampling = 4; /* This, squared, number of samples / pixel */
if ( set_arg_float(gctx, 1, k) ) return;
+ if ( set_arg_float(gctx, 2, weight) ) return;
/* Iterate over panels */
for ( i=0; i<image->det->n_panels; i++ ) {
size_t dims[2];
- size_t ldims[2] = {1, 1};
+ size_t ldims[2];
struct panel *p;
cl_mem diff;
size_t diff_size;
@@ -203,20 +207,29 @@ static void do_panels(struct gpu_context *gctx, struct image *image,
}
if ( set_arg_mem(gctx, 0, diff) ) return;
- if ( set_arg_int(gctx, 2, pan_width) ) return;
- if ( set_arg_float(gctx, 3, p->cnx) ) return;
- if ( set_arg_float(gctx, 4, p->cny) ) return;
- if ( set_arg_float(gctx, 5, p->res) ) return;
- if ( set_arg_float(gctx, 6, p->clen) ) return;
- if ( set_arg_float(gctx, 13, p->fsx) ) return;
- if ( set_arg_float(gctx, 14, p->fsy) ) return;
- if ( set_arg_float(gctx, 15, p->ssx) ) return;
- if ( set_arg_float(gctx, 16, p->ssy) ) return;
- if ( set_arg_float(gctx, 17, xo) ) return;
- if ( set_arg_float(gctx, 18, yo) ) return;
-
- dims[0] = pan_width;
- dims[1] = pan_height;
+
+ if ( set_arg_int(gctx, 3, pan_width) ) return;
+ if ( set_arg_float(gctx, 4, p->cnx) ) return;
+ if ( set_arg_float(gctx, 5, p->cny) ) return;
+ if ( set_arg_float(gctx, 6, p->fsx) ) return;
+ if ( set_arg_float(gctx, 7, p->fsy) ) return;
+ if ( set_arg_float(gctx, 8, p->ssx) ) return;
+ if ( set_arg_float(gctx, 9, p->ssy) ) return;
+ if ( set_arg_float(gctx, 10, p->res) ) return;
+ if ( set_arg_float(gctx, 11, p->clen) ) return;
+
+ dims[0] = pan_width * sampling;
+ dims[1] = pan_height * sampling;
+
+ ldims[0] = sampling;
+ ldims[1] = sampling;
+
+ err = clSetKernelArg(gctx->kern, 18,
+ sampling*sampling*sizeof(cl_float), NULL);
+ if ( err != CL_SUCCESS ) {
+ ERROR("Couldn't set local memory: %s\n", clError(err));
+ return;
+ }
err = clEnqueueNDRangeKernel(gctx->cq, gctx->kern, 2, NULL,
dims, ldims, 0, NULL, NULL);
@@ -250,7 +263,7 @@ static void do_panels(struct gpu_context *gctx, struct image *image,
tfs = p->min_fs + fs;
tss = p->min_ss + ss;
- image->data[tfs + image->width*tss] += frac*val;
+ image->data[tfs + image->width*tss] += val;
}
}
@@ -276,78 +289,54 @@ void get_diffraction_gpu(struct gpu_context *gctx, struct image *image,
int n_neg = 0;
int n_nan = 0;
int fs, ss;
- const int nkstep = 10;
- const int nxs = 4;
- const int nys = 4;
- int kstep;
- double klow, khigh, kinc, w;
- double frac;
- int xs, ys;
- int n, ntot;
+ int i;
if ( gctx == NULL ) {
ERROR("GPU setup failed.\n");
return;
}
- cell_get_cartesian(ucell, &ax, &ay, &az, &bx, &by, &bz, &cx, &cy, &cz);
- cell.s[0] = ax; cell.s[1] = ay; cell.s[2] = az;
- cell.s[3] = bx; cell.s[4] = by; cell.s[5] = bz;
- cell.s[6] = cx; cell.s[7] = cy; cell.s[8] = cz;
-
/* Ensure all required LUTs are available */
check_sinc_lut(gctx, na);
check_sinc_lut(gctx, nb);
check_sinc_lut(gctx, nc);
- if ( set_arg_mem(gctx, 8, gctx->intensities) ) return;
- if ( set_arg_mem(gctx, 9, gctx->sinc_luts[na-1]) ) return;
- if ( set_arg_mem(gctx, 10, gctx->sinc_luts[nb-1]) ) return;
- if ( set_arg_mem(gctx, 11, gctx->sinc_luts[nc-1]) ) return;
- if ( set_arg_mem(gctx, 12, gctx->flags) ) return;
-
/* Unit cell */
- err = clSetKernelArg(gctx->kern, 7, sizeof(cl_float16), &cell);
+ cell_get_cartesian(ucell, &ax, &ay, &az, &bx, &by, &bz, &cx, &cy, &cz);
+ cell.s[0] = ax; cell.s[1] = ay; cell.s[2] = az;
+ cell.s[3] = bx; cell.s[4] = by; cell.s[5] = bz;
+ cell.s[6] = cx; cell.s[7] = cy; cell.s[8] = cz;
+
+ err = clSetKernelArg(gctx->kern, 12, sizeof(cl_float16), &cell);
if ( err != CL_SUCCESS ) {
ERROR("Couldn't set unit cell: %s\n", clError(err));
return;
}
+ if ( set_arg_mem(gctx, 13, gctx->intensities) ) return;
+ if ( set_arg_mem(gctx, 14, gctx->flags) ) return;
+ if ( set_arg_mem(gctx, 15, gctx->sinc_luts[na-1]) ) return;
+ if ( set_arg_mem(gctx, 16, gctx->sinc_luts[nb-1]) ) return;
+ if ( set_arg_mem(gctx, 17, gctx->sinc_luts[nc-1]) ) return;
+
/* Allocate memory for the result */
image->data = calloc(image->width * image->height, sizeof(float));
- w = image->lambda * image->bw;
- klow = 1.0 / (image->lambda + (w/2.0)); /* Smallest k */
- khigh = 1.0 / (image->lambda - (w/2.0)); /* Largest k */
- kinc = (khigh-klow) / (nkstep+1);
-
- ntot = nkstep * nxs * nys;
- frac = 1.0 / ntot;
-
- n = 0;
- for ( xs=0; xs<nxs; xs++ ) {
- for ( ys=0; ys<nys; ys++ ) {
-
- double xo, yo;
-
- xo = (1.0/nxs) * xs;
- yo = (1.0/nys) * ys;
+ double tot = 0.0;
+ for ( i=0; i<image->nsamples; i++ ) {
- for ( kstep=0; kstep<nkstep; kstep++ ) {
+ printf("%.1f eV, weight = %.5f\n",
+ ph_lambda_to_eV(1.0/image->spectrum[i].k),
+ image->spectrum[i].weight);
- double k;
+ do_panels(gctx, image, image->spectrum[i].k,
+ image->spectrum[i].weight,
+ &n_inf, &n_neg, &n_nan);
- k = klow + kstep*kinc;
+ tot += image->spectrum[i].weight;
- do_panels(gctx, image, &n_inf, &n_neg, &n_nan, k, frac,
- xo, yo);
-
- n++;
- progress_bar(n, ntot, "Simulating");
-
- }
- }
}
+ printf("total weight = %f\n", tot);
if ( n_neg + n_inf + n_nan ) {
ERROR("WARNING: The GPU calculation produced %i negative"
@@ -360,16 +349,9 @@ void get_diffraction_gpu(struct gpu_context *gctx, struct image *image,
for ( fs=0; fs<image->width; fs++ ) {
for ( ss=0; ss<image->height; ss++ ) {
- double twotheta, k;
- int idx;
-
- /* Calculate k this time round */
- k = 1.0/image->lambda;
-
- get_q(image, fs, ss, &twotheta, k);
-
- idx = fs + image->width*ss;
- image->twotheta[idx] = twotheta;
+ double twotheta;
+ get_q(image, fs, ss, &twotheta, 1.0/image->lambda);
+ image->twotheta[fs + image->width*ss] = twotheta;
}
}
@@ -437,7 +419,7 @@ struct gpu_context *setup_gpu(int no_sfac,
}
} else {
for ( i=0; i<IDIM*IDIM*IDIM; i++ ) {
- intensities_ptr[i] = 1e5;
+ intensities_ptr[i] = 100.0; /* Does nothing */
}
strncat(cflags, "-DFLAT_INTENSITIES ", 511-strlen(cflags));
}
@@ -468,6 +450,8 @@ struct gpu_context *setup_gpu(int no_sfac,
strncat(cflags, "-DPG23 ", 511-strlen(cflags));
} else if ( strcmp(sym, "m-3") == 0 ) {
strncat(cflags, "-DPGM3 ", 511-strlen(cflags));
+ } else if ( strcmp(sym, "321_H") == 0 ) {
+ strncat(cflags, "-DPG321H ", 511-strlen(cflags));
} else {
ERROR("Sorry! Point group '%s' is not currently"
" supported on the GPU."