ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/OpenCL/OpenCL.xs
(Generate patch)

Comparing OpenCL/OpenCL.xs (file contents):
Revision 1.59 by root, Sun Apr 29 19:38:05 2012 UTC vs.
Revision 1.83 by root, Mon Apr 7 22:36:54 2014 UTC

1#include <stddef.h>
2
1#include "EXTERN.h" 3#include "EXTERN.h"
2#include "perl.h" 4#include "perl.h"
3#include "XSUB.h" 5#include "XSUB.h"
6
7#define ECB_NO_THREADS 1
8#define ECB_NO_LIBM 1
9#include "ecb.h"
4 10
5#define X_STACKSIZE sizeof (void *) * 512 * 1024 // 2-4mb should be enough, really 11#define X_STACKSIZE sizeof (void *) * 512 * 1024 // 2-4mb should be enough, really
6#include "xthread.h" 12#include "xthread.h"
7#include "schmorp.h" 13#include "schmorp.h"
8 14
26#if PREFER_1_1 32#if PREFER_1_1
27 #define CL_USE_DEPRECATED_OPENCL_1_1_APIS 33 #define CL_USE_DEPRECATED_OPENCL_1_1_APIS
28#endif 34#endif
29 35
30#ifdef __APPLE__ 36#ifdef __APPLE__
31 #include <OpenCL/opencl.h> 37 #define CLHDR(name) <OpenCL/name>
32#else 38#else
39 #define CLHDR(name) <CL/name>
40#endif
41
33 #include <CL/opencl.h> 42#include CLHDR(opencl.h)
43
44#if 0
45#ifndef CL_VERSION_1_2
46 #include CLHDR(cl_d3d9.h)
47#endif
48#endif
49
50#if _WIN32
51 #include CLHDR(cl_d3d10.h)
52 #if CL_VERSION_1_2
53 #include CLHDR(cl_d3d11.h)
54 #endif
55 #include CLHDR(cl_dx9_media_sharing.h)
34#endif 56#endif
35 57
36#ifndef CL_VERSION_1_2 58#ifndef CL_VERSION_1_2
37 #undef PREFER_1_1 59 #undef PREFER_1_1
38 #define PREFER_1_1 1 60 #define PREFER_1_1 1
39#endif 61#endif
40 62
63// make sure all constants we might use are actually defined
64#include "default.h"
65
41typedef cl_platform_id OpenCL__Platform; 66typedef cl_platform_id OpenCL__Platform;
42typedef cl_device_id OpenCL__Device; 67typedef cl_device_id OpenCL__Device;
68typedef cl_device_id OpenCL__SubDevice;
43typedef cl_context OpenCL__Context; 69typedef cl_context OpenCL__Context;
44typedef cl_command_queue OpenCL__Queue; 70typedef cl_command_queue OpenCL__Queue;
45typedef cl_mem OpenCL__Memory; 71typedef cl_mem OpenCL__Memory;
46typedef cl_mem OpenCL__Buffer; 72typedef cl_mem OpenCL__Buffer;
47typedef cl_mem OpenCL__BufferObj; 73typedef cl_mem OpenCL__BufferObj;
48typedef cl_mem OpenCL__Image; 74typedef cl_mem OpenCL__Image;
49typedef cl_mem OpenCL__Image2D;
50typedef cl_mem OpenCL__Image3D;
51typedef cl_mem OpenCL__Memory_ornull; 75typedef cl_mem OpenCL__Memory_ornull;
52typedef cl_mem OpenCL__Buffer_ornull; 76typedef cl_mem OpenCL__Buffer_ornull;
53typedef cl_mem OpenCL__Image_ornull; 77typedef cl_mem OpenCL__Image_ornull;
54typedef cl_mem OpenCL__Image2D_ornull;
55typedef cl_mem OpenCL__Image3D_ornull;
56typedef cl_sampler OpenCL__Sampler; 78typedef cl_sampler OpenCL__Sampler;
57typedef cl_program OpenCL__Program; 79typedef cl_program OpenCL__Program;
58typedef cl_kernel OpenCL__Kernel; 80typedef cl_kernel OpenCL__Kernel;
59typedef cl_event OpenCL__Event; 81typedef cl_event OpenCL__Event;
60typedef cl_event OpenCL__UserEvent; 82typedef cl_event OpenCL__UserEvent;
61 83
62typedef SV *FUTURE; 84typedef struct mapped * OpenCL__Mapped;
85
86static HV
87 *stash_platform,
88 *stash_device,
89 *stash_subdevice,
90 *stash_context,
91 *stash_queue,
92 *stash_program,
93 *stash_kernel,
94 *stash_sampler,
95 *stash_event,
96 *stash_userevent,
97 *stash_memory,
98 *stash_buffer,
99 *stash_bufferobj,
100 *stash_image,
101 *stash_image1d,
102 *stash_image1darray,
103 *stash_image1dbuffer,
104 *stash_image2d,
105 *stash_image2darray,
106 *stash_image3d,
107 *stash_mapped,
108 *stash_mappedbuffer,
109 *stash_mappedimage;
63 110
64/*****************************************************************************/ 111/*****************************************************************************/
65 112
66// name must include a leading underscore 113// name must include a leading underscore
67// all of this horrors would be unneceesary if somebody wrote a proper OpenGL module 114// all of this horrors would be unneceesary if somebody wrote a proper OpenGL module
107} 154}
108 155
109/*****************************************************************************/ 156/*****************************************************************************/
110 157
111/* up to two temporary buffers */ 158/* up to two temporary buffers */
112static void * 159static void * ecb_noinline
113tmpbuf (size_t size) 160tmpbuf (size_t size)
114{ 161{
115 enum { buffers = 3 }; 162 enum { buffers = 4 };
116 static int idx; 163 static int idx;
117 static void *buf [buffers]; 164 static void *buf [buffers];
118 static size_t len [buffers]; 165 static size_t len [buffers];
119 166
120 idx = (idx + 1) % buffers; 167 idx = (idx + 1) % buffers;
127 } 174 }
128 175
129 return buf [idx]; 176 return buf [idx];
130} 177}
131 178
179static const char * ecb_noinline
180cv_get_name (CV *cv)
181{
182 static char fullname [256];
183
184 GV *gv = CvGV (cv); // gv better be nonzero
185
186 HV *stash = GvSTASH (gv);
187 const char *hvname = HvNAME_get (stash); // stash also better be nonzero
188 const char *gvname = GvNAME (gv);
189
190 snprintf (fullname, sizeof (fullname), "%s::%s", hvname, gvname);
191 return fullname;
192}
193
132/*****************************************************************************/ 194/*****************************************************************************/
133 195
134typedef struct 196typedef struct
135{ 197{
136 IV iv; 198 IV iv;
137 const char *name; 199 const char *name;
138 #define const_iv(name) { (IV)CL_ ## name, # name }, 200 #define const_iv(name) { (IV)CL_ ## name, # name },
139} ivstr; 201} ivstr;
202
203typedef struct
204{
205 NV nv;
206 const char *name;
207 #define const_nv(name) { (NV)CL_ ## name, # name },
208} nvstr;
140 209
141static const char * 210static const char *
142iv2str (IV value, const ivstr *base, int count, const char *fallback) 211iv2str (IV value, const ivstr *base, int count, const char *fallback)
143{ 212{
144 int i; 213 int i;
193 if (res) \ 262 if (res) \
194 FAIL (name); 263 FAIL (name);
195 264
196/*****************************************************************************/ 265/*****************************************************************************/
197 266
267static SV *
268new_clobj (HV *stash, IV id)
269{
270 return sv_2mortal (sv_bless (newRV_noinc (newSViv (id)), stash));
271}
272
273#define PUSH_CLOBJ(stash,id) PUSHs (new_clobj ((stash), (IV)(id)))
274#define XPUSH_CLOBJ(stash,id) XPUSHs (new_clobj ((stash), (IV)(id)))
275
276/* cl objects are either \$iv, or [$iv, ...] */
277/* they can be upgraded at runtime to the array form */
278static void * ecb_noinline
279SvCLOBJ (CV *cv, const char *svname, SV *sv, const char *pkg)
280{
281 // sv_derived_from is quite slow :(
282 if (SvROK (sv) && sv_derived_from (sv, pkg))
283 return (void *)SvIV (SvRV (sv));
284
285 croak ("%s: %s is not of type %s", cv_get_name (cv), svname, pkg);
286}
287
288// the "no-inherit" version of the above
289static void * ecb_noinline
290SvCLOBJ_ni (CV *cv, const char *svname, SV *sv, HV *stash)
291{
292 if (SvROK (sv) && SvSTASH (SvRV (sv)) == stash)
293 return (void *)SvIV (SvRV (sv));
294
295 croak ("%s: %s is not of type %s", cv_get_name (cv), svname, HvNAME (stash));
296}
297
298/*****************************************************************************/
299
198static cl_context_properties * 300static cl_context_properties * ecb_noinline
199SvCONTEXTPROPERTIES (const char *func, const char *svname, SV *sv, cl_context_properties *extra, int extracount) 301SvCONTEXTPROPERTIES (CV *cv, const char *svname, SV *sv, cl_context_properties *extra, int extracount)
200{ 302{
201 if (!sv || !SvOK (sv)) 303 if (!sv || !SvOK (sv))
202 if (extra) 304 if (extra)
203 sv = sv_2mortal (newRV_noinc ((SV *)newAV ())); // slow, but rarely used hopefully 305 sv = sv_2mortal (newRV_noinc ((SV *)newAV ())); // slow, but rarely used hopefully
204 else 306 else
210 int i, len = av_len (av) + 1; 312 int i, len = av_len (av) + 1;
211 cl_context_properties *p = tmpbuf (sizeof (cl_context_properties) * (len + extracount + 1)); 313 cl_context_properties *p = tmpbuf (sizeof (cl_context_properties) * (len + extracount + 1));
212 cl_context_properties *l = p; 314 cl_context_properties *l = p;
213 315
214 if (len & 1) 316 if (len & 1)
215 croak ("%s: %s is not a property list (must be even number of elements)", func, svname); 317 croak ("%s: %s is not a property list (must contain an even number of elements)", cv_get_name (cv), svname);
216 318
217 while (extracount--) 319 while (extracount--)
218 *l++ = *extra++; 320 *l++ = *extra++;
219 321
220 for (i = 0; i < len; i += 2) 322 for (i = 0; i < len; i += 2)
223 SV *p_sv = *av_fetch (av, i + 1, 0); 325 SV *p_sv = *av_fetch (av, i + 1, 0);
224 cl_context_properties v = SvIV (p_sv); // code below can override 326 cl_context_properties v = SvIV (p_sv); // code below can override
225 327
226 switch (t) 328 switch (t)
227 { 329 {
330 case CL_CONTEXT_PLATFORM:
331 if (SvROK (p_sv))
332 v = (cl_context_properties)SvCLOBJ (cv, svname, p_sv, "OpenCL::Platform");
333 break;
334
228 case CL_GLX_DISPLAY_KHR: 335 case CL_GLX_DISPLAY_KHR:
229 if (!SvOK (p_sv)) 336 if (!SvOK (p_sv))
230 { 337 {
231 void *func = glsym ("_glXGetCurrentDisplay"); 338 void *func = glsym ("_glXGetCurrentDisplay");
232 if (func) 339 if (func)
255 *l = 0; 362 *l = 0;
256 363
257 return p; 364 return p;
258 } 365 }
259 366
260 croak ("%s: %s is not a property list (either undef or [type => value, ...])", func, svname); 367 croak ("%s: %s is not a property list (either undef or [type => value, ...])", cv_get_name (cv), svname);
261} 368}
262 369
263/*****************************************************************************/ 370// parse an array of CLOBJ into a void ** array in C - works only for CLOBJs whose representation
264 371// is a pointer (and only on well-behaved systems).
265#define NEW_CLOBJ(class,ptr) sv_setref_pv (sv_newmortal (), class, ptr) 372static void * ecb_noinline
266#define PUSH_CLOBJ(class,ptr) PUSHs (NEW_CLOBJ (class, ptr)) 373object_list (CV *cv, int or_undef, const char *argname, SV *arg, const char *klass, cl_uint *rcount)
267#define XPUSH_CLOBJ(class,ptr) XPUSHs (NEW_CLOBJ (class, ptr))
268
269/* cl objects are either \$iv, or [$iv, ...] */
270/* they can be upgraded at runtime to the array form */
271static void *
272SvCLOBJ (const char *func, const char *svname, SV *sv, const char *pkg)
273{ 374{
274 if (SvROK (sv) && sv_derived_from (sv, pkg)) 375 if (!SvROK (arg) || SvTYPE (SvRV (arg)) != SVt_PVAV)
275 return (void *)SvIV (SvRV (sv)); 376 croak ("%s: '%s' parameter must be %sa reference to an array of %s objects",
377 cv_get_name (cv), argname, or_undef ? "undef or " : "", klass);
276 378
277 croak ("%s: %s is not of type %s", func, svname, pkg); 379 AV *av = (AV *)SvRV (arg);
380 void **list = 0;
381 cl_uint count = av_len (av) + 1;
382
383 if (count)
384 {
385 list = tmpbuf (sizeof (*list) * count);
386 int i;
387 for (i = 0; i < count; ++i)
388 list [i] = SvCLOBJ (cv, argname, *av_fetch (av, i, 1), klass);
389 }
390
391 if (!count && !or_undef)
392 croak ("%s: '%s' must contain at least one %s object",
393 cv_get_name (cv), argname, klass);
394
395 *rcount = count;
396 return (void *)list;
278} 397}
279 398
280/*****************************************************************************/ 399/*****************************************************************************/
281/* callback stuff */ 400/* callback stuff */
282 401
304static void (*eq_signal_func)(void *signal_arg, int value); 423static void (*eq_signal_func)(void *signal_arg, int value);
305static void *eq_signal_arg; 424static void *eq_signal_arg;
306static xmutex_t eq_lock = X_MUTEX_INIT; 425static xmutex_t eq_lock = X_MUTEX_INIT;
307static eq_item *eq_head, *eq_tail; 426static eq_item *eq_head, *eq_tail;
308 427
309static void 428static void ecb_noinline
310eq_enq (eq_vtbl *vtbl, SV *cb, void *data1, void *data2, void *data3) 429eq_enq (eq_vtbl *vtbl, SV *cb, void *data1, void *data2, void *data3)
311{ 430{
312 eq_item *item = malloc (sizeof (eq_item)); 431 eq_item *item = malloc (sizeof (eq_item));
313 432
314 item->next = 0; 433 item->next = 0;
381} 500}
382 501
383/*****************************************************************************/ 502/*****************************************************************************/
384/* context notify */ 503/* context notify */
385 504
386static void 505static void ecb_noinline
387eq_context_push (void *data1, void *data2, void *data3) 506eq_context_push (void *data1, void *data2, void *data3)
388{ 507{
389 dSP; 508 dSP;
390 PUSHs (sv_2mortal (newSVpv (data1, 0))); 509 PUSHs (sv_2mortal (newSVpv (data1, 0)));
391 PUSHs (sv_2mortal (newSVpvn (data2, (STRLEN)data3))); 510 PUSHs (sv_2mortal (newSVpvn (data2, (STRLEN)data3)));
413 { \ 532 { \
414 pfn_notify = eq_context_notify; \ 533 pfn_notify = eq_context_notify; \
415 user_data = s_get_cv (notify); \ 534 user_data = s_get_cv (notify); \
416 } 535 }
417 536
418static SV * 537static SV * ecb_noinline
419new_clobj_context (cl_context ctx, void *user_data) 538new_clobj_context (cl_context ctx, void *user_data)
420{ 539{
421 SV *sv = NEW_CLOBJ ("OpenCL::Context", ctx); 540 SV *sv = new_clobj (stash_context, (IV)ctx);
422 541
423 if (user_data) 542 if (user_data)
424 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0); 543 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0);
425 544
426 return sv; 545 return sv;
433 552
434static void 553static void
435eq_program_push (void *data1, void *data2, void *data3) 554eq_program_push (void *data1, void *data2, void *data3)
436{ 555{
437 dSP; 556 dSP;
438 PUSH_CLOBJ ("OpenCL::Program", data1); 557 PUSH_CLOBJ (stash_program, data1);
439 PUTBACK; 558 PUTBACK;
440} 559}
441 560
442static eq_vtbl eq_program_vtbl = { 1, eq_program_push }; 561static eq_vtbl eq_program_vtbl = { 1, eq_program_push };
443 562
444static void CL_CALLBACK 563static void CL_CALLBACK
445eq_program_notify (cl_program program, void *user_data) 564eq_program_notify (cl_program program, void *user_data)
446{ 565{
566 clRetainProgram (program);
567
447 eq_enq (&eq_program_vtbl, user_data, (void *)program, 0, 0); 568 eq_enq (&eq_program_vtbl, user_data, (void *)program, 0, 0);
569}
570
571typedef void (CL_CALLBACK *program_callback)(cl_program program, void *user_data);
572
573static program_callback ecb_noinline
574make_program_callback (SV *notify, void **ruser_data)
575{
576 if (SvOK (notify))
577 {
578 *ruser_data = SvREFCNT_inc (s_get_cv (notify));
579 return eq_program_notify;
580 }
581 else
582 {
583 *ruser_data = 0;
584 return 0;
585 }
448} 586}
449 587
450struct build_args 588struct build_args
451{ 589{
452 cl_program program; 590 cl_program program;
466 else 604 else
467 clReleaseProgram (arg->program); 605 clReleaseProgram (arg->program);
468 606
469 free (arg->options); 607 free (arg->options);
470 free (arg); 608 free (arg);
609
610 return 0;
471} 611}
472 612
473static void 613static void
474build_program_async (cl_program program, cl_uint num_devices, const cl_device_id *device_list, const char *options, void *user_data) 614build_program_async (cl_program program, cl_uint num_devices, const cl_device_id *device_list, const char *options, void *user_data)
475{ 615{
480 arg->user_data = user_data; 620 arg->user_data = user_data;
481 arg->num_devices = num_devices; 621 arg->num_devices = num_devices;
482 memcpy (arg + 1, device_list, sizeof (*device_list) * num_devices); 622 memcpy (arg + 1, device_list, sizeof (*device_list) * num_devices);
483 623
484 xthread_t id; 624 xthread_t id;
485 thread_create (&id, build_program_thread, arg); 625 xthread_create (&id, build_program_thread, arg);
626}
627
628/*****************************************************************************/
629/* mem object destructor notify */
630
631static void ecb_noinline
632eq_destructor_push (void *data1, void *data2, void *data3)
633{
634}
635
636static eq_vtbl eq_destructor_vtbl = { 0, eq_destructor_push };
637
638static void CL_CALLBACK
639eq_destructor_notify (cl_mem memobj, void *user_data)
640{
641 eq_enq (&eq_destructor_vtbl, (SV *)user_data, (void *)memobj, 0, 0);
486} 642}
487 643
488/*****************************************************************************/ 644/*****************************************************************************/
489/* event objects */ 645/* event objects */
490 646
491static void 647static void
492eq_event_push (void *data1, void *data2, void *data3) 648eq_event_push (void *data1, void *data2, void *data3)
493{ 649{
494 dSP; 650 dSP;
495 PUSH_CLOBJ ("OpenCL::Event", data1); 651 PUSH_CLOBJ (stash_event, data1);
496 PUSHs (sv_2mortal (newSViv ((IV)data2))); 652 PUSHs (sv_2mortal (newSViv ((IV)data2)));
497 PUTBACK; 653 PUTBACK;
498} 654}
499 655
500static eq_vtbl eq_event_vtbl = { 1, eq_event_push }; 656static eq_vtbl eq_event_vtbl = { 1, eq_event_push };
505 clRetainEvent (event); 661 clRetainEvent (event);
506 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0); 662 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0);
507} 663}
508 664
509/*****************************************************************************/ 665/*****************************************************************************/
666/* utilities for XS code */
510 667
511static size_t 668static size_t
512img_row_pitch (cl_mem img) 669img_row_pitch (cl_mem img)
513{ 670{
514 size_t res; 671 size_t res;
515 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0); 672 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0);
516 return res; 673 return res;
517} 674}
518 675
519static cl_event * 676static cl_event * ecb_noinline
520event_list (SV **items, cl_uint *rcount) 677event_list (CV *cv, SV **items, cl_uint *rcount, cl_event extra)
521{ 678{
522 cl_uint count = *rcount; 679 cl_uint count = *rcount;
523 680
524 if (!count) 681 if (count > 0x7fffffffU) // yeah, it's a hack - the caller might have underflowed
682 *rcount = count = 0;
683
684 if (!count && !extra)
525 return 0; 685 return 0;
526 686
527 cl_event *list = tmpbuf (sizeof (cl_event) * count); 687 cl_event *list = tmpbuf (sizeof (cl_event) * (count + 1));
528 int i = 0; 688 int i = 0;
529 689
530 do 690 while (count--)
531 {
532 --count;
533 if (SvOK (items [count])) 691 if (SvOK (items [count]))
534 list [i++] = SvCLOBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event"); 692 list [i++] = SvCLOBJ (cv, "wait_events", items [count], "OpenCL::Event");
535 } 693
536 while (count); 694 if (extra)
695 list [i++] = extra;
537 696
538 *rcount = i; 697 *rcount = i;
539 698
540 return i ? list : 0; 699 return i ? list : 0;
541} 700}
542 701
543#define EVENT_LIST(items,count) \ 702#define EVENT_LIST(skip) \
544 cl_uint event_list_count = (count); \ 703 cl_uint event_list_count = items - (skip); \
545 cl_event *event_list_ptr = event_list (&ST (items), &event_list_count) 704 cl_event *event_list_ptr = event_list (cv, &ST (skip), &event_list_count, 0)
546 705
547#define INFO(class) \ 706#define INFO(class) \
548{ \ 707{ \
549 size_t size; \ 708 size_t size; \
550 NEED_SUCCESS (Get ## class ## Info, (self, name, 0, 0, &size)); \ 709 NEED_SUCCESS (Get ## class ## Info, (self, name, 0, 0, &size)); \
554 SvCUR_set (sv, size); \ 713 SvCUR_set (sv, size); \
555 NEED_SUCCESS (Get ## class ## Info, (self, name, size, SvPVX (sv), 0)); \ 714 NEED_SUCCESS (Get ## class ## Info, (self, name, size, SvPVX (sv), 0)); \
556 XPUSHs (sv); \ 715 XPUSHs (sv); \
557} 716}
558 717
718/*****************************************************************************/
719/* mapped_xxx */
720
721static OpenCL__Mapped
722SvMAPPED (SV *sv)
723{
724 // no typechecking atm., keep your fingers crossed
725 return (OpenCL__Mapped)SvMAGIC (SvRV (sv))->mg_ptr;
726}
727
728struct mapped
729{
730 cl_command_queue queue;
731 cl_mem memobj;
732 void *ptr;
733 size_t cb;
734 cl_event event;
735 size_t row_pitch;
736 size_t slice_pitch;
737
738 size_t element_size;
739 size_t width, height, depth;
740};
741
742static SV *
743mapped_new (
744 HV *stash, cl_command_queue queue, cl_mem memobj, cl_map_flags flags,
745 void *ptr, size_t cb, cl_event ev,
746 size_t row_pitch, size_t slice_pitch, size_t element_size,
747 size_t width, size_t height, size_t depth
748)
749{
750 SV *data = newSV (0);
751 SvUPGRADE (data, SVt_PVMG);
752
753 OpenCL__Mapped mapped;
754 New (0, mapped, 1, struct mapped);
755
756 clRetainCommandQueue (queue);
757
758 mapped->queue = queue;
759 mapped->memobj = memobj;
760 mapped->ptr = ptr;
761 mapped->cb = cb;
762 mapped->event = ev;
763 mapped->row_pitch = row_pitch;
764 mapped->slice_pitch = slice_pitch;
765
766 mapped->element_size = element_size;
767 mapped->width = width;
768 mapped->height = height;
769 mapped->depth = depth;
770
771 sv_magicext (data, 0, PERL_MAGIC_ext, 0, (char *)mapped, 0);
772
773 if (SvLEN (data))
774 Safefree (data);
775
776 SvPVX (data) = (char *)ptr;
777 SvCUR_set (data, cb);
778 SvLEN_set (data, 0);
779 SvPOK_only (data);
780
781 SV *obj = sv_2mortal (sv_bless (newRV_noinc (data), stash));
782
783 if (!(flags & CL_MAP_WRITE))
784 SvREADONLY_on (data);
785
786 return obj;
787}
788
789static void
790mapped_detach (SV *sv, OpenCL__Mapped mapped)
791{
792 SV *data = SvRV (sv);
793
794 // the next check checks both before AND after detach, where SvPVX should STILL be 0
795 if (SvPVX (data) != (char *)mapped->ptr)
796 warn ("FATAL: OpenCL memory mapped scalar changed location, detected");
797 else
798 {
799 SvREADONLY_off (data);
800 SvCUR_set (data, 0);
801 SvPVX (data) = 0;
802 SvOK_off (data);
803 }
804
805 mapped->ptr = 0;
806}
807
808static void
809mapped_unmap (CV *cv, SV *self, OpenCL__Mapped mapped, cl_command_queue queue, SV **wait_list, cl_uint event_list_count)
810{
811 cl_event *event_list_ptr = event_list (cv, wait_list, &event_list_count, mapped->event);
812 cl_event ev;
813
814 NEED_SUCCESS (EnqueueUnmapMemObject, (queue, mapped->memobj, mapped->ptr, event_list_count, event_list_ptr, &ev));
815
816 clReleaseEvent (mapped->event);
817 mapped->event = ev;
818
819 mapped_detach (self, mapped);
820}
821
822static size_t
823mapped_element_size (OpenCL__Mapped self)
824{
825 if (!self->element_size)
826 clGetImageInfo (self->memobj, CL_IMAGE_ELEMENT_SIZE, sizeof (self->element_size), &self->element_size, 0);
827
828 return self->element_size;
829}
830
831/*****************************************************************************/
832
559MODULE = OpenCL PACKAGE = OpenCL 833MODULE = OpenCL PACKAGE = OpenCL
560 834
561PROTOTYPES: ENABLE 835PROTOTYPES: ENABLE
562 836
563void 837void
572 eq_signal_arg = (void*)arg; 846 eq_signal_arg = (void*)arg;
573 847
574BOOT: 848BOOT:
575{ 849{
576 HV *stash = gv_stashpv ("OpenCL", 1); 850 HV *stash = gv_stashpv ("OpenCL", 1);
851
577 static const ivstr *civ, const_iv[] = { 852 static const ivstr *civ, const_iv[] = {
578 { sizeof (cl_char ), "SIZEOF_CHAR" }, 853 { sizeof (cl_char ), "SIZEOF_CHAR" },
579 { sizeof (cl_uchar ), "SIZEOF_UCHAR" }, 854 { sizeof (cl_uchar ), "SIZEOF_UCHAR" },
580 { sizeof (cl_short ), "SIZEOF_SHORT" }, 855 { sizeof (cl_short ), "SIZEOF_SHORT" },
581 { sizeof (cl_ushort), "SIZEOF_USHORT" }, 856 { sizeof (cl_ushort), "SIZEOF_USHORT" },
584 { sizeof (cl_long ), "SIZEOF_LONG" }, 859 { sizeof (cl_long ), "SIZEOF_LONG" },
585 { sizeof (cl_ulong ), "SIZEOF_ULONG" }, 860 { sizeof (cl_ulong ), "SIZEOF_ULONG" },
586 { sizeof (cl_half ), "SIZEOF_HALF" }, 861 { sizeof (cl_half ), "SIZEOF_HALF" },
587 { sizeof (cl_float ), "SIZEOF_FLOAT" }, 862 { sizeof (cl_float ), "SIZEOF_FLOAT" },
588 { sizeof (cl_double), "SIZEOF_DOUBLE" }, 863 { sizeof (cl_double), "SIZEOF_DOUBLE" },
864 { PREFER_1_1 , "PREFER_1_1" },
589#include "constiv.h" 865#include "constiv.h"
590 }; 866 };
591 867
592 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--) 868 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--)
593 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv)); 869 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv));
870
871 static const nvstr *cnv, const_nv[] = {
872#include "constnv.h"
873 };
874
875 for (cnv = const_nv + sizeof (const_nv) / sizeof (const_nv [0]); cnv > const_nv; cnv--)
876 newCONSTSUB (stash, (char *)cnv[-1].name, newSVnv (cnv[-1].nv));
877
878 newCONSTSUB (stash, "NAN", newSVnv (CL_NAN)); // CL_NAN might be a function call
879
880 stash_platform = gv_stashpv ("OpenCL::Platform", GV_ADD);
881 stash_device = gv_stashpv ("OpenCL::Device", GV_ADD);
882 stash_subdevice = gv_stashpv ("OpenCL::SubDevice", GV_ADD);
883 stash_context = gv_stashpv ("OpenCL::Context", GV_ADD);
884 stash_queue = gv_stashpv ("OpenCL::Queue", GV_ADD);
885 stash_program = gv_stashpv ("OpenCL::Program", GV_ADD);
886 stash_kernel = gv_stashpv ("OpenCL::Kernel", GV_ADD);
887 stash_sampler = gv_stashpv ("OpenCL::Sampler", GV_ADD);
888 stash_event = gv_stashpv ("OpenCL::Event", GV_ADD);
889 stash_userevent = gv_stashpv ("OpenCL::UserEvent", GV_ADD);
890 stash_memory = gv_stashpv ("OpenCL::Memory", GV_ADD);
891 stash_buffer = gv_stashpv ("OpenCL::Buffer", GV_ADD);
892 stash_bufferobj = gv_stashpv ("OpenCL::BufferObj", GV_ADD);
893 stash_image = gv_stashpv ("OpenCL::Image", GV_ADD);
894 stash_image1d = gv_stashpv ("OpenCL::Image1D", GV_ADD);
895 stash_image1darray = gv_stashpv ("OpenCL::Image1DArray", GV_ADD);
896 stash_image1dbuffer = gv_stashpv ("OpenCL::Image1DBuffer", GV_ADD);
897 stash_image2d = gv_stashpv ("OpenCL::Image2D", GV_ADD);
898 stash_image2darray = gv_stashpv ("OpenCL::Image2DArray", GV_ADD);
899 stash_image3d = gv_stashpv ("OpenCL::Image3D", GV_ADD);
900 stash_mapped = gv_stashpv ("OpenCL::Mapped", GV_ADD);
901 stash_mappedbuffer = gv_stashpv ("OpenCL::MappedBuffer", GV_ADD);
902 stash_mappedimage = gv_stashpv ("OpenCL::MappedImage", GV_ADD);
594 903
595 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt); 904 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt);
596} 905}
597 906
598cl_int 907cl_int
619 list = tmpbuf (sizeof (*list) * count); 928 list = tmpbuf (sizeof (*list) * count);
620 NEED_SUCCESS (GetPlatformIDs, (count, list, 0)); 929 NEED_SUCCESS (GetPlatformIDs, (count, list, 0));
621 930
622 EXTEND (SP, count); 931 EXTEND (SP, count);
623 for (i = 0; i < count; ++i) 932 for (i = 0; i < count; ++i)
624 PUSH_CLOBJ ("OpenCL::Platform", list [i]); 933 PUSH_CLOBJ (stash_platform, list [i]);
625 934
626void 935void
627context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef) 936context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
628 PPCODE: 937 PPCODE:
629 CONTEXT_NOTIFY_CALLBACK; 938 CONTEXT_NOTIFY_CALLBACK;
630 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, 0, 0, &res)); 939 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, pfn_notify, user_data, &res));
631 XPUSH_CLOBJ_CONTEXT; 940 XPUSH_CLOBJ_CONTEXT;
632 941
633void 942void
634context (FUTURE properties, FUTURE devices, FUTURE notify) 943context (cl_context_properties *properties, SV *devices, SV *notify = &PL_sv_undef)
635 PPCODE: 944 PPCODE:
636 /* der Gipfel der Kunst */ 945 cl_uint device_count;
946 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
947
948 CONTEXT_NOTIFY_CALLBACK;
949 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (properties, device_count, device_list, pfn_notify, user_data, &res));
950 XPUSH_CLOBJ_CONTEXT;
637 951
638void 952void
639wait_for_events (...) 953wait_for_events (...)
640 CODE: 954 CODE:
641 EVENT_LIST (0, items); 955 EVENT_LIST (0);
642 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr)); 956 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr));
643 957
644PROTOTYPES: DISABLE 958PROTOTYPES: DISABLE
645 959
646MODULE = OpenCL PACKAGE = OpenCL::Platform 960MODULE = OpenCL PACKAGE = OpenCL::Platform
689 list = tmpbuf (sizeof (*list) * count); 1003 list = tmpbuf (sizeof (*list) * count);
690 NEED_SUCCESS (GetDeviceIDs, (self, type, count, list, 0)); 1004 NEED_SUCCESS (GetDeviceIDs, (self, type, count, list, 0));
691 1005
692 EXTEND (SP, count); 1006 EXTEND (SP, count);
693 for (i = 0; i < count; ++i) 1007 for (i = 0; i < count; ++i)
694 PUSHs (sv_setref_pv (sv_newmortal (), "OpenCL::Device", list [i])); 1008 PUSH_CLOBJ (stash_device, list [i]);
695 1009
696void 1010void
697context (OpenCL::Platform self, cl_context_properties *properties, SV *devices, SV *notify = &PL_sv_undef) 1011context (OpenCL::Platform self, SV *properties, SV *devices, SV *notify = &PL_sv_undef)
698 PPCODE: 1012 PPCODE:
1013 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
1014 cl_context_properties *props = SvCONTEXTPROPERTIES (cv, "properties", properties, extra, 2);
1015
1016 cl_uint device_count;
1017 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
1018
1019 CONTEXT_NOTIFY_CALLBACK;
1020 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (props, device_count, device_list, pfn_notify, user_data, &res));
1021 XPUSH_CLOBJ_CONTEXT;
1022
1023void
1024context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
1025 PPCODE:
1026 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
1027 cl_context_properties *props = SvCONTEXTPROPERTIES (cv, "properties", properties, extra, 2);
1028
1029 CONTEXT_NOTIFY_CALLBACK;
1030 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, pfn_notify, user_data, &res));
1031 XPUSH_CLOBJ_CONTEXT;
1032
1033MODULE = OpenCL PACKAGE = OpenCL::Device
1034
1035void
1036info (OpenCL::Device self, cl_device_info name)
1037 PPCODE:
1038 INFO (Device)
1039
1040#if CL_VERSION_1_2
1041
1042void
1043sub_devices (OpenCL::Device self, SV *properties)
1044 PPCODE:
699 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV) 1045 if (!SvROK (properties) || SvTYPE (SvRV (properties)) != SVt_PVAV)
700 croak ("OpenCL::Platform::context argument 'device' must be an arrayref with device objects, in call"); 1046 croak ("OpenCL::Device::sub_devices: properties must be specified as reference to an array of property-value pairs");
701 1047
702 AV *av = (AV *)SvRV (devices); 1048 properties = SvRV (properties);
703 cl_uint num_devices = av_len (av) + 1; 1049
704 cl_device_id *device_list = tmpbuf (sizeof (cl_device_id) * num_devices); 1050 cl_uint count = av_len ((AV *)properties) + 1;
1051 cl_device_partition_property *props = tmpbuf (sizeof (*props) * count + 1);
705 1052
706 int i; 1053 int i;
707 for (i = num_devices; i--; ) 1054 for (i = 0; i < count; ++i)
708 device_list [i] = SvCLOBJ ("clCreateContext", "devices", *av_fetch (av, i, 0), "OpenCL::Device"); 1055 props [i] = (cl_device_partition_property)SvIV (*av_fetch ((AV *)properties, i, 0));
709 1056
710 CONTEXT_NOTIFY_CALLBACK; 1057 props [count] = 0;
711 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (properties, num_devices, device_list, pfn_notify, user_data, &res));
712 XPUSH_CLOBJ_CONTEXT;
713 1058
714void 1059 cl_uint num_devices;
715context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef) 1060 NEED_SUCCESS (CreateSubDevices, (self, props, 0, 0, &num_devices));
716 PPCODE: 1061 cl_device_id *list = tmpbuf (sizeof (*list) * num_devices);
717 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self }; 1062 NEED_SUCCESS (CreateSubDevices, (self, props, num_devices, list, 0));
718 cl_context_properties *props = SvCONTEXTPROPERTIES ("OpenCL::Platform::context_from_type", "properties", properties, extra, 2);
719 1063
720 CONTEXT_NOTIFY_CALLBACK; 1064 EXTEND (SP, num_devices);
721 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, 0, 0, &res)); 1065 for (i = 0; i < count; ++i)
722 XPUSH_CLOBJ_CONTEXT; 1066 PUSH_CLOBJ (stash_subdevice, list [i]);
723 1067
724MODULE = OpenCL PACKAGE = OpenCL::Device 1068#endif
725
726void
727info (OpenCL::Device self, cl_device_info name)
728 PPCODE:
729 INFO (Device)
730 1069
731#BEGIN:device 1070#BEGIN:device
732 1071
733void 1072void
734type (OpenCL::Device self) 1073type (OpenCL::Device self)
1074 ALIAS:
1075 type = CL_DEVICE_TYPE
1076 address_bits = CL_DEVICE_ADDRESS_BITS
1077 max_mem_alloc_size = CL_DEVICE_MAX_MEM_ALLOC_SIZE
1078 single_fp_config = CL_DEVICE_SINGLE_FP_CONFIG
1079 global_mem_cache_size = CL_DEVICE_GLOBAL_MEM_CACHE_SIZE
1080 global_mem_size = CL_DEVICE_GLOBAL_MEM_SIZE
1081 max_constant_buffer_size = CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE
1082 local_mem_size = CL_DEVICE_LOCAL_MEM_SIZE
1083 execution_capabilities = CL_DEVICE_EXECUTION_CAPABILITIES
1084 properties = CL_DEVICE_QUEUE_PROPERTIES
1085 double_fp_config = CL_DEVICE_DOUBLE_FP_CONFIG
1086 half_fp_config = CL_DEVICE_HALF_FP_CONFIG
735 PPCODE: 1087 PPCODE:
736 cl_device_type value [1]; 1088 cl_ulong value [1];
737 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_TYPE, sizeof (value), value, 0)); 1089 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
738 EXTEND (SP, 1); 1090 EXTEND (SP, 1);
739 const int i = 0; 1091 const int i = 0;
740 PUSHs (sv_2mortal (newSViv (value [i]))); 1092 PUSHs (sv_2mortal (newSVuv (value [i])));
741 1093
742void 1094void
743vendor_id (OpenCL::Device self) 1095vendor_id (OpenCL::Device self)
744 ALIAS: 1096 ALIAS:
745 vendor_id = CL_DEVICE_VENDOR_ID 1097 vendor_id = CL_DEVICE_VENDOR_ID
756 max_write_image_args = CL_DEVICE_MAX_WRITE_IMAGE_ARGS 1108 max_write_image_args = CL_DEVICE_MAX_WRITE_IMAGE_ARGS
757 image_support = CL_DEVICE_IMAGE_SUPPORT 1109 image_support = CL_DEVICE_IMAGE_SUPPORT
758 max_samplers = CL_DEVICE_MAX_SAMPLERS 1110 max_samplers = CL_DEVICE_MAX_SAMPLERS
759 mem_base_addr_align = CL_DEVICE_MEM_BASE_ADDR_ALIGN 1111 mem_base_addr_align = CL_DEVICE_MEM_BASE_ADDR_ALIGN
760 min_data_type_align_size = CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE 1112 min_data_type_align_size = CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE
1113 global_mem_cache_type = CL_DEVICE_GLOBAL_MEM_CACHE_TYPE
761 global_mem_cacheline_size = CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE 1114 global_mem_cacheline_size = CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE
762 max_constant_args = CL_DEVICE_MAX_CONSTANT_ARGS 1115 max_constant_args = CL_DEVICE_MAX_CONSTANT_ARGS
1116 local_mem_type = CL_DEVICE_LOCAL_MEM_TYPE
763 preferred_vector_width_half = CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF 1117 preferred_vector_width_half = CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF
764 native_vector_width_char = CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR 1118 native_vector_width_char = CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR
765 native_vector_width_short = CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT 1119 native_vector_width_short = CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT
766 native_vector_width_int = CL_DEVICE_NATIVE_VECTOR_WIDTH_INT 1120 native_vector_width_int = CL_DEVICE_NATIVE_VECTOR_WIDTH_INT
767 native_vector_width_long = CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG 1121 native_vector_width_long = CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG
805 EXTEND (SP, n); 1159 EXTEND (SP, n);
806 for (i = 0; i < n; ++i) 1160 for (i = 0; i < n; ++i)
807 PUSHs (sv_2mortal (newSVuv (value [i]))); 1161 PUSHs (sv_2mortal (newSVuv (value [i])));
808 1162
809void 1163void
810address_bits (OpenCL::Device self)
811 PPCODE:
812 cl_bitfield value [1];
813 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_ADDRESS_BITS, sizeof (value), value, 0));
814 EXTEND (SP, 1);
815 const int i = 0;
816 PUSHs (sv_2mortal (newSVuv (value [i])));
817
818void
819max_mem_alloc_size (OpenCL::Device self)
820 ALIAS:
821 max_mem_alloc_size = CL_DEVICE_MAX_MEM_ALLOC_SIZE
822 global_mem_cache_size = CL_DEVICE_GLOBAL_MEM_CACHE_SIZE
823 global_mem_size = CL_DEVICE_GLOBAL_MEM_SIZE
824 max_constant_buffer_size = CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE
825 local_mem_size = CL_DEVICE_LOCAL_MEM_SIZE
826 PPCODE:
827 cl_ulong value [1];
828 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
829 EXTEND (SP, 1);
830 const int i = 0;
831 PUSHs (sv_2mortal (newSVuv (value [i])));
832
833void
834single_fp_config (OpenCL::Device self)
835 ALIAS:
836 single_fp_config = CL_DEVICE_SINGLE_FP_CONFIG
837 double_fp_config = CL_DEVICE_DOUBLE_FP_CONFIG
838 half_fp_config = CL_DEVICE_HALF_FP_CONFIG
839 PPCODE:
840 cl_device_fp_config value [1];
841 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
842 EXTEND (SP, 1);
843 const int i = 0;
844 PUSHs (sv_2mortal (newSVuv (value [i])));
845
846void
847global_mem_cache_type (OpenCL::Device self)
848 PPCODE:
849 cl_device_mem_cache_type value [1];
850 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, sizeof (value), value, 0));
851 EXTEND (SP, 1);
852 const int i = 0;
853 PUSHs (sv_2mortal (newSVuv (value [i])));
854
855void
856local_mem_type (OpenCL::Device self)
857 PPCODE:
858 cl_device_local_mem_type value [1];
859 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_LOCAL_MEM_TYPE, sizeof (value), value, 0));
860 EXTEND (SP, 1);
861 const int i = 0;
862 PUSHs (sv_2mortal (newSVuv (value [i])));
863
864void
865error_correction_support (OpenCL::Device self) 1164error_correction_support (OpenCL::Device self)
866 ALIAS: 1165 ALIAS:
867 error_correction_support = CL_DEVICE_ERROR_CORRECTION_SUPPORT 1166 error_correction_support = CL_DEVICE_ERROR_CORRECTION_SUPPORT
868 endian_little = CL_DEVICE_ENDIAN_LITTLE 1167 endian_little = CL_DEVICE_ENDIAN_LITTLE
869 available = CL_DEVICE_AVAILABLE 1168 available = CL_DEVICE_AVAILABLE
870 compiler_available = CL_DEVICE_COMPILER_AVAILABLE 1169 compiler_available = CL_DEVICE_COMPILER_AVAILABLE
871 host_unified_memory = CL_DEVICE_HOST_UNIFIED_MEMORY 1170 host_unified_memory = CL_DEVICE_HOST_UNIFIED_MEMORY
872 PPCODE: 1171 PPCODE:
873 cl_bool value [1]; 1172 cl_uint value [1];
874 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0)); 1173 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
875 EXTEND (SP, 1); 1174 EXTEND (SP, 1);
876 const int i = 0; 1175 const int i = 0;
877 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no)); 1176 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no));
878
879void
880execution_capabilities (OpenCL::Device self)
881 PPCODE:
882 cl_device_exec_capabilities value [1];
883 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_EXECUTION_CAPABILITIES, sizeof (value), value, 0));
884 EXTEND (SP, 1);
885 const int i = 0;
886 PUSHs (sv_2mortal (newSVuv (value [i])));
887
888void
889properties (OpenCL::Device self)
890 PPCODE:
891 cl_command_queue_properties value [1];
892 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0));
893 EXTEND (SP, 1);
894 const int i = 0;
895 PUSHs (sv_2mortal (newSViv (value [i])));
896 1177
897void 1178void
898platform (OpenCL::Device self) 1179platform (OpenCL::Device self)
899 PPCODE: 1180 PPCODE:
900 cl_platform_id value [1]; 1181 cl_platform_id value [1];
901 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0)); 1182 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0));
902 EXTEND (SP, 1); 1183 EXTEND (SP, 1);
903 const int i = 0; 1184 const int i = 0;
904 {
905 PUSH_CLOBJ ("OpenCL::Platform", value [i]); 1185 PUSH_CLOBJ (stash_platform, value [i]);
906 }
907 1186
908void 1187void
909name (OpenCL::Device self) 1188name (OpenCL::Device self)
910 ALIAS: 1189 ALIAS:
911 name = CL_DEVICE_NAME 1190 name = CL_DEVICE_NAME
928 PPCODE: 1207 PPCODE:
929 cl_device_id value [1]; 1208 cl_device_id value [1];
930 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0)); 1209 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0));
931 EXTEND (SP, 1); 1210 EXTEND (SP, 1);
932 const int i = 0; 1211 const int i = 0;
933 {
934 PUSH_CLOBJ ("OpenCL::Device", value [i]); 1212 PUSH_CLOBJ (stash_device, value [i]);
935 }
936 1213
937void 1214void
938partition_types_ext (OpenCL::Device self) 1215partition_types_ext (OpenCL::Device self)
939 ALIAS: 1216 ALIAS:
940 partition_types_ext = CL_DEVICE_PARTITION_TYPES_EXT 1217 partition_types_ext = CL_DEVICE_PARTITION_TYPES_EXT
950 for (i = 0; i < n; ++i) 1227 for (i = 0; i < n; ++i)
951 PUSHs (sv_2mortal (newSVuv (value [i]))); 1228 PUSHs (sv_2mortal (newSVuv (value [i])));
952 1229
953#END:device 1230#END:device
954 1231
1232MODULE = OpenCL PACKAGE = OpenCL::SubDevice
1233
1234#if CL_VERSION_1_2
1235
1236void
1237DESTROY (OpenCL::SubDevice self)
1238 CODE:
1239 clReleaseDevice (self);
1240
1241#endif
1242
955MODULE = OpenCL PACKAGE = OpenCL::Context 1243MODULE = OpenCL PACKAGE = OpenCL::Context
956 1244
957void 1245void
958DESTROY (OpenCL::Context context) 1246DESTROY (OpenCL::Context self)
959 CODE: 1247 CODE:
960 clReleaseContext (context); 1248 clReleaseContext (self);
961 1249
962void 1250void
963info (OpenCL::Context self, cl_context_info name) 1251info (OpenCL::Context self, cl_context_info name)
964 PPCODE: 1252 PPCODE:
965 INFO (Context) 1253 INFO (Context)
966 1254
967void 1255void
968queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0) 1256queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0)
969 PPCODE: 1257 PPCODE:
970 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res)); 1258 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res));
971 XPUSH_CLOBJ ("OpenCL::Queue", queue); 1259 XPUSH_CLOBJ (stash_queue, queue);
972 1260
973void 1261void
974user_event (OpenCL::Context self) 1262user_event (OpenCL::Context self)
975 PPCODE: 1263 PPCODE:
976 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res)); 1264 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res));
977 XPUSH_CLOBJ ("OpenCL::UserEvent", ev); 1265 XPUSH_CLOBJ (stash_userevent, ev);
978 1266
979void 1267void
980buffer (OpenCL::Context self, cl_mem_flags flags, size_t len) 1268buffer (OpenCL::Context self, cl_mem_flags flags, size_t len)
981 PPCODE: 1269 PPCODE:
982 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) 1270 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))
983 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?"); 1271 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?");
984 1272
985 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res)); 1273 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res));
986 XPUSH_CLOBJ ("OpenCL::BufferObj", mem); 1274 XPUSH_CLOBJ (stash_bufferobj, mem);
987 1275
988void 1276void
989buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data) 1277buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data)
990 PPCODE: 1278 PPCODE:
991 STRLEN len; 1279 STRLEN len;
992 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0; 1280 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
993 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) 1281 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)))
994 croak ("OpenCL::Context::buffer_sv: you have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?"); 1282 croak ("OpenCL::Context::buffer_sv: you have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?");
995 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res)); 1283 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res));
996 XPUSH_CLOBJ ("OpenCL::BufferObj", mem); 1284 XPUSH_CLOBJ (stash_bufferobj, mem);
997 1285
998#if CL_VERSION_1_2 1286#if CL_VERSION_1_2
999 1287
1000void 1288void
1001image (OpenCL::Context self, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, cl_mem_object_type type, size_t width, size_t height, size_t depth = 0, size_t array_size = 0, size_t row_pitch = 0, size_t slice_pitch = 0, cl_uint num_mip_level = 0, cl_uint num_samples = 0, SV *data = &PL_sv_undef) 1289image (OpenCL::Context self, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, cl_mem_object_type type, size_t width, size_t height, size_t depth = 0, size_t array_size = 0, size_t row_pitch = 0, size_t slice_pitch = 0, cl_uint num_mip_level = 0, cl_uint num_samples = 0, SV *data = &PL_sv_undef)
1006 const cl_image_desc desc = { 1294 const cl_image_desc desc = {
1007 type, 1295 type,
1008 width, height, depth, 1296 width, height, depth,
1009 array_size, row_pitch, slice_pitch, 1297 array_size, row_pitch, slice_pitch,
1010 num_mip_level, num_samples, 1298 num_mip_level, num_samples,
1011 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ ("OpenCL::Context::Image", "data", data, "OpenCL::Buffer") : 0 1299 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ (cv, "data", data, "OpenCL::Buffer") : 0
1012 }; 1300 };
1013 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res)); 1301 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1014 char *klass = "OpenCL::Image"; 1302 HV *stash = stash_image;
1015 switch (type) 1303 switch (type)
1016 { 1304 {
1017 case CL_MEM_OBJECT_IMAGE1D_BUFFER: klass = "OpenCL::Image1DBuffer"; break; 1305 case CL_MEM_OBJECT_IMAGE1D_BUFFER: stash = stash_image1dbuffer; break;
1018 case CL_MEM_OBJECT_IMAGE1D: klass = "OpenCL::Image1D"; break; 1306 case CL_MEM_OBJECT_IMAGE1D: stash = stash_image1d; break;
1019 case CL_MEM_OBJECT_IMAGE1D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1307 case CL_MEM_OBJECT_IMAGE1D_ARRAY: stash = stash_image2darray; break;
1020 case CL_MEM_OBJECT_IMAGE2D: klass = "OpenCL::Image2D"; break; 1308 case CL_MEM_OBJECT_IMAGE2D: stash = stash_image2d; break;
1021 case CL_MEM_OBJECT_IMAGE2D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1309 case CL_MEM_OBJECT_IMAGE2D_ARRAY: stash = stash_image2darray; break;
1022 case CL_MEM_OBJECT_IMAGE3D: klass = "OpenCL::Image3D"; break; 1310 case CL_MEM_OBJECT_IMAGE3D: stash = stash_image3d; break;
1023 } 1311 }
1024 XPUSH_CLOBJ (klass, mem); 1312 XPUSH_CLOBJ (stash, mem);
1025 1313
1026#endif 1314#endif
1027 1315
1028void 1316void
1029image2d (OpenCL::Context self, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, size_t width, size_t height, size_t row_pitch = 0, SV *data = &PL_sv_undef) 1317image2d (OpenCL::Context self, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, size_t width, size_t height, size_t row_pitch = 0, SV *data = &PL_sv_undef)
1035 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res)); 1323 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res));
1036#else 1324#else
1037 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, row_pitch, 0, 0, 0, 0 }; 1325 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, row_pitch, 0, 0, 0, 0 };
1038 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res)); 1326 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1039#endif 1327#endif
1040 XPUSH_CLOBJ ("OpenCL::Image2D", mem); 1328 XPUSH_CLOBJ (stash_image2d, mem);
1041 1329
1042void 1330void
1043image3d (OpenCL::Context self, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, size_t width, size_t height, size_t depth, size_t row_pitch = 0, size_t slice_pitch = 0, SV *data = &PL_sv_undef) 1331image3d (OpenCL::Context self, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, size_t width, size_t height, size_t depth, size_t row_pitch = 0, size_t slice_pitch = 0, SV *data = &PL_sv_undef)
1044 PPCODE: 1332 PPCODE:
1045 STRLEN len; 1333 STRLEN len;
1049 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res)); 1337 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res));
1050#else 1338#else
1051 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, row_pitch, slice_pitch, 0, 0, 0 }; 1339 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, row_pitch, slice_pitch, 0, 0, 0 };
1052 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res)); 1340 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1053#endif 1341#endif
1054 XPUSH_CLOBJ ("OpenCL::Image3D", mem); 1342 XPUSH_CLOBJ (stash_image3d, mem);
1055 1343
1056#if cl_apple_gl_sharing || cl_khr_gl_sharing 1344#if cl_apple_gl_sharing || cl_khr_gl_sharing
1057 1345
1058void 1346void
1059gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj) 1347gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj)
1060 PPCODE: 1348 PPCODE:
1061 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res)); 1349 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res));
1062 XPUSH_CLOBJ ("OpenCL::BufferObj", mem); 1350 XPUSH_CLOBJ (stash_bufferobj, mem);
1063 1351
1064void 1352void
1065gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer) 1353gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer)
1066 PPCODE: 1354 PPCODE:
1067 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res)); 1355 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res));
1068 XPUSH_CLOBJ ("OpenCL::Image2D", mem); 1356 XPUSH_CLOBJ (stash_image2d, mem);
1069 1357
1070#if CL_VERSION_1_2 1358#if CL_VERSION_1_2
1071 1359
1072void 1360void
1073gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1361gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1074 ALIAS: 1362 ALIAS:
1075 PPCODE: 1363 PPCODE:
1076 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture, (self, flags, target, miplevel, texture, &res)); 1364 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture, (self, flags, target, miplevel, texture, &res));
1077 cl_gl_object_type type; 1365 cl_gl_object_type type;
1078 NEED_SUCCESS (GetGLObjectInfo, (mem, &type, 0)); // TODO: use target instead? 1366 NEED_SUCCESS (GetGLObjectInfo, (mem, &type, 0)); // TODO: use target instead?
1079 char *klass = "OpenCL::Memory"; 1367 HV *stash = stash_memory;
1080 switch (type) 1368 switch (type)
1081 { 1369 {
1082 case CL_GL_OBJECT_TEXTURE_BUFFER: klass = "OpenCL::Image1DBuffer"; break; 1370 case CL_GL_OBJECT_TEXTURE_BUFFER: stash = stash_image1dbuffer; break;
1083 case CL_GL_OBJECT_TEXTURE1D: klass = "OpenCL::Image1D"; break; 1371 case CL_GL_OBJECT_TEXTURE1D: stash = stash_image1d; break;
1084 case CL_GL_OBJECT_TEXTURE1D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1372 case CL_GL_OBJECT_TEXTURE1D_ARRAY: stash = stash_image2darray; break;
1085 case CL_GL_OBJECT_TEXTURE2D: klass = "OpenCL::Image2D"; break; 1373 case CL_GL_OBJECT_TEXTURE2D: stash = stash_image2d; break;
1086 case CL_GL_OBJECT_TEXTURE2D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1374 case CL_GL_OBJECT_TEXTURE2D_ARRAY: stash = stash_image2darray; break;
1087 case CL_GL_OBJECT_TEXTURE3D: klass = "OpenCL::Image3D"; break; 1375 case CL_GL_OBJECT_TEXTURE3D: stash = stash_image3d; break;
1088 } 1376 }
1089 XPUSH_CLOBJ (klass, mem); 1377 XPUSH_CLOBJ (stash, mem);
1090 1378
1091#endif 1379#endif
1092 1380
1093void 1381void
1094gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1382gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1096#if PREFER_1_1 1384#if PREFER_1_1
1097 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res)); 1385 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res));
1098#else 1386#else
1099 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res)); 1387 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1100#endif 1388#endif
1101 XPUSH_CLOBJ ("OpenCL::Image2D", mem); 1389 XPUSH_CLOBJ (stash_image2d, mem);
1102 1390
1103void 1391void
1104gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1392gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1105 PPCODE: 1393 PPCODE:
1106#if PREFER_1_1 1394#if PREFER_1_1
1107 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res)); 1395 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res));
1108#else 1396#else
1109 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res)); 1397 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1110#endif 1398#endif
1111 XPUSH_CLOBJ ("OpenCL::Image3D", mem); 1399 XPUSH_CLOBJ (stash_image3d, mem);
1112 1400
1113#endif 1401#endif
1114 1402
1115void 1403void
1116supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type) 1404supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type)
1136 1424
1137void 1425void
1138sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode) 1426sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode)
1139 PPCODE: 1427 PPCODE:
1140 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res)); 1428 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res));
1141 XPUSH_CLOBJ ("OpenCL::Sampler", sampler); 1429 XPUSH_CLOBJ (stash_sampler, sampler);
1142 1430
1143void 1431void
1144program_with_source (OpenCL::Context self, SV *program) 1432program_with_source (OpenCL::Context self, SV *program)
1145 PPCODE: 1433 PPCODE:
1146 STRLEN len; 1434 STRLEN len;
1147 size_t len2; 1435 size_t len2;
1148 const char *ptr = SvPVbyte (program, len); 1436 const char *ptr = SvPVbyte (program, len);
1149 1437
1150 len2 = len; 1438 len2 = len;
1151 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res)); 1439 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res));
1152 XPUSH_CLOBJ ("OpenCL::Program", prog); 1440 XPUSH_CLOBJ (stash_program, prog);
1441
1442void
1443program_with_binary (OpenCL::Context self, SV *devices, SV *binaries)
1444 PPCODE:
1445 cl_uint device_count;
1446 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
1447
1448 if (!SvROK (binaries) || SvTYPE (SvRV (binaries)) != SVt_PVAV)
1449 croak ("OpenCL::Context::program_with_binary: binaries must be specified as reference to an array of strings");
1450
1451 binaries = SvRV (binaries);
1452
1453 if (device_count != av_len ((AV *)binaries) + 1)
1454 croak ("OpenCL::Context::program_with_binary: differing numbers of devices and binaries are not allowed");
1455
1456 size_t *length_list = tmpbuf (sizeof (*length_list) * device_count);
1457 const unsigned char **binary_list = tmpbuf (sizeof (*binary_list) * device_count);
1458 cl_int *status_list = tmpbuf (sizeof (*status_list) * device_count);
1459
1460 int i;
1461 for (i = 0; i < device_count; ++i)
1462 {
1463 STRLEN len;
1464 binary_list [i] = (const unsigned char *)SvPVbyte (*av_fetch ((AV *)binaries, i, 0), len);
1465 length_list [i] = len;
1466 }
1467
1468 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithBinary, (self, device_count, device_list,
1469 length_list, binary_list,
1470 GIMME_V == G_ARRAY ? status_list : 0, &res));
1471
1472 EXTEND (SP, 2);
1473 PUSH_CLOBJ (stash_program, prog);
1474
1475 if (GIMME_V == G_ARRAY)
1476 {
1477 AV *av = newAV ();
1478 PUSHs (sv_2mortal (newRV_noinc ((SV *)av)));
1479
1480 for (i = device_count; i--; )
1481 av_store (av, i, newSViv (status_list [i]));
1482 }
1483
1484#if CL_VERSION_1_2
1485
1486void
1487program_with_built_in_kernels (OpenCL::Context self, SV *devices, SV *kernel_names)
1488 PPCODE:
1489 cl_uint device_count;
1490 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
1491
1492 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithBuiltInKernels, (self, device_count, device_list, SvPVbyte_nolen (kernel_names), &res));
1493
1494 XPUSH_CLOBJ (stash_program, prog);
1495
1496void
1497link_program (OpenCL::Context self, SV *devices, SV *options, SV *programs, SV *notify = &PL_sv_undef)
1498 CODE:
1499 cl_uint device_count = 0;
1500 cl_device_id *device_list = 0;
1501
1502 if (SvOK (devices))
1503 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
1504
1505 cl_uint program_count;
1506 cl_program *program_list = object_list (cv, 0, "programs", programs, "OpenCL::Program", &program_count);
1507
1508 void *user_data;
1509 program_callback pfn_notify = make_program_callback (notify, &user_data);
1510
1511 NEED_SUCCESS_ARG (cl_program prog, LinkProgram, (self, device_count, device_list, SvPVbyte_nolen (options),
1512 program_count, program_list, pfn_notify, user_data, &res));
1513
1514 XPUSH_CLOBJ (stash_program, prog);
1515
1516#endif
1153 1517
1154#BEGIN:context 1518#BEGIN:context
1155 1519
1156void 1520void
1157reference_count (OpenCL::Context self) 1521reference_count (OpenCL::Context self)
1173 cl_device_id *value = tmpbuf (size); 1537 cl_device_id *value = tmpbuf (size);
1174 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0)); 1538 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0));
1175 int i, n = size / sizeof (*value); 1539 int i, n = size / sizeof (*value);
1176 EXTEND (SP, n); 1540 EXTEND (SP, n);
1177 for (i = 0; i < n; ++i) 1541 for (i = 0; i < n; ++i)
1178 {
1179 PUSH_CLOBJ ("OpenCL::Device", value [i]); 1542 PUSH_CLOBJ (stash_device, value [i]);
1180 }
1181 1543
1182void 1544void
1183properties (OpenCL::Context self) 1545properties (OpenCL::Context self)
1184 PPCODE: 1546 PPCODE:
1185 size_t size; 1547 size_t size;
1203void 1565void
1204read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...) 1566read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...)
1205 ALIAS: 1567 ALIAS:
1206 enqueue_read_buffer = 0 1568 enqueue_read_buffer = 0
1207 PPCODE: 1569 PPCODE:
1208 cl_event ev = 0;
1209 EVENT_LIST (6, items - 6); 1570 EVENT_LIST (6);
1210 1571
1211 SvUPGRADE (data, SVt_PV); 1572 SvUPGRADE (data, SVt_PV);
1212 SvGROW (data, len); 1573 SvGROW (data, len);
1213 SvPOK_only (data); 1574 SvPOK_only (data);
1214 SvCUR_set (data, len); 1575 SvCUR_set (data, len);
1576
1577 cl_event ev = 0;
1215 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1578 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1216 1579
1217 if (ev) 1580 if (ev)
1218 XPUSH_CLOBJ ("OpenCL::Event", ev); 1581 XPUSH_CLOBJ (stash_event, ev);
1219 1582
1220void 1583void
1221write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...) 1584write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...)
1222 ALIAS: 1585 ALIAS:
1223 enqueue_write_buffer = 0 1586 enqueue_write_buffer = 0
1224 PPCODE: 1587 PPCODE:
1225 cl_event ev = 0; 1588 EVENT_LIST (5);
1589
1226 STRLEN len; 1590 STRLEN len;
1227 char *ptr = SvPVbyte (data, len); 1591 char *ptr = SvPVbyte (data, len);
1228 EVENT_LIST (5, items - 5);
1229 1592
1593 cl_event ev = 0;
1230 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1594 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1231 1595
1232 if (ev) 1596 if (ev)
1233 XPUSH_CLOBJ ("OpenCL::Event", ev); 1597 XPUSH_CLOBJ (stash_event, ev);
1234 1598
1235#if CL_VERSION_1_2 1599#if CL_VERSION_1_2
1236 1600
1237void 1601void
1238fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...) 1602fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...)
1239 ALIAS: 1603 ALIAS:
1240 enqueue_fill_buffer = 0 1604 enqueue_fill_buffer = 0
1241 PPCODE: 1605 PPCODE:
1242 cl_event ev = 0; 1606 EVENT_LIST (5);
1607
1243 STRLEN len; 1608 STRLEN len;
1244 char *ptr = SvPVbyte (data, len); 1609 char *ptr = SvPVbyte (data, len);
1245 EVENT_LIST (5, items - 5);
1246 1610
1611 cl_event ev = 0;
1247 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1612 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1248 1613
1249 if (ev) 1614 if (ev)
1250 XPUSH_CLOBJ ("OpenCL::Event", ev); 1615 XPUSH_CLOBJ (stash_event, ev);
1251 1616
1252void 1617void
1253fill_image (OpenCL::Queue self, OpenCL::Image img, NV r, NV g, NV b, NV a, size_t x, size_t y, size_t z, size_t width, size_t height, size_t depth, ...) 1618fill_image (OpenCL::Queue self, OpenCL::Image img, NV r, NV g, NV b, NV a, size_t x, size_t y, size_t z, size_t width, size_t height, size_t depth, ...)
1254 ALIAS: 1619 ALIAS:
1255 enqueue_fill_image = 0 1620 enqueue_fill_image = 0
1256 PPCODE: 1621 PPCODE:
1257 cl_event ev = 0; 1622 EVENT_LIST (12);
1258 STRLEN len; 1623
1259 const size_t origin [3] = { x, y, z }; 1624 const size_t origin [3] = { x, y, z };
1260 const size_t region [3] = { width, height, depth }; 1625 const size_t region [3] = { width, height, depth };
1261 EVENT_LIST (12, items - 12);
1262 1626
1263 const cl_float c_f [4] = { r, g, b, a }; 1627 const cl_float c_f [4] = { r, g, b, a };
1264 const cl_uint c_u [4] = { r, g, b, a }; 1628 const cl_uint c_u [4] = { r, g, b, a };
1265 const cl_int c_s [4] = { r, g, b, a }; 1629 const cl_int c_s [4] = { r, g, b, a };
1266 const void *c_fus [3] = { &c_f, &c_u, &c_s }; 1630 const void *c_fus [3] = { &c_f, &c_u, &c_s };
1269 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0)); 1633 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
1270 assert (sizeof (fus) == CL_FLOAT + 1 - CL_SNORM_INT8); 1634 assert (sizeof (fus) == CL_FLOAT + 1 - CL_SNORM_INT8);
1271 if (format.image_channel_data_type < CL_SNORM_INT8 || CL_FLOAT < format.image_channel_data_type) 1635 if (format.image_channel_data_type < CL_SNORM_INT8 || CL_FLOAT < format.image_channel_data_type)
1272 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported."); 1636 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported.");
1273 1637
1638 cl_event ev = 0;
1274 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type - CL_SNORM_INT8]], 1639 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type - CL_SNORM_INT8]],
1275 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1640 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1276 1641
1277 if (ev) 1642 if (ev)
1278 XPUSH_CLOBJ ("OpenCL::Event", ev); 1643 XPUSH_CLOBJ (stash_event, ev);
1279 1644
1280#endif 1645#endif
1281 1646
1282void 1647void
1283copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...) 1648copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...)
1284 ALIAS: 1649 ALIAS:
1285 enqueue_copy_buffer = 0 1650 enqueue_copy_buffer = 0
1286 PPCODE: 1651 PPCODE:
1652 EVENT_LIST (6);
1653
1287 cl_event ev = 0; 1654 cl_event ev = 0;
1288 EVENT_LIST (6, items - 6);
1289
1290 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1655 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1291 1656
1292 if (ev) 1657 if (ev)
1293 XPUSH_CLOBJ ("OpenCL::Event", ev); 1658 XPUSH_CLOBJ (stash_event, ev);
1294 1659
1295void 1660void
1296read_buffer_rect (OpenCL::Queue self, OpenCL::Memory buf, cl_bool blocking, size_t buf_x, size_t buf_y, size_t buf_z, size_t host_x, size_t host_y, size_t host_z, size_t width, size_t height, size_t depth, size_t buf_row_pitch, size_t buf_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, SV *data, ...) 1661read_buffer_rect (OpenCL::Queue self, OpenCL::Memory buf, cl_bool blocking, size_t buf_x, size_t buf_y, size_t buf_z, size_t host_x, size_t host_y, size_t host_z, size_t width, size_t height, size_t depth, size_t buf_row_pitch, size_t buf_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, SV *data, ...)
1297 ALIAS: 1662 ALIAS:
1298 enqueue_read_buffer_rect = 0 1663 enqueue_read_buffer_rect = 0
1299 PPCODE: 1664 PPCODE:
1300 cl_event ev = 0; 1665 EVENT_LIST (17);
1666
1301 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1667 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
1302 const size_t host_origin[3] = { host_x, host_y, host_z }; 1668 const size_t host_origin[3] = { host_x, host_y, host_z };
1303 const size_t region[3] = { width, height, depth }; 1669 const size_t region[3] = { width, height, depth };
1304 EVENT_LIST (17, items - 17);
1305 1670
1306 if (!buf_row_pitch) 1671 if (!buf_row_pitch)
1307 buf_row_pitch = region [0]; 1672 buf_row_pitch = region [0];
1308 1673
1309 if (!buf_slice_pitch) 1674 if (!buf_slice_pitch)
1319 1684
1320 SvUPGRADE (data, SVt_PV); 1685 SvUPGRADE (data, SVt_PV);
1321 SvGROW (data, len); 1686 SvGROW (data, len);
1322 SvPOK_only (data); 1687 SvPOK_only (data);
1323 SvCUR_set (data, len); 1688 SvCUR_set (data, len);
1689
1690 cl_event ev = 0;
1324 NEED_SUCCESS (EnqueueReadBufferRect, (self, buf, blocking, buf_origin, host_origin, region, buf_row_pitch, buf_slice_pitch, host_row_pitch, host_slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1691 NEED_SUCCESS (EnqueueReadBufferRect, (self, buf, blocking, buf_origin, host_origin, region, buf_row_pitch, buf_slice_pitch, host_row_pitch, host_slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1325 1692
1326 if (ev) 1693 if (ev)
1327 XPUSH_CLOBJ ("OpenCL::Event", ev); 1694 XPUSH_CLOBJ (stash_event, ev);
1328 1695
1329void 1696void
1330write_buffer_rect (OpenCL::Queue self, OpenCL::Memory buf, cl_bool blocking, size_t buf_x, size_t buf_y, size_t buf_z, size_t host_x, size_t host_y, size_t host_z, size_t width, size_t height, size_t depth, size_t buf_row_pitch, size_t buf_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, SV *data, ...) 1697write_buffer_rect (OpenCL::Queue self, OpenCL::Memory buf, cl_bool blocking, size_t buf_x, size_t buf_y, size_t buf_z, size_t host_x, size_t host_y, size_t host_z, size_t width, size_t height, size_t depth, size_t buf_row_pitch, size_t buf_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, SV *data, ...)
1331 ALIAS: 1698 ALIAS:
1332 enqueue_write_buffer_rect = 0 1699 enqueue_write_buffer_rect = 0
1333 PPCODE: 1700 PPCODE:
1334 cl_event ev = 0; 1701 EVENT_LIST (17);
1702
1335 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1703 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
1336 const size_t host_origin[3] = { host_x, host_y, host_z }; 1704 const size_t host_origin[3] = { host_x, host_y, host_z };
1337 const size_t region[3] = { width, height, depth }; 1705 const size_t region[3] = { width, height, depth };
1338 STRLEN len; 1706 STRLEN len;
1339 char *ptr = SvPVbyte (data, len); 1707 char *ptr = SvPVbyte (data, len);
1340 EVENT_LIST (17, items - 17);
1341 1708
1342 if (!buf_row_pitch) 1709 if (!buf_row_pitch)
1343 buf_row_pitch = region [0]; 1710 buf_row_pitch = region [0];
1344 1711
1345 if (!buf_slice_pitch) 1712 if (!buf_slice_pitch)
1354 size_t min_len = host_row_pitch * host_slice_pitch * region [2]; 1721 size_t min_len = host_row_pitch * host_slice_pitch * region [2];
1355 1722
1356 if (len < min_len) 1723 if (len < min_len)
1357 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1724 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
1358 1725
1726 cl_event ev = 0;
1359 NEED_SUCCESS (EnqueueWriteBufferRect, (self, buf, blocking, buf_origin, host_origin, region, buf_row_pitch, buf_slice_pitch, host_row_pitch, host_slice_pitch, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1727 NEED_SUCCESS (EnqueueWriteBufferRect, (self, buf, blocking, buf_origin, host_origin, region, buf_row_pitch, buf_slice_pitch, host_row_pitch, host_slice_pitch, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1360 1728
1361 if (ev) 1729 if (ev)
1362 XPUSH_CLOBJ ("OpenCL::Event", ev); 1730 XPUSH_CLOBJ (stash_event, ev);
1363 1731
1364void 1732void
1365copy_buffer_rect (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_x, size_t src_y, size_t src_z, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, size_t src_row_pitch, size_t src_slice_pitch, size_t dst_row_pitch, size_t dst_slice_pitch, ...) 1733copy_buffer_rect (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_x, size_t src_y, size_t src_z, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, size_t src_row_pitch, size_t src_slice_pitch, size_t dst_row_pitch, size_t dst_slice_pitch, ...)
1366 ALIAS: 1734 ALIAS:
1367 enqueue_copy_buffer_rect = 0 1735 enqueue_copy_buffer_rect = 0
1368 PPCODE: 1736 PPCODE:
1369 cl_event ev = 0; 1737 EVENT_LIST (16);
1738
1370 const size_t src_origin[3] = { src_x, src_y, src_z }; 1739 const size_t src_origin[3] = { src_x, src_y, src_z };
1371 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1740 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1372 const size_t region[3] = { width, height, depth }; 1741 const size_t region[3] = { width, height, depth };
1373 EVENT_LIST (16, items - 16);
1374 1742
1743 cl_event ev = 0;
1375 NEED_SUCCESS (EnqueueCopyBufferRect, (self, src, dst, src_origin, dst_origin, region, src_row_pitch, src_slice_pitch, dst_row_pitch, dst_slice_pitch, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1744 NEED_SUCCESS (EnqueueCopyBufferRect, (self, src, dst, src_origin, dst_origin, region, src_row_pitch, src_slice_pitch, dst_row_pitch, dst_slice_pitch, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1376 1745
1377 if (ev) 1746 if (ev)
1378 XPUSH_CLOBJ ("OpenCL::Event", ev); 1747 XPUSH_CLOBJ (stash_event, ev);
1379 1748
1380void 1749void
1381read_image (OpenCL::Queue self, OpenCL::Image src, cl_bool blocking, size_t src_x, size_t src_y, size_t src_z, size_t width, size_t height, size_t depth, size_t row_pitch, size_t slice_pitch, SV *data, ...) 1750read_image (OpenCL::Queue self, OpenCL::Image src, cl_bool blocking, size_t src_x, size_t src_y, size_t src_z, size_t width, size_t height, size_t depth, size_t row_pitch, size_t slice_pitch, SV *data, ...)
1382 ALIAS: 1751 ALIAS:
1383 enqueue_read_image = 0 1752 enqueue_read_image = 0
1384 PPCODE: 1753 PPCODE:
1385 cl_event ev = 0; 1754 EVENT_LIST (12);
1755
1386 const size_t src_origin[3] = { src_x, src_y, src_z }; 1756 const size_t src_origin[3] = { src_x, src_y, src_z };
1387 const size_t region[3] = { width, height, depth }; 1757 const size_t region[3] = { width, height, depth };
1388 EVENT_LIST (12, items - 12);
1389 1758
1390 if (!row_pitch) 1759 if (!row_pitch)
1391 row_pitch = img_row_pitch (src); 1760 row_pitch = img_row_pitch (src);
1392 1761
1393 if (depth > 1 && !slice_pitch) 1762 if (depth > 1 && !slice_pitch)
1397 1766
1398 SvUPGRADE (data, SVt_PV); 1767 SvUPGRADE (data, SVt_PV);
1399 SvGROW (data, len); 1768 SvGROW (data, len);
1400 SvPOK_only (data); 1769 SvPOK_only (data);
1401 SvCUR_set (data, len); 1770 SvCUR_set (data, len);
1771
1772 cl_event ev = 0;
1402 NEED_SUCCESS (EnqueueReadImage, (self, src, blocking, src_origin, region, row_pitch, slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1773 NEED_SUCCESS (EnqueueReadImage, (self, src, blocking, src_origin, region, row_pitch, slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1403 1774
1404 if (ev) 1775 if (ev)
1405 XPUSH_CLOBJ ("OpenCL::Event", ev); 1776 XPUSH_CLOBJ (stash_event, ev);
1406 1777
1407void 1778void
1408write_image (OpenCL::Queue self, OpenCL::Image dst, cl_bool blocking, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, size_t row_pitch, size_t slice_pitch, SV *data, ...) 1779write_image (OpenCL::Queue self, OpenCL::Image dst, cl_bool blocking, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, size_t row_pitch, size_t slice_pitch, SV *data, ...)
1409 ALIAS: 1780 ALIAS:
1410 enqueue_write_image = 0 1781 enqueue_write_image = 0
1411 PPCODE: 1782 PPCODE:
1412 cl_event ev = 0; 1783 EVENT_LIST (12);
1784
1413 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1785 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1414 const size_t region[3] = { width, height, depth }; 1786 const size_t region[3] = { width, height, depth };
1415 STRLEN len; 1787 STRLEN len;
1416 char *ptr = SvPVbyte (data, len); 1788 char *ptr = SvPVbyte (data, len);
1417 EVENT_LIST (12, items - 12);
1418 1789
1419 if (!row_pitch) 1790 if (!row_pitch)
1420 row_pitch = img_row_pitch (dst); 1791 row_pitch = img_row_pitch (dst);
1421 1792
1422 if (depth > 1 && !slice_pitch) 1793 if (depth > 1 && !slice_pitch)
1425 size_t min_len = slice_pitch ? slice_pitch * depth : row_pitch * height; 1796 size_t min_len = slice_pitch ? slice_pitch * depth : row_pitch * height;
1426 1797
1427 if (len < min_len) 1798 if (len < min_len)
1428 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1799 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
1429 1800
1801 cl_event ev = 0;
1430 NEED_SUCCESS (EnqueueWriteImage, (self, dst, blocking, dst_origin, region, row_pitch, slice_pitch, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1802 NEED_SUCCESS (EnqueueWriteImage, (self, dst, blocking, dst_origin, region, row_pitch, slice_pitch, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1431 1803
1432 if (ev) 1804 if (ev)
1433 XPUSH_CLOBJ ("OpenCL::Event", ev); 1805 XPUSH_CLOBJ (stash_event, ev);
1434 1806
1435void 1807void
1436copy_image (OpenCL::Queue self, OpenCL::Image src, OpenCL::Image dst, size_t src_x, size_t src_y, size_t src_z, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, ...) 1808copy_image (OpenCL::Queue self, OpenCL::Image src, OpenCL::Image dst, size_t src_x, size_t src_y, size_t src_z, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, ...)
1437 ALIAS: 1809 ALIAS:
1438 enqueue_copy_image = 0 1810 enqueue_copy_image = 0
1439 PPCODE: 1811 PPCODE:
1440 cl_event ev = 0; 1812 EVENT_LIST (12);
1813
1441 const size_t src_origin[3] = { src_x, src_y, src_z }; 1814 const size_t src_origin[3] = { src_x, src_y, src_z };
1442 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1815 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1443 const size_t region[3] = { width, height, depth }; 1816 const size_t region[3] = { width, height, depth };
1444 EVENT_LIST (12, items - 12);
1445 1817
1818 cl_event ev = 0;
1446 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1819 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1447 1820
1448 if (ev) 1821 if (ev)
1449 XPUSH_CLOBJ ("OpenCL::Event", ev); 1822 XPUSH_CLOBJ (stash_event, ev);
1450 1823
1451void 1824void
1452copy_image_to_buffer (OpenCL::Queue self, OpenCL::Image src, OpenCL::Buffer dst, size_t src_x, size_t src_y, size_t src_z, size_t width, size_t height, size_t depth, size_t dst_offset, ...) 1825copy_image_to_buffer (OpenCL::Queue self, OpenCL::Image src, OpenCL::Buffer dst, size_t src_x, size_t src_y, size_t src_z, size_t width, size_t height, size_t depth, size_t dst_offset, ...)
1453 ALIAS: 1826 ALIAS:
1454 enqueue_copy_image_to_buffer = 0 1827 enqueue_copy_image_to_buffer = 0
1455 PPCODE: 1828 PPCODE:
1829 EVENT_LIST (10);
1830
1831 const size_t src_origin[3] = { src_x, src_y, src_z };
1832 const size_t region [3] = { width, height, depth };
1833
1456 cl_event ev = 0; 1834 cl_event ev = 0;
1457 const size_t src_origin[3] = { src_x, src_y, src_z };
1458 const size_t region[3] = { width, height, depth };
1459 EVENT_LIST (10, items - 10);
1460
1461 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1835 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1462 1836
1463 if (ev) 1837 if (ev)
1464 XPUSH_CLOBJ ("OpenCL::Event", ev); 1838 XPUSH_CLOBJ (stash_event, ev);
1465 1839
1466void 1840void
1467copy_buffer_to_image (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Image dst, size_t src_offset, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, ...) 1841copy_buffer_to_image (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Image dst, size_t src_offset, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, ...)
1468 ALIAS: 1842 ALIAS:
1469 enqueue_copy_buffer_to_image = 0 1843 enqueue_copy_buffer_to_image = 0
1470 PPCODE: 1844 PPCODE:
1845 EVENT_LIST (10);
1846
1847 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1848 const size_t region [3] = { width, height, depth };
1849
1471 cl_event ev = 0; 1850 cl_event ev = 0;
1851 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1852
1853 if (ev)
1854 XPUSH_CLOBJ (stash_event, ev);
1855
1856void
1857map_buffer (OpenCL::Queue self, OpenCL::Buffer buf, cl_bool blocking = 1, cl_map_flags map_flags = CL_MAP_READ | CL_MAP_WRITE, size_t offset = 0, SV *cb_ = &PL_sv_undef, ...)
1858 ALIAS:
1859 enqueue_map_buffer = 0
1860 PPCODE:
1861 EVENT_LIST (6);
1862
1863 size_t cb = SvIV (cb_);
1864
1865 if (!SvOK (cb_))
1866 {
1867 NEED_SUCCESS (GetMemObjectInfo, (buf, CL_MEM_SIZE, sizeof (cb), &cb, 0));
1868 cb -= offset;
1869 }
1870
1871 cl_event ev;
1872 NEED_SUCCESS_ARG (void *ptr, EnqueueMapBuffer, (self, buf, blocking, map_flags, offset, cb, event_list_count, event_list_ptr, &ev, &res));
1873 XPUSHs (mapped_new (stash_mappedbuffer, self, buf, map_flags, ptr, cb, ev, 0, 0, 1, cb, 1, 1));
1874
1875void
1876map_image (OpenCL::Queue self, OpenCL::Image img, cl_bool blocking = 1, cl_map_flags map_flags = CL_MAP_READ | CL_MAP_WRITE, size_t x = 0, size_t y = 0, size_t z = 0, SV *width_ = &PL_sv_undef, SV *height_ = &PL_sv_undef, SV *depth_ = &PL_sv_undef, ...)
1877 ALIAS:
1878 enqueue_map_image = 0
1879 PPCODE:
1880 size_t width = SvIV (width_);
1881 if (!SvOK (width_))
1882 {
1883 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_WIDTH, sizeof (width), &width, 0));
1884 width -= x;
1885 }
1886
1887 size_t height = SvIV (width_);
1888 if (!SvOK (height_))
1889 {
1890 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_HEIGHT, sizeof (height), &height, 0));
1891 height -= y;
1892
1893 // stupid opencl returns 0 for depth, but requires 1 for 2d images
1894 if (!height)
1895 height = 1;
1896 }
1897
1898 size_t depth = SvIV (width_);
1899 if (!SvOK (depth_))
1900 {
1901 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_DEPTH, sizeof (depth), &depth, 0));
1902 depth -= z;
1903
1904 // stupid opencl returns 0 for depth, but requires 1 for 2d images
1905 if (!depth)
1906 depth = 1;
1907 }
1908
1472 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1909 const size_t origin[3] = { x, y, z };
1473 const size_t region[3] = { width, height, depth }; 1910 const size_t region[3] = { width, height, depth };
1911 size_t row_pitch, slice_pitch;
1474 EVENT_LIST (10, items - 10); 1912 EVENT_LIST (10);
1475 1913
1476 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1914 cl_event ev;
1915 NEED_SUCCESS_ARG (void *ptr, EnqueueMapImage, (self, img, blocking, map_flags, origin, region, &row_pitch, &slice_pitch, event_list_count, event_list_ptr, &ev, &res));
1477 1916
1478 if (ev) 1917 size_t cb = slice_pitch ? slice_pitch * region [2]
1479 XPUSH_CLOBJ ("OpenCL::Event", ev); 1918 : row_pitch ? row_pitch * region [1]
1919 : region [0];
1920
1921 XPUSHs (mapped_new (stash_mappedimage, self, img, map_flags, ptr, cb, ev, row_pitch, slice_pitch, 0, width, height, depth));
1922
1923void
1924unmap (OpenCL::Queue self, OpenCL::Mapped mapped, ...)
1925 PPCODE:
1926 mapped_unmap (cv, ST (1), mapped, self, &ST (2), items - 2);
1927 if (GIMME_V != G_VOID)
1928 {
1929 clRetainEvent (mapped->event);
1930 XPUSH_CLOBJ (stash_event, mapped->event);
1931 }
1480 1932
1481void 1933void
1482task (OpenCL::Queue self, OpenCL::Kernel kernel, ...) 1934task (OpenCL::Queue self, OpenCL::Kernel kernel, ...)
1483 ALIAS: 1935 ALIAS:
1484 enqueue_task = 0 1936 enqueue_task = 0
1485 PPCODE: 1937 PPCODE:
1938 EVENT_LIST (2);
1939
1486 cl_event ev = 0; 1940 cl_event ev = 0;
1487 EVENT_LIST (2, items - 2);
1488
1489 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1941 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1490 1942
1491 if (ev) 1943 if (ev)
1492 XPUSH_CLOBJ ("OpenCL::Event", ev); 1944 XPUSH_CLOBJ (stash_event, ev);
1493 1945
1494void 1946void
1495nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...) 1947nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...)
1496 ALIAS: 1948 ALIAS:
1497 enqueue_nd_range_kernel = 0 1949 enqueue_nd_range_kernel = 0
1498 PPCODE: 1950 PPCODE:
1499 cl_event ev = 0; 1951 EVENT_LIST (5);
1952
1500 size_t *gwo = 0, *gws, *lws = 0; 1953 size_t *gwo = 0, *gws, *lws = 0;
1501 int gws_len; 1954 int gws_len;
1502 size_t *lists; 1955 size_t *lists;
1503 int i; 1956 int i;
1504 EVENT_LIST (5, items - 5);
1505 1957
1506 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV) 1958 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV)
1507 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference"); 1959 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference");
1508 1960
1509 gws_len = AvFILLp (SvRV (global_work_size)) + 1; 1961 gws_len = AvFILLp (SvRV (global_work_size)) + 1;
1548 if (!lws [i]) 2000 if (!lws [i])
1549 croak ("clEnqueueNDRangeKernel: local_work_size[%d] is zero, must be non-zero", i); 2001 croak ("clEnqueueNDRangeKernel: local_work_size[%d] is zero, must be non-zero", i);
1550 } 2002 }
1551 } 2003 }
1552 2004
2005 cl_event ev = 0;
1553 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 2006 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1554 2007
1555 if (ev) 2008 if (ev)
1556 XPUSH_CLOBJ ("OpenCL::Event", ev); 2009 XPUSH_CLOBJ (stash_event, ev);
2010
2011#if CL_VERSION_1_2
2012
2013void
2014migrate_mem_objects (OpenCL::Queue self, SV *objects, cl_mem_migration_flags flags, ...)
2015 ALIAS:
2016 enqueue_migrate_mem_objects = 0
2017 PPCODE:
2018 EVENT_LIST (3);
2019
2020 cl_uint obj_count;
2021 cl_mem *obj_list = object_list (cv, 0, "objects", objects, "OpenCL::Memory", &obj_count);
2022
2023 cl_event ev = 0;
2024 NEED_SUCCESS (EnqueueMigrateMemObjects, (self, obj_count, obj_list, flags, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2025
2026 if (ev)
2027 XPUSH_CLOBJ (stash_event, ev);
2028
2029#endif
1557 2030
1558#if cl_apple_gl_sharing || cl_khr_gl_sharing 2031#if cl_apple_gl_sharing || cl_khr_gl_sharing
1559 2032
1560void 2033void
1561acquire_gl_objects (OpenCL::Queue self, SV *objects, ...) 2034acquire_gl_objects (OpenCL::Queue self, SV *objects, ...)
1562 ALIAS: 2035 ALIAS:
2036 release_gl_objects = 1
1563 enqueue_acquire_gl_objects = 0 2037 enqueue_acquire_gl_objects = 0
1564 ALIAS:
1565 enqueue_release_gl_objects = 1 2038 enqueue_release_gl_objects = 1
1566 PPCODE: 2039 PPCODE:
1567 if (!SvROK (objects) || SvTYPE (SvRV (objects)) != SVt_PVAV) 2040 EVENT_LIST (2);
1568 croak ("OpenCL::Queue::enqueue_acquire/release_gl_objects argument 'objects' must be an arrayref with memory objects, in call"); 2041
2042 cl_uint obj_count;
2043 cl_mem *obj_list = object_list (cv, 0, "objects", objects, "OpenCL::Memory", &obj_count);
1569 2044
1570 cl_event ev = 0; 2045 cl_event ev = 0;
1571 EVENT_LIST (2, items - 2);
1572 AV *av = (AV *)SvRV (objects);
1573 cl_uint num_objects = av_len (av) + 1;
1574 cl_mem *object_list = tmpbuf (sizeof (cl_mem) * num_objects);
1575 int i;
1576
1577 for (i = num_objects; i--; )
1578 object_list [i] = SvCLOBJ ("OpenCL::Queue::enqueue_acquire/release_gl_objects", "objects", *av_fetch (av, i, 0), "OpenCL::Memory");
1579 2046
1580 if (ix) 2047 if (ix)
1581 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 2048 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, obj_count, obj_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1582 else 2049 else
1583 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 2050 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, obj_count, obj_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1584 2051
1585 if (ev) 2052 if (ev)
1586 XPUSH_CLOBJ ("OpenCL::Event", ev); 2053 XPUSH_CLOBJ (stash_event, ev);
1587 2054
1588#endif 2055#endif
1589 2056
1590void 2057void
1591wait_for_events (OpenCL::Queue self, ...) 2058wait_for_events (OpenCL::Queue self, ...)
1592 ALIAS: 2059 ALIAS:
1593 enqueue_wait_for_events = 0 2060 enqueue_wait_for_events = 0
1594 CODE: 2061 CODE:
1595 EVENT_LIST (1, items - 1); 2062 EVENT_LIST (1);
1596#if PREFER_1_1 2063#if PREFER_1_1
1597 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr)); 2064 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
1598#else 2065#else
1599 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0)); 2066 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0));
1600#endif 2067#endif
1602void 2069void
1603marker (OpenCL::Queue self, ...) 2070marker (OpenCL::Queue self, ...)
1604 ALIAS: 2071 ALIAS:
1605 enqueue_marker = 0 2072 enqueue_marker = 0
1606 PPCODE: 2073 PPCODE:
2074 EVENT_LIST (1);
1607 cl_event ev = 0; 2075 cl_event ev = 0;
1608 EVENT_LIST (1, items - 1);
1609#if PREFER_1_1 2076#if PREFER_1_1
1610 if (!event_list_count) 2077 if (!event_list_count)
1611 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0)); 2078 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
1612 else 2079 else
1613#if CL_VERSION_1_2 2080#if CL_VERSION_1_2
1620#endif 2087#endif
1621#else 2088#else
1622 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 2089 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1623#endif 2090#endif
1624 if (ev) 2091 if (ev)
1625 XPUSH_CLOBJ ("OpenCL::Event", ev); 2092 XPUSH_CLOBJ (stash_event, ev);
1626 2093
1627void 2094void
1628barrier (OpenCL::Queue self, ...) 2095barrier (OpenCL::Queue self, ...)
1629 ALIAS: 2096 ALIAS:
1630 enqueue_barrier = 0 2097 enqueue_barrier = 0
1631 PPCODE: 2098 PPCODE:
2099 EVENT_LIST (1);
1632 cl_event ev = 0; 2100 cl_event ev = 0;
1633 EVENT_LIST (1, items - 1);
1634#if PREFER_1_1 2101#if PREFER_1_1
1635 if (!event_list_count && GIMME_V == G_VOID) 2102 if (!event_list_count && GIMME_V == G_VOID)
1636 NEED_SUCCESS (EnqueueBarrier, (self)); 2103 NEED_SUCCESS (EnqueueBarrier, (self));
1637 else 2104 else
1638#if CL_VERSION_1_2 2105#if CL_VERSION_1_2
1648#endif 2115#endif
1649#else 2116#else
1650 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 2117 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1651#endif 2118#endif
1652 if (ev) 2119 if (ev)
1653 XPUSH_CLOBJ ("OpenCL::Event", ev); 2120 XPUSH_CLOBJ (stash_event, ev);
1654 2121
1655void 2122void
1656flush (OpenCL::Queue self) 2123flush (OpenCL::Queue self)
1657 CODE: 2124 CODE:
1658 NEED_SUCCESS (Flush, (self)); 2125 NEED_SUCCESS (Flush, (self));
1674 PPCODE: 2141 PPCODE:
1675 cl_context value [1]; 2142 cl_context value [1];
1676 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0)); 2143 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0));
1677 EXTEND (SP, 1); 2144 EXTEND (SP, 1);
1678 const int i = 0; 2145 const int i = 0;
1679 {
1680 NEED_SUCCESS (RetainContext, (value [i])); 2146 NEED_SUCCESS (RetainContext, (value [i]));
1681 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2147 PUSH_CLOBJ (stash_context, value [i]);
1682 }
1683 2148
1684void 2149void
1685device (OpenCL::Queue self) 2150device (OpenCL::Queue self)
1686 PPCODE: 2151 PPCODE:
1687 cl_device_id value [1]; 2152 cl_device_id value [1];
1688 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0)); 2153 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0));
1689 EXTEND (SP, 1); 2154 EXTEND (SP, 1);
1690 const int i = 0; 2155 const int i = 0;
1691 {
1692 PUSH_CLOBJ ("OpenCL::Device", value [i]); 2156 PUSH_CLOBJ (stash_device, value [i]);
1693 }
1694 2157
1695void 2158void
1696reference_count (OpenCL::Queue self) 2159reference_count (OpenCL::Queue self)
1697 PPCODE: 2160 PPCODE:
1698 cl_uint value [1]; 2161 cl_uint value [1];
1702 PUSHs (sv_2mortal (newSVuv (value [i]))); 2165 PUSHs (sv_2mortal (newSVuv (value [i])));
1703 2166
1704void 2167void
1705properties (OpenCL::Queue self) 2168properties (OpenCL::Queue self)
1706 PPCODE: 2169 PPCODE:
1707 cl_command_queue_properties value [1]; 2170 cl_ulong value [1];
1708 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_PROPERTIES, sizeof (value), value, 0)); 2171 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_PROPERTIES, sizeof (value), value, 0));
1709 EXTEND (SP, 1); 2172 EXTEND (SP, 1);
1710 const int i = 0; 2173 const int i = 0;
1711 PUSHs (sv_2mortal (newSViv (value [i]))); 2174 PUSHs (sv_2mortal (newSVuv (value [i])));
1712 2175
1713#END:command_queue 2176#END:command_queue
1714 2177
1715MODULE = OpenCL PACKAGE = OpenCL::Memory 2178MODULE = OpenCL PACKAGE = OpenCL::Memory
1716 2179
1722void 2185void
1723info (OpenCL::Memory self, cl_mem_info name) 2186info (OpenCL::Memory self, cl_mem_info name)
1724 PPCODE: 2187 PPCODE:
1725 INFO (MemObject) 2188 INFO (MemObject)
1726 2189
2190void
2191destructor_callback (OpenCL::Memory self, SV *notify)
2192 PPCODE:
2193 clSetMemObjectDestructorCallback (self, eq_destructor_notify, SvREFCNT_inc (s_get_cv (notify)));
2194
1727#BEGIN:mem 2195#BEGIN:mem
1728 2196
1729void 2197void
1730type (OpenCL::Memory self) 2198type (OpenCL::Memory self)
2199 ALIAS:
2200 type = CL_MEM_TYPE
2201 map_count = CL_MEM_MAP_COUNT
2202 reference_count = CL_MEM_REFERENCE_COUNT
1731 PPCODE: 2203 PPCODE:
1732 cl_mem_object_type value [1]; 2204 cl_uint value [1];
1733 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_TYPE, sizeof (value), value, 0)); 2205 NEED_SUCCESS (GetMemObjectInfo, (self, ix, sizeof (value), value, 0));
1734 EXTEND (SP, 1); 2206 EXTEND (SP, 1);
1735 const int i = 0; 2207 const int i = 0;
1736 PUSHs (sv_2mortal (newSViv (value [i]))); 2208 PUSHs (sv_2mortal (newSVuv (value [i])));
1737 2209
1738void 2210void
1739flags (OpenCL::Memory self) 2211flags (OpenCL::Memory self)
1740 PPCODE: 2212 PPCODE:
1741 cl_mem_flags value [1]; 2213 cl_ulong value [1];
1742 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_FLAGS, sizeof (value), value, 0)); 2214 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_FLAGS, sizeof (value), value, 0));
1743 EXTEND (SP, 1); 2215 EXTEND (SP, 1);
1744 const int i = 0; 2216 const int i = 0;
1745 PUSHs (sv_2mortal (newSViv (value [i]))); 2217 PUSHs (sv_2mortal (newSVuv (value [i])));
1746 2218
1747void 2219void
1748size (OpenCL::Memory self) 2220size (OpenCL::Memory self)
1749 ALIAS: 2221 ALIAS:
1750 size = CL_MEM_SIZE 2222 size = CL_MEM_SIZE
1764 EXTEND (SP, 1); 2236 EXTEND (SP, 1);
1765 const int i = 0; 2237 const int i = 0;
1766 PUSHs (sv_2mortal (newSVuv ((IV)(intptr_t)value [i]))); 2238 PUSHs (sv_2mortal (newSVuv ((IV)(intptr_t)value [i])));
1767 2239
1768void 2240void
1769map_count (OpenCL::Memory self)
1770 ALIAS:
1771 map_count = CL_MEM_MAP_COUNT
1772 reference_count = CL_MEM_REFERENCE_COUNT
1773 PPCODE:
1774 cl_uint value [1];
1775 NEED_SUCCESS (GetMemObjectInfo, (self, ix, sizeof (value), value, 0));
1776 EXTEND (SP, 1);
1777 const int i = 0;
1778 PUSHs (sv_2mortal (newSVuv (value [i])));
1779
1780void
1781context (OpenCL::Memory self) 2241context (OpenCL::Memory self)
1782 PPCODE: 2242 PPCODE:
1783 cl_context value [1]; 2243 cl_context value [1];
1784 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0)); 2244 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0));
1785 EXTEND (SP, 1); 2245 EXTEND (SP, 1);
1786 const int i = 0; 2246 const int i = 0;
1787 {
1788 NEED_SUCCESS (RetainContext, (value [i])); 2247 NEED_SUCCESS (RetainContext, (value [i]));
1789 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2248 PUSH_CLOBJ (stash_context, value [i]);
1790 }
1791 2249
1792void 2250void
1793associated_memobject (OpenCL::Memory self) 2251associated_memobject (OpenCL::Memory self)
1794 PPCODE: 2252 PPCODE:
1795 cl_mem value [1]; 2253 cl_mem value [1];
1796 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0)); 2254 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0));
1797 EXTEND (SP, 1); 2255 EXTEND (SP, 1);
1798 const int i = 0; 2256 const int i = 0;
1799 {
1800 NEED_SUCCESS (RetainMemObject, (value [i])); 2257 NEED_SUCCESS (RetainMemObject, (value [i]));
1801 PUSH_CLOBJ ("OpenCL::Memory", value [i]); 2258 PUSH_CLOBJ (stash_memory, value [i]);
1802 }
1803 2259
1804#END:mem 2260#END:mem
1805 2261
1806#if cl_apple_gl_sharing || cl_khr_gl_sharing 2262#if cl_apple_gl_sharing || cl_khr_gl_sharing
1807 2263
1826 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!"); 2282 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!");
1827 2283
1828 cl_buffer_region crdata = { origin, size }; 2284 cl_buffer_region crdata = { origin, size };
1829 2285
1830 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res)); 2286 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res));
1831 XPUSH_CLOBJ ("OpenCL::Buffer", mem); 2287 XPUSH_CLOBJ (stash_buffer, mem);
1832 2288
1833MODULE = OpenCL PACKAGE = OpenCL::Image 2289MODULE = OpenCL PACKAGE = OpenCL::Image
1834 2290
1835void 2291void
1836image_info (OpenCL::Image self, cl_image_info name) 2292image_info (OpenCL::Image self, cl_image_info name)
1906 2362
1907#BEGIN:sampler 2363#BEGIN:sampler
1908 2364
1909void 2365void
1910reference_count (OpenCL::Sampler self) 2366reference_count (OpenCL::Sampler self)
2367 ALIAS:
2368 reference_count = CL_SAMPLER_REFERENCE_COUNT
2369 normalized_coords = CL_SAMPLER_NORMALIZED_COORDS
2370 addressing_mode = CL_SAMPLER_ADDRESSING_MODE
1911 PPCODE: 2371 PPCODE:
1912 cl_uint value [1]; 2372 cl_uint value [1];
1913 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_REFERENCE_COUNT, sizeof (value), value, 0)); 2373 NEED_SUCCESS (GetSamplerInfo, (self, ix, sizeof (value), value, 0));
1914 EXTEND (SP, 1); 2374 EXTEND (SP, 1);
1915 const int i = 0; 2375 const int i = 0;
1916 PUSHs (sv_2mortal (newSVuv (value [i]))); 2376 PUSHs (sv_2mortal (newSVuv (value [i])));
1917 2377
1918void 2378void
1920 PPCODE: 2380 PPCODE:
1921 cl_context value [1]; 2381 cl_context value [1];
1922 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0)); 2382 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0));
1923 EXTEND (SP, 1); 2383 EXTEND (SP, 1);
1924 const int i = 0; 2384 const int i = 0;
1925 {
1926 NEED_SUCCESS (RetainContext, (value [i])); 2385 NEED_SUCCESS (RetainContext, (value [i]));
1927 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2386 PUSH_CLOBJ (stash_context, value [i]);
1928 }
1929
1930void
1931normalized_coords (OpenCL::Sampler self)
1932 PPCODE:
1933 cl_addressing_mode value [1];
1934 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0));
1935 EXTEND (SP, 1);
1936 const int i = 0;
1937 PUSHs (sv_2mortal (newSViv (value [i])));
1938
1939void
1940addressing_mode (OpenCL::Sampler self)
1941 PPCODE:
1942 cl_filter_mode value [1];
1943 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0));
1944 EXTEND (SP, 1);
1945 const int i = 0;
1946 PUSHs (sv_2mortal (newSViv (value [i])));
1947 2387
1948void 2388void
1949filter_mode (OpenCL::Sampler self) 2389filter_mode (OpenCL::Sampler self)
1950 PPCODE: 2390 PPCODE:
1951 cl_bool value [1]; 2391 cl_uint value [1];
1952 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_FILTER_MODE, sizeof (value), value, 0)); 2392 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_FILTER_MODE, sizeof (value), value, 0));
1953 EXTEND (SP, 1); 2393 EXTEND (SP, 1);
1954 const int i = 0; 2394 const int i = 0;
1955 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no)); 2395 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no));
1956 2396
1966void 2406void
1967build (OpenCL::Program self, SV *devices = &PL_sv_undef, SV *options = &PL_sv_undef, SV *notify = &PL_sv_undef) 2407build (OpenCL::Program self, SV *devices = &PL_sv_undef, SV *options = &PL_sv_undef, SV *notify = &PL_sv_undef)
1968 ALIAS: 2408 ALIAS:
1969 build_async = 1 2409 build_async = 1
1970 CODE: 2410 CODE:
1971 void (CL_CALLBACK *pfn_notify)(cl_program program, void *user_data) = 0;
1972 void *user_data = 0;
1973 cl_uint num_devices = 0; 2411 cl_uint device_count = 0;
1974 cl_device_id *device_list = 0; 2412 cl_device_id *device_list = 0;
1975 2413
1976 if (SvOK (devices)) 2414 if (SvOK (devices))
1977 { 2415 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
2416
2417 void *user_data;
2418 program_callback pfn_notify = make_program_callback (notify, &user_data);
2419
2420 if (ix)
2421 build_program_async (self, device_count, device_list, SvPVbyte_nolen (options), user_data);
2422 else
2423 NEED_SUCCESS (BuildProgram, (self, device_count, device_list, SvPVbyte_nolen (options), pfn_notify, user_data));
2424
2425#if CL_VERSION_1_2
2426
2427void
2428compile (OpenCL::Program self, SV *devices, SV *options = &PL_sv_undef, SV *headers = &PL_sv_undef, SV *notify = &PL_sv_undef)
2429 CODE:
2430 cl_uint device_count = 0;
2431 cl_device_id *device_list = 0;
2432
2433 if (SvOK (devices))
2434 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
2435
2436 cl_uint header_count = 0;
2437 cl_program *header_list = 0;
2438 const char **header_name = 0;
2439
2440 if (SvOK (headers))
2441 {
1978 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV) 2442 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVHV)
1979 croak ("clProgramBuild: devices must be undef or an array of OpenCL::Device objects."); 2443 croak ("clCompileProgram: headers must be undef or a hashref of name => OpenCL::Program pairs");
1980 2444
1981 AV *av = (AV *)SvRV (devices); 2445 HV *hv = (HV *)SvRV (devices);
1982 num_devices = av_len (av) + 1;
1983 2446
1984 if (num_devices) 2447 header_count = hv_iterinit (hv);
2448 header_list = tmpbuf (sizeof (*header_list) * header_count);
2449 header_name = tmpbuf (sizeof (*header_name) * header_count);
2450
2451 HE *he;
2452 int i = 0;
2453 while (he = hv_iternext (hv))
1985 { 2454 {
1986 device_list = tmpbuf (sizeof (*device_list) * num_devices); 2455 header_name [i] = SvPVbyte_nolen (HeSVKEY_force (he));
2456 header_list [i] = SvCLOBJ (cv, "headers", HeVAL (he), "OpenCL::Program");
1987 int count; 2457 ++i;
1988 for (count = 0; count < num_devices; ++count)
1989 device_list [count] = SvCLOBJ ("clBuildProgram", "devices", *av_fetch (av, count, 1), "OpenCL::Device");
1990 } 2458 }
1991 } 2459 }
1992 2460
1993 if (SvOK (notify)) 2461 void *user_data;
1994 { 2462 program_callback pfn_notify = make_program_callback (notify, &user_data);
1995 NEED_SUCCESS (RetainProgram, (self));
1996 pfn_notify = eq_program_notify;
1997 user_data = SvREFCNT_inc (s_get_cv (notify));
1998 }
1999 2463
2000 if (ix)
2001 build_program_async (self, num_devices, device_list, SvPVbyte_nolen (options), user_data);
2002 else
2003 NEED_SUCCESS (BuildProgram, (self, num_devices, device_list, SvPVbyte_nolen (options), pfn_notify, user_data)); 2464 NEED_SUCCESS (CompileProgram, (self, device_count, device_list, SvPVbyte_nolen (options),
2465 header_count, header_list, header_name, pfn_notify, user_data));
2466
2467#endif
2004 2468
2005void 2469void
2006build_info (OpenCL::Program self, OpenCL::Device device, cl_program_build_info name) 2470build_info (OpenCL::Program self, OpenCL::Device device, cl_program_build_info name)
2007 PPCODE: 2471 PPCODE:
2008 size_t size; 2472 size_t size;
2017#BEGIN:program_build 2481#BEGIN:program_build
2018 2482
2019void 2483void
2020build_status (OpenCL::Program self, OpenCL::Device device) 2484build_status (OpenCL::Program self, OpenCL::Device device)
2021 PPCODE: 2485 PPCODE:
2022 cl_build_status value [1]; 2486 cl_int value [1];
2023 NEED_SUCCESS (GetProgramBuildInfo, (self, device, CL_PROGRAM_BUILD_STATUS, sizeof (value), value, 0)); 2487 NEED_SUCCESS (GetProgramBuildInfo, (self, device, CL_PROGRAM_BUILD_STATUS, sizeof (value), value, 0));
2024 EXTEND (SP, 1); 2488 EXTEND (SP, 1);
2025 const int i = 0; 2489 const int i = 0;
2026 PUSHs (sv_2mortal (newSViv (value [i]))); 2490 PUSHs (sv_2mortal (newSViv (value [i])));
2027 2491
2037 NEED_SUCCESS (GetProgramBuildInfo, (self, device, ix, size, value, 0)); 2501 NEED_SUCCESS (GetProgramBuildInfo, (self, device, ix, size, value, 0));
2038 EXTEND (SP, 1); 2502 EXTEND (SP, 1);
2039 const int i = 0; 2503 const int i = 0;
2040 PUSHs (sv_2mortal (newSVpv (value, 0))); 2504 PUSHs (sv_2mortal (newSVpv (value, 0)));
2041 2505
2506void
2507binary_type (OpenCL::Program self, OpenCL::Device device)
2508 PPCODE:
2509 cl_uint value [1];
2510 NEED_SUCCESS (GetProgramBuildInfo, (self, device, CL_PROGRAM_BINARY_TYPE, sizeof (value), value, 0));
2511 EXTEND (SP, 1);
2512 const int i = 0;
2513 PUSHs (sv_2mortal (newSVuv ((UV)value [i])));
2514
2042#END:program_build 2515#END:program_build
2043 2516
2044void 2517void
2045kernel (OpenCL::Program program, SV *function) 2518kernel (OpenCL::Program program, SV *function)
2046 PPCODE: 2519 PPCODE:
2047 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res)); 2520 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res));
2048 XPUSH_CLOBJ ("OpenCL::Kernel", kernel); 2521 XPUSH_CLOBJ (stash_kernel, kernel);
2049 2522
2050void 2523void
2051kernels_in_program (OpenCL::Program program) 2524kernels_in_program (OpenCL::Program program)
2052 PPCODE: 2525 PPCODE:
2053 cl_uint num_kernels; 2526 cl_uint num_kernels;
2056 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0)); 2529 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0));
2057 2530
2058 int i; 2531 int i;
2059 EXTEND (SP, num_kernels); 2532 EXTEND (SP, num_kernels);
2060 for (i = 0; i < num_kernels; ++i) 2533 for (i = 0; i < num_kernels; ++i)
2061 PUSH_CLOBJ ("OpenCL::Kernel", kernels [i]); 2534 PUSH_CLOBJ (stash_kernel, kernels [i]);
2062 2535
2063void 2536void
2064info (OpenCL::Program self, cl_program_info name) 2537info (OpenCL::Program self, cl_program_info name)
2065 PPCODE: 2538 PPCODE:
2066 INFO (Program) 2539 INFO (Program)
2112 PPCODE: 2585 PPCODE:
2113 cl_context value [1]; 2586 cl_context value [1];
2114 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0)); 2587 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0));
2115 EXTEND (SP, 1); 2588 EXTEND (SP, 1);
2116 const int i = 0; 2589 const int i = 0;
2117 {
2118 NEED_SUCCESS (RetainContext, (value [i])); 2590 NEED_SUCCESS (RetainContext, (value [i]));
2119 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2591 PUSH_CLOBJ (stash_context, value [i]);
2120 }
2121 2592
2122void 2593void
2123devices (OpenCL::Program self) 2594devices (OpenCL::Program self)
2124 PPCODE: 2595 PPCODE:
2125 size_t size; 2596 size_t size;
2127 cl_device_id *value = tmpbuf (size); 2598 cl_device_id *value = tmpbuf (size);
2128 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0)); 2599 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0));
2129 int i, n = size / sizeof (*value); 2600 int i, n = size / sizeof (*value);
2130 EXTEND (SP, n); 2601 EXTEND (SP, n);
2131 for (i = 0; i < n; ++i) 2602 for (i = 0; i < n; ++i)
2132 {
2133 PUSH_CLOBJ ("OpenCL::Device", value [i]); 2603 PUSH_CLOBJ (stash_device, value [i]);
2134 }
2135 2604
2136void 2605void
2137source (OpenCL::Program self) 2606source (OpenCL::Program self)
2138 PPCODE: 2607 PPCODE:
2139 size_t size; 2608 size_t size;
2212 case 'f': arg.cf = SvNV (sv); size = sizeof (arg.cf); break; 2681 case 'f': arg.cf = SvNV (sv); size = sizeof (arg.cf); break;
2213 case 'd': arg.cd = SvNV (sv); size = sizeof (arg.cd); break; 2682 case 'd': arg.cd = SvNV (sv); size = sizeof (arg.cd); break;
2214 2683
2215 case 'z': nullarg = 1; size = SvIV (sv); break; 2684 case 'z': nullarg = 1; size = SvIV (sv); break;
2216 2685
2217 case 'm': nullarg = !SvOK (sv); arg.cm = SvCLOBJ ("OpenCL::Kernel::setf", "m", sv, "OpenCL::Memory" ); size = sizeof (arg.cm); break; 2686 case 'm': nullarg = !SvOK (sv); arg.cm = SvCLOBJ (cv, "m", sv, "OpenCL::Memory" ); size = sizeof (arg.cm); break;
2218 case 'a': nullarg = !SvOK (sv); arg.ca = SvCLOBJ ("OpenCL::Kernel::setf", "a", sv, "OpenCL::Sampler"); size = sizeof (arg.ca); break; 2687 case 'a': nullarg = !SvOK (sv); arg.ca = SvCLOBJ (cv, "a", sv, "OpenCL::Sampler"); size = sizeof (arg.ca); break;
2219 case 'e': nullarg = !SvOK (sv); arg.ca = SvCLOBJ ("OpenCL::Kernel::setf", "e", sv, "OpenCL::Event" ); size = sizeof (arg.ce); break; 2688 case 'e': nullarg = !SvOK (sv); arg.ca = SvCLOBJ (cv, "e", sv, "OpenCL::Event" ); size = sizeof (arg.ce); break;
2220 2689
2221 default: 2690 default:
2222 croak ("OpenCL::Kernel::setf format character '%c' not supported", type); 2691 croak ("OpenCL::Kernel::setf format character '%c' not supported", type);
2223 } 2692 }
2224 2693
2295 CODE: 2764 CODE:
2296 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0); 2765 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
2297 2766
2298void 2767void
2299set_image (OpenCL::Kernel self, cl_uint idx, OpenCL::Image_ornull value) 2768set_image (OpenCL::Kernel self, cl_uint idx, OpenCL::Image_ornull value)
2300 ALIAS:
2301 set_image2d = 0
2302 set_image3d = 0
2303 CODE: 2769 CODE:
2304 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0); 2770 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
2305 2771
2306void 2772void
2307set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value) 2773set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value)
2353 PPCODE: 2819 PPCODE:
2354 cl_context value [1]; 2820 cl_context value [1];
2355 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0)); 2821 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0));
2356 EXTEND (SP, 1); 2822 EXTEND (SP, 1);
2357 const int i = 0; 2823 const int i = 0;
2358 {
2359 NEED_SUCCESS (RetainContext, (value [i])); 2824 NEED_SUCCESS (RetainContext, (value [i]));
2360 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2825 PUSH_CLOBJ (stash_context, value [i]);
2361 }
2362 2826
2363void 2827void
2364program (OpenCL::Kernel self) 2828program (OpenCL::Kernel self)
2365 PPCODE: 2829 PPCODE:
2366 cl_program value [1]; 2830 cl_program value [1];
2367 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0)); 2831 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0));
2368 EXTEND (SP, 1); 2832 EXTEND (SP, 1);
2369 const int i = 0; 2833 const int i = 0;
2370 {
2371 NEED_SUCCESS (RetainProgram, (value [i])); 2834 NEED_SUCCESS (RetainProgram, (value [i]));
2372 PUSH_CLOBJ ("OpenCL::Program", value [i]); 2835 PUSH_CLOBJ (stash_program, value [i]);
2373 }
2374 2836
2375#END:kernel 2837#END:kernel
2376 2838
2377void 2839void
2378work_group_info (OpenCL::Kernel self, OpenCL::Device device, cl_kernel_work_group_info name) 2840work_group_info (OpenCL::Kernel self, OpenCL::Device device, cl_kernel_work_group_info name)
2424 const int i = 0; 2886 const int i = 0;
2425 PUSHs (sv_2mortal (newSVuv (value [i]))); 2887 PUSHs (sv_2mortal (newSVuv (value [i])));
2426 2888
2427#END:kernel_work_group 2889#END:kernel_work_group
2428 2890
2891#if CL_VERSION_1_2
2892
2893void
2894arg_info (OpenCL::Kernel self, cl_uint idx, cl_kernel_arg_info name)
2895 PPCODE:
2896 size_t size;
2897 NEED_SUCCESS (GetKernelArgInfo, (self, idx, name, 0, 0, &size));
2898 SV *sv = sv_2mortal (newSV (size));
2899 SvUPGRADE (sv, SVt_PV);
2900 SvPOK_only (sv);
2901 SvCUR_set (sv, size);
2902 NEED_SUCCESS (GetKernelArgInfo, (self, idx, name, size, SvPVX (sv), 0));
2903 XPUSHs (sv);
2904
2905#BEGIN:kernel_arg
2906
2907void
2908arg_address_qualifier (OpenCL::Kernel self, cl_uint idx)
2909 ALIAS:
2910 arg_address_qualifier = CL_KERNEL_ARG_ADDRESS_QUALIFIER
2911 arg_access_qualifier = CL_KERNEL_ARG_ACCESS_QUALIFIER
2912 PPCODE:
2913 cl_uint value [1];
2914 NEED_SUCCESS (GetKernelArgInfo, (self, idx, ix, sizeof (value), value, 0));
2915 EXTEND (SP, 1);
2916 const int i = 0;
2917 PUSHs (sv_2mortal (newSVuv (value [i])));
2918
2919void
2920arg_type_name (OpenCL::Kernel self, cl_uint idx)
2921 ALIAS:
2922 arg_type_name = CL_KERNEL_ARG_TYPE_NAME
2923 arg_name = CL_KERNEL_ARG_NAME
2924 PPCODE:
2925 size_t size;
2926 NEED_SUCCESS (GetKernelArgInfo, (self, idx, ix, 0, 0, &size));
2927 char *value = tmpbuf (size);
2928 NEED_SUCCESS (GetKernelArgInfo, (self, idx, ix, size, value, 0));
2929 EXTEND (SP, 1);
2930 const int i = 0;
2931 PUSHs (sv_2mortal (newSVpv (value, 0)));
2932
2933void
2934arg_type_qualifier (OpenCL::Kernel self, cl_uint idx)
2935 PPCODE:
2936 cl_ulong value [1];
2937 NEED_SUCCESS (GetKernelArgInfo, (self, idx, CL_KERNEL_ARG_TYPE_QUALIFIER, sizeof (value), value, 0));
2938 EXTEND (SP, 1);
2939 const int i = 0;
2940 PUSHs (sv_2mortal (newSVuv (value [i])));
2941
2942#END:kernel_arg
2943
2944#endif
2945
2429MODULE = OpenCL PACKAGE = OpenCL::Event 2946MODULE = OpenCL PACKAGE = OpenCL::Event
2430 2947
2431void 2948void
2432DESTROY (OpenCL::Event self) 2949DESTROY (OpenCL::Event self)
2433 CODE: 2950 CODE:
2455 PPCODE: 2972 PPCODE:
2456 cl_command_queue value [1]; 2973 cl_command_queue value [1];
2457 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0)); 2974 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0));
2458 EXTEND (SP, 1); 2975 EXTEND (SP, 1);
2459 const int i = 0; 2976 const int i = 0;
2460 {
2461 NEED_SUCCESS (RetainCommandQueue, (value [i])); 2977 NEED_SUCCESS (RetainCommandQueue, (value [i]));
2462 PUSH_CLOBJ ("OpenCL::Queue", value [i]); 2978 PUSH_CLOBJ (stash_queue, value [i]);
2463 }
2464 2979
2465void 2980void
2466command_type (OpenCL::Event self) 2981command_type (OpenCL::Event self)
2467 PPCODE:
2468 cl_command_type value [1];
2469 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_TYPE, sizeof (value), value, 0));
2470 EXTEND (SP, 1);
2471 const int i = 0;
2472 PUSHs (sv_2mortal (newSVuv (value [i])));
2473
2474void
2475reference_count (OpenCL::Event self)
2476 ALIAS: 2982 ALIAS:
2983 command_type = CL_EVENT_COMMAND_TYPE
2477 reference_count = CL_EVENT_REFERENCE_COUNT 2984 reference_count = CL_EVENT_REFERENCE_COUNT
2478 command_execution_status = CL_EVENT_COMMAND_EXECUTION_STATUS 2985 command_execution_status = CL_EVENT_COMMAND_EXECUTION_STATUS
2479 PPCODE: 2986 PPCODE:
2480 cl_uint value [1]; 2987 cl_uint value [1];
2481 NEED_SUCCESS (GetEventInfo, (self, ix, sizeof (value), value, 0)); 2988 NEED_SUCCESS (GetEventInfo, (self, ix, sizeof (value), value, 0));
2488 PPCODE: 2995 PPCODE:
2489 cl_context value [1]; 2996 cl_context value [1];
2490 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0)); 2997 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0));
2491 EXTEND (SP, 1); 2998 EXTEND (SP, 1);
2492 const int i = 0; 2999 const int i = 0;
2493 {
2494 NEED_SUCCESS (RetainContext, (value [i])); 3000 NEED_SUCCESS (RetainContext, (value [i]));
2495 PUSH_CLOBJ ("OpenCL::Context", value [i]); 3001 PUSH_CLOBJ (stash_context, value [i]);
2496 }
2497 3002
2498#END:event 3003#END:event
2499 3004
2500void 3005void
2501profiling_info (OpenCL::Event self, cl_profiling_info name) 3006profiling_info (OpenCL::Event self, cl_profiling_info name)
2525void 3030void
2526set_status (OpenCL::UserEvent self, cl_int execution_status) 3031set_status (OpenCL::UserEvent self, cl_int execution_status)
2527 CODE: 3032 CODE:
2528 clSetUserEventStatus (self, execution_status); 3033 clSetUserEventStatus (self, execution_status);
2529 3034
3035MODULE = OpenCL PACKAGE = OpenCL::Mapped
3036
3037void
3038DESTROY (SV *self)
3039 CODE:
3040 OpenCL__Mapped mapped = SvMAPPED (self);
3041
3042 clEnqueueUnmapMemObject (mapped->queue, mapped->memobj, mapped->ptr, 1, &mapped->event, 0);
3043 mapped_detach (self, mapped);
3044
3045 clReleaseCommandQueue (mapped->queue);
3046 clReleaseEvent (mapped->event);
3047 Safefree (mapped);
3048
3049void
3050unmap (OpenCL::Mapped self, ...)
3051 CODE:
3052 mapped_unmap (cv, ST (0), self, self->queue, &ST (1), items - 1);
3053
3054bool
3055mapped (OpenCL::Mapped self)
3056 CODE:
3057 RETVAL = !!self->ptr;
3058 OUTPUT:
3059 RETVAL
3060
3061void
3062wait (OpenCL::Mapped self)
3063 PPCODE:
3064 if (self->event)
3065 NEED_SUCCESS (WaitForEvents, (1, &self->event));
3066
3067void
3068event (OpenCL::Mapped self)
3069 PPCODE:
3070 if (!self->event)
3071 XSRETURN_UNDEF;
3072
3073 clRetainEvent (self->event);
3074 XPUSH_CLOBJ (stash_event, self->event);
3075
3076#define MAPPED_OFFSET_CB offsetof (struct mapped, cb)
3077#define MAPPED_OFFSET_ROW_PITCH offsetof (struct mapped, row_pitch)
3078#define MAPPED_OFFSET_SLICE_PITCH offsetof (struct mapped, slice_pitch)
3079#define MAPPED_OFFSET_WIDTH offsetof (struct mapped, width)
3080#define MAPPED_OFFSET_HEIGHT offsetof (struct mapped, height)
3081#define MAPPED_OFFSET_DEPTH offsetof (struct mapped, depth)
3082
3083IV
3084size (OpenCL::Mapped self)
3085 ALIAS:
3086 size = MAPPED_OFFSET_CB
3087 row_pitch = MAPPED_OFFSET_ROW_PITCH
3088 slice_pitch = MAPPED_OFFSET_SLICE_PITCH
3089 width = MAPPED_OFFSET_WIDTH
3090 height = MAPPED_OFFSET_HEIGHT
3091 depth = MAPPED_OFFSET_DEPTH
3092 CODE:
3093 RETVAL = *(size_t *)((char *)self + ix);
3094 OUTPUT:
3095 RETVAL
3096
3097IV
3098ptr (OpenCL::Mapped self)
3099 CODE:
3100 RETVAL = PTR2IV (self->ptr);
3101 OUTPUT:
3102 RETVAL
3103
3104void
3105set (OpenCL::Mapped self, size_t offset, SV *data)
3106 CODE:
3107 STRLEN len;
3108 const char *ptr = SvPVbyte (data, len);
3109
3110 if (offset + len > self->cb)
3111 croak ("OpenCL::Mapped::set out of bound condition detected");
3112
3113 memcpy (offset + (char *)self->ptr, ptr, len);
3114
3115void
3116get_row (OpenCL::Mapped self, size_t count, size_t x = 0, size_t y = 0, size_t z = 0)
3117 PPCODE:
3118 if (!SvOK (ST (1)))
3119 count = self->width - x;
3120
3121 if (x + count > self->width)
3122 croak ("OpenCL::Mapped::get: x + count crosses a row boundary");
3123
3124 if (y >= self->height)
3125 croak ("OpenCL::Mapped::get: y coordinate out of bounds");
3126
3127 if (z >= self->depth)
3128 croak ("OpenCL::Mapped::get: z coordinate out of bounds");
3129
3130 size_t element = mapped_element_size (self);
3131
3132 count *= element;
3133 x *= element;
3134
3135 char *ptr = (char *)self->ptr + x + y * self->row_pitch + z * self->slice_pitch;
3136 XPUSHs (sv_2mortal (newSVpvn (ptr, count)));
3137
3138void
3139set_row (OpenCL::Mapped self, SV *data, size_t x = 0, size_t y = 0, size_t z = 0)
3140 PPCODE:
3141 STRLEN count;
3142 char *dataptr = SvPVbyte (data, count);
3143 size_t element = mapped_element_size (self);
3144
3145 x *= element;
3146
3147 if (x + count > self->width * element)
3148 croak ("OpenCL::Mapped::set: x + data size crosses a row boundary");
3149
3150 if (y >= self->height)
3151 croak ("OpenCL::Mapped::set: y coordinate out of bounds");
3152
3153 if (z >= self->depth)
3154 croak ("OpenCL::Mapped::set: z coordinate out of bounds");
3155
3156 char *ptr = (char *)self->ptr + x + y * self->row_pitch + z * self->slice_pitch;
3157 memcpy (ptr, dataptr, count);
3158
3159MODULE = OpenCL PACKAGE = OpenCL::MappedBuffer
3160
3161MODULE = OpenCL PACKAGE = OpenCL::MappedImage
3162
3163IV
3164element_size (OpenCL::Mapped self)
3165 CODE:
3166 RETVAL = mapped_element_size (self);
3167 OUTPUT:
3168 RETVAL
3169

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines