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

Comparing OpenCL/OpenCL.xs (file contents):
Revision 1.53 by root, Wed Apr 25 00:02:25 2012 UTC vs.
Revision 1.64 by root, Thu May 3 23:30:08 2012 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4
5#include "ecb.h"//D
4 6
5#define X_STACKSIZE sizeof (void *) * 512 * 1024 // 2-4mb should be enough, really 7#define X_STACKSIZE sizeof (void *) * 512 * 1024 // 2-4mb should be enough, really
6#include "xthread.h" 8#include "xthread.h"
7#include "schmorp.h" 9#include "schmorp.h"
8 10
44typedef cl_command_queue OpenCL__Queue; 46typedef cl_command_queue OpenCL__Queue;
45typedef cl_mem OpenCL__Memory; 47typedef cl_mem OpenCL__Memory;
46typedef cl_mem OpenCL__Buffer; 48typedef cl_mem OpenCL__Buffer;
47typedef cl_mem OpenCL__BufferObj; 49typedef cl_mem OpenCL__BufferObj;
48typedef cl_mem OpenCL__Image; 50typedef cl_mem OpenCL__Image;
49typedef cl_mem OpenCL__Image2D;
50typedef cl_mem OpenCL__Image3D;
51typedef cl_mem OpenCL__Memory_ornull; 51typedef cl_mem OpenCL__Memory_ornull;
52typedef cl_mem OpenCL__Buffer_ornull; 52typedef cl_mem OpenCL__Buffer_ornull;
53typedef cl_mem OpenCL__Image_ornull; 53typedef cl_mem OpenCL__Image_ornull;
54typedef cl_mem OpenCL__Image2D_ornull;
55typedef cl_mem OpenCL__Image3D_ornull;
56typedef cl_sampler OpenCL__Sampler; 54typedef cl_sampler OpenCL__Sampler;
57typedef cl_program OpenCL__Program; 55typedef cl_program OpenCL__Program;
58typedef cl_kernel OpenCL__Kernel; 56typedef cl_kernel OpenCL__Kernel;
59typedef cl_event OpenCL__Event; 57typedef cl_event OpenCL__Event;
60typedef cl_event OpenCL__UserEvent; 58typedef cl_event OpenCL__UserEvent;
61 59
60typedef struct mapped * OpenCL__Mapped;
61
62typedef SV *FUTURE; 62typedef SV *FUTURE;
63
64static HV
65 *stash_platform,
66 *stash_device,
67 *stash_context,
68 *stash_queue,
69 *stash_program,
70 *stash_kernel,
71 *stash_sampler,
72 *stash_event,
73 *stash_userevent,
74 *stash_memory,
75 *stash_buffer,
76 *stash_bufferobj,
77 *stash_image,
78 *stash_image1d,
79 *stash_image1darray,
80 *stash_image1dbuffer,
81 *stash_image2d,
82 *stash_image2darray,
83 *stash_image3d,
84 *stash_mapped,
85 *stash_mappedbuffer,
86 *stash_mappedimage;
63 87
64/*****************************************************************************/ 88/*****************************************************************************/
65 89
66// name must include a leading underscore 90// name must include a leading underscore
67// all of this horrors would be unneceesary if somebody wrote a proper OpenGL module 91// all of this horrors would be unneceesary if somebody wrote a proper OpenGL module
107} 131}
108 132
109/*****************************************************************************/ 133/*****************************************************************************/
110 134
111/* up to two temporary buffers */ 135/* up to two temporary buffers */
112static void * 136static void * ecb_noinline
113tmpbuf (size_t size) 137tmpbuf (size_t size)
114{ 138{
115 enum { buffers = 3 }; 139 enum { buffers = 4 };
116 static int idx; 140 static int idx;
117 static void *buf [buffers]; 141 static void *buf [buffers];
118 static size_t len [buffers]; 142 static size_t len [buffers];
119 143
120 idx = (idx + 1) % buffers; 144 idx = (idx + 1) % buffers;
193 if (res) \ 217 if (res) \
194 FAIL (name); 218 FAIL (name);
195 219
196/*****************************************************************************/ 220/*****************************************************************************/
197 221
198static cl_context_properties * 222static cl_context_properties * ecb_noinline
199SvCONTEXTPROPERTIES (const char *func, const char *svname, SV *sv, cl_context_properties *extra, int extracount) 223SvCONTEXTPROPERTIES (const char *func, const char *svname, SV *sv, cl_context_properties *extra, int extracount)
200{ 224{
201 if (!sv || !SvOK (sv)) 225 if (!sv || !SvOK (sv))
202 if (extra) 226 if (extra)
203 sv = sv_2mortal (newRV_noinc ((SV *)newAV ())); // slow, but rarely used hopefully 227 sv = sv_2mortal (newRV_noinc ((SV *)newAV ())); // slow, but rarely used hopefully
260 croak ("%s: %s is not a property list (either undef or [type => value, ...])", func, svname); 284 croak ("%s: %s is not a property list (either undef or [type => value, ...])", func, svname);
261} 285}
262 286
263/*****************************************************************************/ 287/*****************************************************************************/
264 288
265#define NEW_CLOBJ(class,ptr) sv_setref_pv (sv_newmortal (), class, ptr) 289static SV *
266#define PUSH_CLOBJ(class,ptr) PUSHs (NEW_CLOBJ (class, ptr)) 290new_clobj (HV *stash, IV id)
267#define XPUSH_CLOBJ(class,ptr) XPUSHs (NEW_CLOBJ (class, ptr)) 291{
292 return sv_2mortal (sv_bless (newRV_noinc (newSViv (id)), stash));
293}
294
295#define PUSH_CLOBJ(stash,id) PUSHs (new_clobj ((stash), (IV)(id)))
296#define XPUSH_CLOBJ(stash,id) XPUSHs (new_clobj ((stash), (IV)(id)))
268 297
269/* cl objects are either \$iv, or [$iv, ...] */ 298/* cl objects are either \$iv, or [$iv, ...] */
270/* they can be upgraded at runtime to the array form */ 299/* they can be upgraded at runtime to the array form */
271static void * 300static void * ecb_noinline
272SvCLOBJ (const char *func, const char *svname, SV *sv, const char *pkg) 301SvCLOBJ (const char *func, const char *svname, SV *sv, const char *pkg)
273{ 302{
303 // sv_derived_from is quite slow :(
274 if (SvROK (sv) && sv_derived_from (sv, pkg)) 304 if (SvROK (sv) && sv_derived_from (sv, pkg))
275 return (void *)SvIV (SvRV (sv)); 305 return (void *)SvIV (SvRV (sv));
276 306
277 croak ("%s: %s is not of type %s", func, svname, pkg); 307 croak ("%s: %s is not of type %s", func, svname, pkg);
308}
309
310// the "no-inherit" version of the above
311static void * ecb_noinline
312SvCLOBJ_ni (const char *func, const char *svname, SV *sv, HV *stash)
313{
314 if (SvROK (sv) && SvSTASH (SvRV (sv)) == stash)
315 return (void *)SvIV (SvRV (sv));
316
317 croak ("%s: %s is not of type %s", func, svname, HvNAME (stash));
278} 318}
279 319
280/*****************************************************************************/ 320/*****************************************************************************/
281/* callback stuff */ 321/* callback stuff */
282 322
304static void (*eq_signal_func)(void *signal_arg, int value); 344static void (*eq_signal_func)(void *signal_arg, int value);
305static void *eq_signal_arg; 345static void *eq_signal_arg;
306static xmutex_t eq_lock = X_MUTEX_INIT; 346static xmutex_t eq_lock = X_MUTEX_INIT;
307static eq_item *eq_head, *eq_tail; 347static eq_item *eq_head, *eq_tail;
308 348
309static void 349static void ecb_noinline
310eq_enq (eq_vtbl *vtbl, SV *cb, void *data1, void *data2, void *data3) 350eq_enq (eq_vtbl *vtbl, SV *cb, void *data1, void *data2, void *data3)
311{ 351{
312 eq_item *item = malloc (sizeof (eq_item)); 352 eq_item *item = malloc (sizeof (eq_item));
313 353
314 item->next = 0; 354 item->next = 0;
381} 421}
382 422
383/*****************************************************************************/ 423/*****************************************************************************/
384/* context notify */ 424/* context notify */
385 425
386static void 426static void ecb_noinline
387eq_context_push (void *data1, void *data2, void *data3) 427eq_context_push (void *data1, void *data2, void *data3)
388{ 428{
389 dSP; 429 dSP;
390 PUSHs (sv_2mortal (newSVpv (data1, 0))); 430 PUSHs (sv_2mortal (newSVpv (data1, 0)));
391 PUSHs (sv_2mortal (newSVpvn (data2, (STRLEN)data3))); 431 PUSHs (sv_2mortal (newSVpvn (data2, (STRLEN)data3)));
413 { \ 453 { \
414 pfn_notify = eq_context_notify; \ 454 pfn_notify = eq_context_notify; \
415 user_data = s_get_cv (notify); \ 455 user_data = s_get_cv (notify); \
416 } 456 }
417 457
418static SV * 458static SV * ecb_noinline
419new_clobj_context (cl_context ctx, void *user_data) 459new_clobj_context (cl_context ctx, void *user_data)
420{ 460{
421 SV *sv = NEW_CLOBJ ("OpenCL::Context", ctx); 461 SV *sv = new_clobj (stash_context, (IV)ctx);
422 462
423 if (user_data) 463 if (user_data)
424 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0); 464 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0);
425 465
426 return sv; 466 return sv;
433 473
434static void 474static void
435eq_program_push (void *data1, void *data2, void *data3) 475eq_program_push (void *data1, void *data2, void *data3)
436{ 476{
437 dSP; 477 dSP;
438 PUSH_CLOBJ ("OpenCL::Program", data1); 478 PUSH_CLOBJ (stash_program, data1);
439 PUTBACK; 479 PUTBACK;
440} 480}
441 481
442static eq_vtbl eq_program_vtbl = { 1, eq_program_push }; 482static eq_vtbl eq_program_vtbl = { 1, eq_program_push };
443 483
466 else 506 else
467 clReleaseProgram (arg->program); 507 clReleaseProgram (arg->program);
468 508
469 free (arg->options); 509 free (arg->options);
470 free (arg); 510 free (arg);
511
512 return 0;
471} 513}
472 514
473static void 515static void
474build_program_async (cl_program program, cl_uint num_devices, const cl_device_id *device_list, const char *options, void *user_data) 516build_program_async (cl_program program, cl_uint num_devices, const cl_device_id *device_list, const char *options, void *user_data)
475{ 517{
490 532
491static void 533static void
492eq_event_push (void *data1, void *data2, void *data3) 534eq_event_push (void *data1, void *data2, void *data3)
493{ 535{
494 dSP; 536 dSP;
495 PUSH_CLOBJ ("OpenCL::Event", data1); 537 PUSH_CLOBJ (stash_event, data1);
496 PUSHs (sv_2mortal (newSViv ((IV)data2))); 538 PUSHs (sv_2mortal (newSViv ((IV)data2)));
497 PUTBACK; 539 PUTBACK;
498} 540}
499 541
500static eq_vtbl eq_event_vtbl = { 1, eq_event_push }; 542static eq_vtbl eq_event_vtbl = { 1, eq_event_push };
505 clRetainEvent (event); 547 clRetainEvent (event);
506 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0); 548 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0);
507} 549}
508 550
509/*****************************************************************************/ 551/*****************************************************************************/
552/* utilities for XS code */
510 553
511static size_t 554static size_t
512img_row_pitch (cl_mem img) 555img_row_pitch (cl_mem img)
513{ 556{
514 size_t res; 557 size_t res;
515 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0); 558 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0);
516 return res; 559 return res;
517} 560}
518 561
519static cl_event * 562static cl_event * ecb_noinline
520event_list (SV **items, cl_uint *rcount) 563event_list (SV **items, cl_uint *rcount, cl_event extra)
521{ 564{
522 cl_uint count = *rcount; 565 cl_uint count = *rcount;
523 566
524 if (!count) 567 if (count > 0x7fffffffU) // yeah, it's a hack - the caller might have underflowed
568 *rcount = count = 0;
569
570 if (!count && !extra)
525 return 0; 571 return 0;
526 572
527 cl_event *list = tmpbuf (sizeof (cl_event) * count); 573 cl_event *list = tmpbuf (sizeof (cl_event) * (count + 1));
528 int i = 0; 574 int i = 0;
529 575
530 do 576 while (count--)
531 {
532 --count;
533 if (SvOK (items [count])) 577 if (SvOK (items [count]))
534 list [i++] = SvCLOBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event"); 578 list [i++] = SvCLOBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event");
535 } 579
536 while (count); 580 if (extra)
581 list [i++] = extra;
537 582
538 *rcount = i; 583 *rcount = i;
539 584
540 return i ? list : 0; 585 return i ? list : 0;
541} 586}
542 587
543#define EVENT_LIST(items,count) \ 588#define EVENT_LIST(skip) \
544 cl_uint event_list_count = (count); \ 589 cl_uint event_list_count = items - (skip); \
545 cl_event *event_list_ptr = event_list (&ST (items), &event_list_count) 590 cl_event *event_list_ptr = event_list (&ST (skip), &event_list_count, 0)
546 591
547#define INFO(class) \ 592#define INFO(class) \
548{ \ 593{ \
549 size_t size; \ 594 size_t size; \
550 NEED_SUCCESS (Get ## class ## Info, (self, name, 0, 0, &size)); \ 595 NEED_SUCCESS (Get ## class ## Info, (self, name, 0, 0, &size)); \
553 SvPOK_only (sv); \ 598 SvPOK_only (sv); \
554 SvCUR_set (sv, size); \ 599 SvCUR_set (sv, size); \
555 NEED_SUCCESS (Get ## class ## Info, (self, name, size, SvPVX (sv), 0)); \ 600 NEED_SUCCESS (Get ## class ## Info, (self, name, size, SvPVX (sv), 0)); \
556 XPUSHs (sv); \ 601 XPUSHs (sv); \
557} 602}
603
604/*****************************************************************************/
605/* mapped_xxx */
606
607static OpenCL__Mapped
608SvMAPPED (SV *sv)
609{
610 // no typechecking atm., keep your fingers crossed
611 return (OpenCL__Mapped)SvMAGIC (SvRV (sv))->mg_ptr;
612}
613
614struct mapped
615{
616 cl_command_queue queue;
617 cl_mem memobj;
618 void *ptr;
619 size_t cb;
620 cl_event event;
621 size_t row_pitch;
622 size_t slice_pitch;
623};
624
625static SV *
626mapped_new (HV *stash, cl_command_queue queue, cl_mem memobj, cl_map_flags flags, void *ptr, size_t cb, cl_event ev, size_t row_pitch, size_t slice_pitch)
627{
628 SV *data = newSV (0);
629 SvUPGRADE (data, SVt_PVMG);
630
631 OpenCL__Mapped mapped;
632 New (0, mapped, 1, struct mapped);
633
634 clRetainCommandQueue (queue);
635
636 mapped->queue = queue;
637 mapped->memobj = memobj;
638 mapped->ptr = ptr;
639 mapped->cb = cb;
640 mapped->event = ev;
641 mapped->row_pitch = row_pitch;
642 mapped->slice_pitch = slice_pitch;
643
644 sv_magicext (data, 0, PERL_MAGIC_ext, 0, (char *)mapped, 0);
645
646 if (SvLEN (data))
647 Safefree (data);
648
649 SvPVX (data) = (char *)ptr;
650 SvCUR_set (data, cb);
651 SvLEN_set (data, 0);
652 SvPOK_only (data);
653
654 SV *obj = sv_2mortal (sv_bless (newRV_noinc (data), stash));
655
656 if (!(flags & CL_MAP_WRITE))
657 SvREADONLY_on (data);
658
659 return obj;
660}
661
662static void
663mapped_detach (SV *sv, OpenCL__Mapped mapped)
664{
665 SV *data = SvRV (sv);
666
667 // the next check checks both before AND after detach, where SvPVX should STILL be 0
668 if (SvPVX (data) != (char *)mapped->ptr)
669 warn ("FATAL: OpenCL memory mapped scalar changed location, detected");
670 else
671 {
672 SvREADONLY_off (data);
673 SvCUR_set (data, 0);
674 SvPVX (data) = 0;
675 SvOK_off (data);
676 }
677
678 mapped->ptr = 0;
679}
680
681static void
682mapped_unmap (SV *self, OpenCL__Mapped mapped, cl_command_queue queue, SV **wait_list, cl_uint event_list_count)
683{
684 cl_event *event_list_ptr = event_list (wait_list, &event_list_count, mapped->event);
685 cl_event ev;
686
687 NEED_SUCCESS (EnqueueUnmapMemObject, (queue, mapped->memobj, mapped->ptr, event_list_count, event_list_ptr, &ev));
688
689 clReleaseEvent (mapped->event);
690 mapped->event = ev;
691
692 mapped_detach (self, mapped);
693}
694
695/*****************************************************************************/
558 696
559MODULE = OpenCL PACKAGE = OpenCL 697MODULE = OpenCL PACKAGE = OpenCL
560 698
561PROTOTYPES: ENABLE 699PROTOTYPES: ENABLE
562 700
590 }; 728 };
591 729
592 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--) 730 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)); 731 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv));
594 732
733 stash_platform = gv_stashpv ("OpenCL::Platform", GV_ADD);
734 stash_device = gv_stashpv ("OpenCL::Device", GV_ADD);
735 stash_context = gv_stashpv ("OpenCL::Context", GV_ADD);
736 stash_queue = gv_stashpv ("OpenCL::Queue", GV_ADD);
737 stash_program = gv_stashpv ("OpenCL::Program", GV_ADD);
738 stash_kernel = gv_stashpv ("OpenCL::Kernel", GV_ADD);
739 stash_sampler = gv_stashpv ("OpenCL::Sampler", GV_ADD);
740 stash_event = gv_stashpv ("OpenCL::Event", GV_ADD);
741 stash_userevent = gv_stashpv ("OpenCL::UserEvent", GV_ADD);
742 stash_memory = gv_stashpv ("OpenCL::Memory", GV_ADD);
743 stash_buffer = gv_stashpv ("OpenCL::Buffer", GV_ADD);
744 stash_bufferobj = gv_stashpv ("OpenCL::BufferObj", GV_ADD);
745 stash_image = gv_stashpv ("OpenCL::Image", GV_ADD);
746 stash_image1d = gv_stashpv ("OpenCL::Image1D", GV_ADD);
747 stash_image1darray = gv_stashpv ("OpenCL::Image1DArray", GV_ADD);
748 stash_image1dbuffer = gv_stashpv ("OpenCL::Image1DBuffer", GV_ADD);
749 stash_image2d = gv_stashpv ("OpenCL::Image2D", GV_ADD);
750 stash_image2darray = gv_stashpv ("OpenCL::Image2DArray", GV_ADD);
751 stash_image3d = gv_stashpv ("OpenCL::Image3D", GV_ADD);
752 stash_mapped = gv_stashpv ("OpenCL::Mapped", GV_ADD);
753 stash_mappedbuffer = gv_stashpv ("OpenCL::MappedBuffer", GV_ADD);
754 stash_mappedimage = gv_stashpv ("OpenCL::MappedImage", GV_ADD);
755
595 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt); 756 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt);
596} 757}
597 758
598cl_int 759cl_int
599errno () 760errno ()
601 RETVAL = res; 762 RETVAL = res;
602 OUTPUT: 763 OUTPUT:
603 RETVAL 764 RETVAL
604 765
605const char * 766const char *
606err2str (cl_int err) 767err2str (cl_int err = res)
607 768
608const char * 769const char *
609enum2str (cl_uint value) 770enum2str (cl_uint value)
610 771
611void 772void
619 list = tmpbuf (sizeof (*list) * count); 780 list = tmpbuf (sizeof (*list) * count);
620 NEED_SUCCESS (GetPlatformIDs, (count, list, 0)); 781 NEED_SUCCESS (GetPlatformIDs, (count, list, 0));
621 782
622 EXTEND (SP, count); 783 EXTEND (SP, count);
623 for (i = 0; i < count; ++i) 784 for (i = 0; i < count; ++i)
624 PUSH_CLOBJ ("OpenCL::Platform", list [i]); 785 PUSH_CLOBJ (stash_platform, (IV)list [i]);
625 786
626void 787void
627context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef) 788context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
628 PPCODE: 789 PPCODE:
629 CONTEXT_NOTIFY_CALLBACK; 790 CONTEXT_NOTIFY_CALLBACK;
630 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, 0, 0, &res)); 791 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, pfn_notify, user_data, &res));
631 XPUSH_CLOBJ_CONTEXT; 792 XPUSH_CLOBJ_CONTEXT;
632 793
633void 794void
634context (FUTURE properties, FUTURE devices, FUTURE notify) 795context (FUTURE properties, FUTURE devices, FUTURE notify)
635 PPCODE: 796 PPCODE:
636 /* der Gipfel der Kunst */ 797 /* der Gipfel der Kunst */
637 798
638void 799void
639wait_for_events (...) 800wait_for_events (...)
640 CODE: 801 CODE:
641 EVENT_LIST (0, items); 802 EVENT_LIST (0);
642 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr)); 803 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr));
643 804
644PROTOTYPES: DISABLE 805PROTOTYPES: DISABLE
645 806
646MODULE = OpenCL PACKAGE = OpenCL::Platform 807MODULE = OpenCL PACKAGE = OpenCL::Platform
689 list = tmpbuf (sizeof (*list) * count); 850 list = tmpbuf (sizeof (*list) * count);
690 NEED_SUCCESS (GetDeviceIDs, (self, type, count, list, 0)); 851 NEED_SUCCESS (GetDeviceIDs, (self, type, count, list, 0));
691 852
692 EXTEND (SP, count); 853 EXTEND (SP, count);
693 for (i = 0; i < count; ++i) 854 for (i = 0; i < count; ++i)
694 PUSHs (sv_setref_pv (sv_newmortal (), "OpenCL::Device", list [i])); 855 PUSH_CLOBJ (stash_device, list [i]);
695 856
696void 857void
697context (OpenCL::Platform self, cl_context_properties *properties, SV *devices, SV *notify = &PL_sv_undef) 858context (OpenCL::Platform self, SV *properties = 0, SV *devices, SV *notify = &PL_sv_undef)
698 PPCODE: 859 PPCODE:
699 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV) 860 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV)
700 croak ("OpenCL::Platform::context argument 'device' must be an arrayref with device objects, in call"); 861 croak ("OpenCL::Platform::context argument 'device' must be an arrayref with device objects, in call");
862
863 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
864 cl_context_properties *props = SvCONTEXTPROPERTIES ("OpenCL::Platform::context", "properties", properties, extra, 2);
701 865
702 AV *av = (AV *)SvRV (devices); 866 AV *av = (AV *)SvRV (devices);
703 cl_uint num_devices = av_len (av) + 1; 867 cl_uint num_devices = av_len (av) + 1;
704 cl_device_id *device_list = tmpbuf (sizeof (cl_device_id) * num_devices); 868 cl_device_id *device_list = tmpbuf (sizeof (cl_device_id) * num_devices);
705 869
706 int i; 870 int i;
707 for (i = num_devices; i--; ) 871 for (i = num_devices; i--; )
708 device_list [i] = SvCLOBJ ("clCreateContext", "devices", *av_fetch (av, i, 0), "OpenCL::Device"); 872 device_list [i] = SvCLOBJ ("clCreateContext", "devices", *av_fetch (av, i, 0), "OpenCL::Device");
709 873
710 CONTEXT_NOTIFY_CALLBACK; 874 CONTEXT_NOTIFY_CALLBACK;
711 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (properties, num_devices, device_list, pfn_notify, user_data, &res)); 875 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (props, num_devices, device_list, pfn_notify, user_data, &res));
712 XPUSH_CLOBJ_CONTEXT; 876 XPUSH_CLOBJ_CONTEXT;
713 877
714void 878void
715context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef) 879context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
716 PPCODE: 880 PPCODE:
717 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self }; 881 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
718 cl_context_properties *props = SvCONTEXTPROPERTIES ("OpenCL::Platform::context_from_type", "properties", properties, extra, 2); 882 cl_context_properties *props = SvCONTEXTPROPERTIES ("OpenCL::Platform::context_from_type", "properties", properties, extra, 2);
719 883
720 CONTEXT_NOTIFY_CALLBACK; 884 CONTEXT_NOTIFY_CALLBACK;
721 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, 0, 0, &res)); 885 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, pfn_notify, user_data, &res));
722 XPUSH_CLOBJ_CONTEXT; 886 XPUSH_CLOBJ_CONTEXT;
723 887
724MODULE = OpenCL PACKAGE = OpenCL::Device 888MODULE = OpenCL PACKAGE = OpenCL::Device
725 889
726void 890void
735 PPCODE: 899 PPCODE:
736 cl_device_type value [1]; 900 cl_device_type value [1];
737 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_TYPE, sizeof (value), value, 0)); 901 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_TYPE, sizeof (value), value, 0));
738 EXTEND (SP, 1); 902 EXTEND (SP, 1);
739 const int i = 0; 903 const int i = 0;
740 PUSHs (sv_2mortal (newSViv (value [i]))); 904 PUSHs (sv_2mortal (newSVuv (value [i])));
741 905
742void 906void
743vendor_id (OpenCL::Device self) 907vendor_id (OpenCL::Device self)
744 ALIAS: 908 ALIAS:
745 vendor_id = CL_DEVICE_VENDOR_ID 909 vendor_id = CL_DEVICE_VENDOR_ID
890 PPCODE: 1054 PPCODE:
891 cl_command_queue_properties value [1]; 1055 cl_command_queue_properties value [1];
892 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0)); 1056 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0));
893 EXTEND (SP, 1); 1057 EXTEND (SP, 1);
894 const int i = 0; 1058 const int i = 0;
895 PUSHs (sv_2mortal (newSViv (value [i]))); 1059 PUSHs (sv_2mortal (newSVuv (value [i])));
896 1060
897void 1061void
898platform (OpenCL::Device self) 1062platform (OpenCL::Device self)
899 PPCODE: 1063 PPCODE:
900 cl_platform_id value [1]; 1064 cl_platform_id value [1];
901 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0)); 1065 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0));
902 EXTEND (SP, 1); 1066 EXTEND (SP, 1);
903 const int i = 0; 1067 const int i = 0;
904 { 1068 {
905 PUSH_CLOBJ ("OpenCL::Platform", value [i]); 1069 PUSH_CLOBJ (stash_platform, value [i]);
906 } 1070 }
907 1071
908void 1072void
909name (OpenCL::Device self) 1073name (OpenCL::Device self)
910 ALIAS: 1074 ALIAS:
929 cl_device_id value [1]; 1093 cl_device_id value [1];
930 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0)); 1094 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0));
931 EXTEND (SP, 1); 1095 EXTEND (SP, 1);
932 const int i = 0; 1096 const int i = 0;
933 { 1097 {
934 PUSH_CLOBJ ("OpenCL::Device", value [i]); 1098 PUSH_CLOBJ (stash_device, value [i]);
935 } 1099 }
936 1100
937void 1101void
938partition_types_ext (OpenCL::Device self) 1102partition_types_ext (OpenCL::Device self)
939 ALIAS: 1103 ALIAS:
966 1130
967void 1131void
968queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0) 1132queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0)
969 PPCODE: 1133 PPCODE:
970 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res)); 1134 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res));
971 XPUSH_CLOBJ ("OpenCL::Queue", queue); 1135 XPUSH_CLOBJ (stash_queue, queue);
972 1136
973void 1137void
974user_event (OpenCL::Context self) 1138user_event (OpenCL::Context self)
975 PPCODE: 1139 PPCODE:
976 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res)); 1140 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res));
977 XPUSH_CLOBJ ("OpenCL::UserEvent", ev); 1141 XPUSH_CLOBJ (stash_userevent, ev);
978 1142
979void 1143void
980buffer (OpenCL::Context self, cl_mem_flags flags, size_t len) 1144buffer (OpenCL::Context self, cl_mem_flags flags, size_t len)
981 PPCODE: 1145 PPCODE:
982 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) 1146 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?"); 1147 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?");
984 1148
985 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res)); 1149 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res));
986 XPUSH_CLOBJ ("OpenCL::BufferObj", mem); 1150 XPUSH_CLOBJ (stash_bufferobj, mem);
987 1151
988void 1152void
989buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data) 1153buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data)
990 PPCODE: 1154 PPCODE:
991 STRLEN len; 1155 STRLEN len;
992 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0; 1156 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
993 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) 1157 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?"); 1158 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)); 1159 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res));
996 XPUSH_CLOBJ ("OpenCL::BufferObj", mem); 1160 XPUSH_CLOBJ (stash_bufferobj, mem);
997 1161
998#if CL_VERSION_1_2 1162#if CL_VERSION_1_2
999 1163
1000void 1164void
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, 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) 1165image (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)
1002 PPCODE: 1166 PPCODE:
1003 STRLEN len; 1167 STRLEN len;
1004 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0; 1168 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
1005 const cl_image_format format = { channel_order, channel_type }; 1169 const cl_image_format format = { channel_order, channel_type };
1006 const cl_image_desc desc = { 1170 const cl_image_desc desc = {
1009 array_size, row_pitch, slice_pitch, 1173 array_size, row_pitch, slice_pitch,
1010 num_mip_level, num_samples, 1174 num_mip_level, num_samples,
1011 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ ("OpenCL::Context::Image", "data", data, "OpenCL::Buffer") : 0 1175 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ ("OpenCL::Context::Image", "data", data, "OpenCL::Buffer") : 0
1012 }; 1176 };
1013 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res)); 1177 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1014 char *klass = "OpenCL::Image"; 1178 HV *stash = stash_image;
1015 switch (type) 1179 switch (type)
1016 { 1180 {
1017 case CL_MEM_OBJECT_IMAGE1D_BUFFER: klass = "OpenCL::Image1DBuffer"; break; 1181 case CL_MEM_OBJECT_IMAGE1D_BUFFER: stash = stash_image1dbuffer; break;
1018 case CL_MEM_OBJECT_IMAGE1D: klass = "OpenCL::Image1D"; break; 1182 case CL_MEM_OBJECT_IMAGE1D: stash = stash_image1d; break;
1019 case CL_MEM_OBJECT_IMAGE1D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1183 case CL_MEM_OBJECT_IMAGE1D_ARRAY: stash = stash_image2darray; break;
1020 case CL_MEM_OBJECT_IMAGE2D: klass = "OpenCL::Image2D"; break; 1184 case CL_MEM_OBJECT_IMAGE2D: stash = stash_image2d; break;
1021 case CL_MEM_OBJECT_IMAGE2D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1185 case CL_MEM_OBJECT_IMAGE2D_ARRAY: stash = stash_image2darray; break;
1022 case CL_MEM_OBJECT_IMAGE3D: klass = "OpenCL::Image3D"; break; 1186 case CL_MEM_OBJECT_IMAGE3D: stash = stash_image3d; break;
1023 } 1187 }
1024 XPUSH_CLOBJ (klass, mem); 1188 XPUSH_CLOBJ (stash, mem);
1025 1189
1026#endif 1190#endif
1027 1191
1028void 1192void
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) 1193image2d (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)); 1199 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res));
1036#else 1200#else
1037 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, row_pitch, 0, 0, 0, 0 }; 1201 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)); 1202 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1039#endif 1203#endif
1040 XPUSH_CLOBJ ("OpenCL::Image2D", mem); 1204 XPUSH_CLOBJ (stash_image2d, mem);
1041 1205
1042void 1206void
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) 1207image3d (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: 1208 PPCODE:
1045 STRLEN len; 1209 STRLEN len;
1049 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res)); 1213 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res));
1050#else 1214#else
1051 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, row_pitch, slice_pitch, 0, 0, 0 }; 1215 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)); 1216 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1053#endif 1217#endif
1054 XPUSH_CLOBJ ("OpenCL::Image3D", mem); 1218 XPUSH_CLOBJ (stash_image3d, mem);
1055 1219
1056#if cl_apple_gl_sharing || cl_khr_gl_sharing 1220#if cl_apple_gl_sharing || cl_khr_gl_sharing
1057 1221
1058void 1222void
1059gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj) 1223gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj)
1060 PPCODE: 1224 PPCODE:
1061 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res)); 1225 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res));
1062 XPUSH_CLOBJ ("OpenCL::BufferObj", mem); 1226 XPUSH_CLOBJ (stash_bufferobj, mem);
1063 1227
1064void 1228void
1065gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer) 1229gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer)
1066 PPCODE: 1230 PPCODE:
1067 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res)); 1231 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res));
1068 XPUSH_CLOBJ ("OpenCL::Image2D", mem); 1232 XPUSH_CLOBJ (stash_image2d, mem);
1069 1233
1070#if CL_VERSION_1_2 1234#if CL_VERSION_1_2
1071 1235
1072void 1236void
1073gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1237gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1074 ALIAS: 1238 ALIAS:
1075 PPCODE: 1239 PPCODE:
1076 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture, (self, flags, target, miplevel, texture, &res)); 1240 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture, (self, flags, target, miplevel, texture, &res));
1077 cl_gl_object_type type; 1241 cl_gl_object_type type;
1078 NEED_SUCCESS (GetGLObjectInfo, (mem, &type, 0)); // TODO: use target instead? 1242 NEED_SUCCESS (GetGLObjectInfo, (mem, &type, 0)); // TODO: use target instead?
1079 char *klass = "OpenCL::Memory"; 1243 HV *stash = stash_memory;
1080 switch (type) 1244 switch (type)
1081 { 1245 {
1082 case CL_GL_OBJECT_TEXTURE_BUFFER: klass = "OpenCL::Image1DBuffer"; break; 1246 case CL_GL_OBJECT_TEXTURE_BUFFER: stash = stash_image1dbuffer; break;
1083 case CL_GL_OBJECT_TEXTURE1D: klass = "OpenCL::Image1D"; break; 1247 case CL_GL_OBJECT_TEXTURE1D: stash = stash_image1d; break;
1084 case CL_GL_OBJECT_TEXTURE1D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1248 case CL_GL_OBJECT_TEXTURE1D_ARRAY: stash = stash_image2darray; break;
1085 case CL_GL_OBJECT_TEXTURE2D: klass = "OpenCL::Image2D"; break; 1249 case CL_GL_OBJECT_TEXTURE2D: stash = stash_image2d; break;
1086 case CL_GL_OBJECT_TEXTURE2D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1250 case CL_GL_OBJECT_TEXTURE2D_ARRAY: stash = stash_image2darray; break;
1087 case CL_GL_OBJECT_TEXTURE3D: klass = "OpenCL::Image3D"; break; 1251 case CL_GL_OBJECT_TEXTURE3D: stash = stash_image3d; break;
1088 } 1252 }
1089 XPUSH_CLOBJ (klass, mem); 1253 XPUSH_CLOBJ (stash, mem);
1090 1254
1091#endif 1255#endif
1092 1256
1093void 1257void
1094gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1258gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1096#if PREFER_1_1 1260#if PREFER_1_1
1097 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res)); 1261 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res));
1098#else 1262#else
1099 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res)); 1263 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1100#endif 1264#endif
1101 XPUSH_CLOBJ ("OpenCL::Image2D", mem); 1265 XPUSH_CLOBJ (stash_image2d, mem);
1102 1266
1103void 1267void
1104gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1268gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1105 PPCODE: 1269 PPCODE:
1106#if PREFER_1_1 1270#if PREFER_1_1
1107 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res)); 1271 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res));
1108#else 1272#else
1109 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res)); 1273 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1110#endif 1274#endif
1111 XPUSH_CLOBJ ("OpenCL::Image3D", mem); 1275 XPUSH_CLOBJ (stash_image3d, mem);
1112 1276
1113#endif 1277#endif
1114 1278
1115void 1279void
1116supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type) 1280supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type)
1136 1300
1137void 1301void
1138sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode) 1302sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode)
1139 PPCODE: 1303 PPCODE:
1140 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res)); 1304 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res));
1141 XPUSH_CLOBJ ("OpenCL::Sampler", sampler); 1305 XPUSH_CLOBJ (stash_sampler, sampler);
1142 1306
1143void 1307void
1144program_with_source (OpenCL::Context self, SV *program) 1308program_with_source (OpenCL::Context self, SV *program)
1145 PPCODE: 1309 PPCODE:
1146 STRLEN len; 1310 STRLEN len;
1147 size_t len2; 1311 size_t len2;
1148 const char *ptr = SvPVbyte (program, len); 1312 const char *ptr = SvPVbyte (program, len);
1149 1313
1150 len2 = len; 1314 len2 = len;
1151 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res)); 1315 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res));
1152 XPUSH_CLOBJ ("OpenCL::Program", prog); 1316 XPUSH_CLOBJ (stash_program, prog);
1317
1318void
1319program_with_binary (OpenCL::Context self, SV *devices, SV *binaries)
1320 PPCODE:
1321 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV)
1322 croak ("OpenCL::Context::program_with_binary: devices must be specified as reference to an array of device objects");
1323
1324 devices = SvRV (devices);
1325
1326 if (!SvROK (binaries) || SvTYPE (SvRV (binaries)) != SVt_PVAV)
1327 croak ("OpenCL::Context::program_with_binary: binaries must be specified as reference to an array of strings");
1328
1329 binaries = SvRV (binaries);
1330
1331 if (av_len ((AV *)devices) != av_len ((AV *)binaries))
1332 croak ("OpenCL::Context::program_with_binary: differing numbers of devices and binaries are not allowed");
1333
1334 int count = av_len ((AV *)devices) + 1;
1335 cl_device_id *device_list = tmpbuf (sizeof (*device_list) * count);
1336 size_t *length_list = tmpbuf (sizeof (*length_list) * count);
1337 const unsigned char **binary_list = tmpbuf (sizeof (*binary_list) * count);
1338 cl_int *status_list = tmpbuf (sizeof (*status_list) * count);
1339
1340 int i;
1341 for (i = 0; i < count; ++i)
1342 {
1343 device_list [i] = SvCLOBJ ("OpenCL::Context::program_with_binary", "devices", *av_fetch ((AV *)devices, i, 0), "OpenCL::Device");
1344 STRLEN len;
1345 binary_list [i] = (const unsigned char *)SvPVbyte (*av_fetch ((AV *)binaries, i, 0), len);
1346 length_list [i] = len;
1347 }
1348
1349 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithBinary, (self, count, device_list, length_list, binary_list, GIMME_V == G_ARRAY ? status_list : 0, &res));
1350
1351 EXTEND (SP, 2);
1352 PUSH_CLOBJ (stash_program, prog);
1353
1354 if (GIMME_V == G_ARRAY)
1355 {
1356 AV *av = newAV ();
1357 PUSHs (sv_2mortal (newRV_noinc ((SV *)av)));
1358
1359 for (i = count; i--; )
1360 av_store (av, i, newSViv (status_list [i]));
1361 }
1153 1362
1154#BEGIN:context 1363#BEGIN:context
1155 1364
1156void 1365void
1157reference_count (OpenCL::Context self) 1366reference_count (OpenCL::Context self)
1174 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0)); 1383 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0));
1175 int i, n = size / sizeof (*value); 1384 int i, n = size / sizeof (*value);
1176 EXTEND (SP, n); 1385 EXTEND (SP, n);
1177 for (i = 0; i < n; ++i) 1386 for (i = 0; i < n; ++i)
1178 { 1387 {
1179 PUSH_CLOBJ ("OpenCL::Device", value [i]); 1388 PUSH_CLOBJ (stash_device, value [i]);
1180 } 1389 }
1181 1390
1182void 1391void
1183properties (OpenCL::Context self) 1392properties (OpenCL::Context self)
1184 PPCODE: 1393 PPCODE:
1199DESTROY (OpenCL::Queue self) 1408DESTROY (OpenCL::Queue self)
1200 CODE: 1409 CODE:
1201 clReleaseCommandQueue (self); 1410 clReleaseCommandQueue (self);
1202 1411
1203void 1412void
1204enqueue_read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...) 1413read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...)
1414 ALIAS:
1415 enqueue_read_buffer = 0
1205 PPCODE: 1416 PPCODE:
1206 cl_event ev = 0; 1417 cl_event ev = 0;
1207 EVENT_LIST (6, items - 6); 1418 EVENT_LIST (6);
1208 1419
1209 SvUPGRADE (data, SVt_PV); 1420 SvUPGRADE (data, SVt_PV);
1210 SvGROW (data, len); 1421 SvGROW (data, len);
1211 SvPOK_only (data); 1422 SvPOK_only (data);
1212 SvCUR_set (data, len); 1423 SvCUR_set (data, len);
1213 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1424 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1214 1425
1215 if (ev) 1426 if (ev)
1216 XPUSH_CLOBJ ("OpenCL::Event", ev); 1427 XPUSH_CLOBJ (stash_event, ev);
1217 1428
1218void 1429void
1219enqueue_write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...) 1430write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...)
1431 ALIAS:
1432 enqueue_write_buffer = 0
1220 PPCODE: 1433 PPCODE:
1221 cl_event ev = 0; 1434 cl_event ev = 0;
1222 STRLEN len; 1435 STRLEN len;
1223 char *ptr = SvPVbyte (data, len); 1436 char *ptr = SvPVbyte (data, len);
1224 EVENT_LIST (5, items - 5); 1437 EVENT_LIST (5);
1225 1438
1226 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1439 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1227 1440
1228 if (ev) 1441 if (ev)
1229 XPUSH_CLOBJ ("OpenCL::Event", ev); 1442 XPUSH_CLOBJ (stash_event, ev);
1230 1443
1231#if CL_VERSION_1_2 1444#if CL_VERSION_1_2
1232 1445
1233void 1446void
1234enqueue_fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...) 1447fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...)
1448 ALIAS:
1449 enqueue_fill_buffer = 0
1235 PPCODE: 1450 PPCODE:
1236 cl_event ev = 0; 1451 cl_event ev = 0;
1237 STRLEN len; 1452 STRLEN len;
1238 char *ptr = SvPVbyte (data, len); 1453 char *ptr = SvPVbyte (data, len);
1239 EVENT_LIST (5, items - 5); 1454 EVENT_LIST (5);
1240 1455
1241 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1456 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1242 1457
1243 if (ev) 1458 if (ev)
1244 XPUSH_CLOBJ ("OpenCL::Event", ev); 1459 XPUSH_CLOBJ (stash_event, ev);
1245 1460
1246void 1461void
1247enqueue_fill_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, ...) 1462fill_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, ...)
1463 ALIAS:
1464 enqueue_fill_image = 0
1248 PPCODE: 1465 PPCODE:
1249 cl_event ev = 0; 1466 cl_event ev = 0;
1250 STRLEN len;
1251 const size_t origin [3] = { x, y, z }; 1467 const size_t origin [3] = { x, y, z };
1252 const size_t region [3] = { width, height, depth }; 1468 const size_t region [3] = { width, height, depth };
1253 EVENT_LIST (12, items - 12); 1469 EVENT_LIST (12);
1254 1470
1255 const cl_float c_f [4] = { r, g, b, a }; 1471 const cl_float c_f [4] = { r, g, b, a };
1256 const cl_uint c_u [4] = { r, g, b, a }; 1472 const cl_uint c_u [4] = { r, g, b, a };
1257 const cl_int c_s [4] = { r, g, b, a }; 1473 const cl_int c_s [4] = { r, g, b, a };
1258 const void *c_fus [3] = { &c_f, &c_u, &c_s }; 1474 const void *c_fus [3] = { &c_f, &c_u, &c_s };
1261 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0)); 1477 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
1262 assert (sizeof (fus) == CL_FLOAT + 1 - CL_SNORM_INT8); 1478 assert (sizeof (fus) == CL_FLOAT + 1 - CL_SNORM_INT8);
1263 if (format.image_channel_data_type < CL_SNORM_INT8 || CL_FLOAT < format.image_channel_data_type) 1479 if (format.image_channel_data_type < CL_SNORM_INT8 || CL_FLOAT < format.image_channel_data_type)
1264 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported."); 1480 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported.");
1265 1481
1266 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type]], 1482 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type - CL_SNORM_INT8]],
1267 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1483 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1268 1484
1269 if (ev) 1485 if (ev)
1270 XPUSH_CLOBJ ("OpenCL::Event", ev); 1486 XPUSH_CLOBJ (stash_event, ev);
1271 1487
1272#endif 1488#endif
1273 1489
1274void 1490void
1275enqueue_copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...) 1491copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...)
1492 ALIAS:
1493 enqueue_copy_buffer = 0
1276 PPCODE: 1494 PPCODE:
1277 cl_event ev = 0; 1495 cl_event ev = 0;
1278 EVENT_LIST (6, items - 6); 1496 EVENT_LIST (6);
1279 1497
1280 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1498 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1281 1499
1282 if (ev) 1500 if (ev)
1283 XPUSH_CLOBJ ("OpenCL::Event", ev); 1501 XPUSH_CLOBJ (stash_event, ev);
1284 1502
1285void 1503void
1286enqueue_read_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, ...) 1504read_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, ...)
1505 ALIAS:
1506 enqueue_read_buffer_rect = 0
1287 PPCODE: 1507 PPCODE:
1288 cl_event ev = 0; 1508 cl_event ev = 0;
1289 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1509 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
1290 const size_t host_origin[3] = { host_x, host_y, host_z }; 1510 const size_t host_origin[3] = { host_x, host_y, host_z };
1291 const size_t region[3] = { width, height, depth }; 1511 const size_t region[3] = { width, height, depth };
1292 EVENT_LIST (17, items - 17); 1512 EVENT_LIST (17);
1293 1513
1294 if (!buf_row_pitch) 1514 if (!buf_row_pitch)
1295 buf_row_pitch = region [0]; 1515 buf_row_pitch = region [0];
1296 1516
1297 if (!buf_slice_pitch) 1517 if (!buf_slice_pitch)
1310 SvPOK_only (data); 1530 SvPOK_only (data);
1311 SvCUR_set (data, len); 1531 SvCUR_set (data, len);
1312 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)); 1532 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));
1313 1533
1314 if (ev) 1534 if (ev)
1315 XPUSH_CLOBJ ("OpenCL::Event", ev); 1535 XPUSH_CLOBJ (stash_event, ev);
1316 1536
1317void 1537void
1318enqueue_write_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, ...) 1538write_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, ...)
1539 ALIAS:
1540 enqueue_write_buffer_rect = 0
1319 PPCODE: 1541 PPCODE:
1320 cl_event ev = 0; 1542 cl_event ev = 0;
1321 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1543 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
1322 const size_t host_origin[3] = { host_x, host_y, host_z }; 1544 const size_t host_origin[3] = { host_x, host_y, host_z };
1323 const size_t region[3] = { width, height, depth }; 1545 const size_t region[3] = { width, height, depth };
1324 STRLEN len; 1546 STRLEN len;
1325 char *ptr = SvPVbyte (data, len); 1547 char *ptr = SvPVbyte (data, len);
1326 EVENT_LIST (17, items - 17); 1548 EVENT_LIST (17);
1327 1549
1328 if (!buf_row_pitch) 1550 if (!buf_row_pitch)
1329 buf_row_pitch = region [0]; 1551 buf_row_pitch = region [0];
1330 1552
1331 if (!buf_slice_pitch) 1553 if (!buf_slice_pitch)
1343 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1565 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
1344 1566
1345 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)); 1567 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));
1346 1568
1347 if (ev) 1569 if (ev)
1348 XPUSH_CLOBJ ("OpenCL::Event", ev); 1570 XPUSH_CLOBJ (stash_event, ev);
1349 1571
1350void 1572void
1351enqueue_copy_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, ...) 1573copy_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, ...)
1574 ALIAS:
1575 enqueue_copy_buffer_rect = 0
1352 PPCODE: 1576 PPCODE:
1353 cl_event ev = 0; 1577 cl_event ev = 0;
1354 const size_t src_origin[3] = { src_x, src_y, src_z }; 1578 const size_t src_origin[3] = { src_x, src_y, src_z };
1355 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1579 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1356 const size_t region[3] = { width, height, depth }; 1580 const size_t region[3] = { width, height, depth };
1357 EVENT_LIST (16, items - 16); 1581 EVENT_LIST (16);
1358 1582
1359 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)); 1583 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));
1360 1584
1361 if (ev) 1585 if (ev)
1362 XPUSH_CLOBJ ("OpenCL::Event", ev); 1586 XPUSH_CLOBJ (stash_event, ev);
1363 1587
1364void 1588void
1365enqueue_read_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, ...) 1589read_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, ...)
1590 ALIAS:
1591 enqueue_read_image = 0
1366 PPCODE: 1592 PPCODE:
1367 cl_event ev = 0; 1593 cl_event ev = 0;
1368 const size_t src_origin[3] = { src_x, src_y, src_z }; 1594 const size_t src_origin[3] = { src_x, src_y, src_z };
1369 const size_t region[3] = { width, height, depth }; 1595 const size_t region[3] = { width, height, depth };
1370 EVENT_LIST (12, items - 12); 1596 EVENT_LIST (12);
1371 1597
1372 if (!row_pitch) 1598 if (!row_pitch)
1373 row_pitch = img_row_pitch (src); 1599 row_pitch = img_row_pitch (src);
1374 1600
1375 if (depth > 1 && !slice_pitch) 1601 if (depth > 1 && !slice_pitch)
1382 SvPOK_only (data); 1608 SvPOK_only (data);
1383 SvCUR_set (data, len); 1609 SvCUR_set (data, len);
1384 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)); 1610 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));
1385 1611
1386 if (ev) 1612 if (ev)
1387 XPUSH_CLOBJ ("OpenCL::Event", ev); 1613 XPUSH_CLOBJ (stash_event, ev);
1388 1614
1389void 1615void
1390enqueue_write_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, ...) 1616write_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, ...)
1617 ALIAS:
1618 enqueue_write_image = 0
1391 PPCODE: 1619 PPCODE:
1392 cl_event ev = 0; 1620 cl_event ev = 0;
1393 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1621 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1394 const size_t region[3] = { width, height, depth }; 1622 const size_t region[3] = { width, height, depth };
1395 STRLEN len; 1623 STRLEN len;
1396 char *ptr = SvPVbyte (data, len); 1624 char *ptr = SvPVbyte (data, len);
1397 EVENT_LIST (12, items - 12); 1625 EVENT_LIST (12);
1398 1626
1399 if (!row_pitch) 1627 if (!row_pitch)
1400 row_pitch = img_row_pitch (dst); 1628 row_pitch = img_row_pitch (dst);
1401 1629
1402 if (depth > 1 && !slice_pitch) 1630 if (depth > 1 && !slice_pitch)
1408 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1636 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
1409 1637
1410 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)); 1638 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));
1411 1639
1412 if (ev) 1640 if (ev)
1413 XPUSH_CLOBJ ("OpenCL::Event", ev); 1641 XPUSH_CLOBJ (stash_event, ev);
1414 1642
1415void 1643void
1416enqueue_copy_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, ...) 1644copy_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, ...)
1645 ALIAS:
1646 enqueue_copy_image = 0
1417 PPCODE: 1647 PPCODE:
1418 cl_event ev = 0; 1648 cl_event ev = 0;
1419 const size_t src_origin[3] = { src_x, src_y, src_z }; 1649 const size_t src_origin[3] = { src_x, src_y, src_z };
1420 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1650 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1421 const size_t region[3] = { width, height, depth }; 1651 const size_t region[3] = { width, height, depth };
1422 EVENT_LIST (12, items - 12); 1652 EVENT_LIST (12);
1423 1653
1424 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1654 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1425 1655
1426 if (ev) 1656 if (ev)
1427 XPUSH_CLOBJ ("OpenCL::Event", ev); 1657 XPUSH_CLOBJ (stash_event, ev);
1428 1658
1429void 1659void
1430enqueue_copy_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, ...) 1660copy_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, ...)
1661 ALIAS:
1662 enqueue_copy_image_to_buffer = 0
1431 PPCODE: 1663 PPCODE:
1432 cl_event ev = 0; 1664 cl_event ev = 0;
1433 const size_t src_origin[3] = { src_x, src_y, src_z }; 1665 const size_t src_origin[3] = { src_x, src_y, src_z };
1666 const size_t region [3] = { width, height, depth };
1667 EVENT_LIST (10);
1668
1669 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1670
1671 if (ev)
1672 XPUSH_CLOBJ (stash_event, ev);
1673
1674void
1675copy_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, ...)
1676 ALIAS:
1677 enqueue_copy_buffer_to_image = 0
1678 PPCODE:
1679 cl_event ev = 0;
1680 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1681 const size_t region [3] = { width, height, depth };
1682 EVENT_LIST (10);
1683
1684 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1685
1686 if (ev)
1687 XPUSH_CLOBJ (stash_event, ev);
1688
1689void
1690map_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, ...)
1691 ALIAS:
1692 enqueue_map_buffer = 0
1693 PPCODE:
1694 cl_event ev;
1695 EVENT_LIST (6);
1696 size_t cb = SvIV (cb_);
1697
1698 if (!SvOK (cb_))
1699 {
1700 NEED_SUCCESS (GetMemObjectInfo, (buf, CL_MEM_SIZE, sizeof (cb), &cb, 0));
1701 cb -= offset;
1702 }
1703
1704 NEED_SUCCESS_ARG (void *ptr, EnqueueMapBuffer, (self, buf, blocking, map_flags, offset, cb, event_list_count, event_list_ptr, &ev, &res));
1705 XPUSHs (mapped_new (stash_mappedbuffer, self, buf, map_flags, ptr, cb, ev, 0, 0));
1706
1707void
1708map_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, ...)
1709 ALIAS:
1710 enqueue_map_image = 0
1711 PPCODE:
1712 size_t width = SvIV (width_);
1713 if (!SvOK (width_))
1714 {
1715 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_WIDTH, sizeof (width), &width, 0));
1716 width -= x;
1717 }
1718
1719 size_t height = SvIV (width_);
1720 if (!SvOK (height_))
1721 {
1722 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_HEIGHT, sizeof (height), &height, 0));
1723 height -= y;
1724 }
1725
1726 size_t depth = SvIV (width_);
1727 if (!SvOK (depth_))
1728 {
1729 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_DEPTH, sizeof (depth), &depth, 0));
1730 depth -= z;
1731
1732 // stupid opencl returns 0 for depth, but requires 1 for 2d images
1733 if (!depth)
1734 depth = 1;
1735 }
1736
1737 const size_t origin[3] = { x, y, z };
1434 const size_t region[3] = { width, height, depth }; 1738 const size_t region[3] = { width, height, depth };
1739 size_t row_pitch, slice_pitch;
1435 EVENT_LIST (10, items - 10); 1740 EVENT_LIST (10);
1436 1741
1742 cl_event ev;
1743 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));
1744
1745 size_t cb = slice_pitch ? slice_pitch * region [2]
1746 : row_pitch ? row_pitch * region [1]
1747 : region [0];
1748
1749 XPUSHs (mapped_new (stash_mappedimage, self, img, map_flags, ptr, cb, ev, row_pitch, slice_pitch));
1750
1751void
1752unmap (OpenCL::Queue self, OpenCL::Mapped mapped, ...)
1753 PPCODE:
1754 mapped_unmap (ST (1), mapped, self, &ST (2), items - 2);
1755 if (GIMME_V != G_VOID)
1756 {
1757 clRetainEvent (mapped->event);
1758 XPUSH_CLOBJ (stash_event, mapped->event);
1759 }
1760
1761void
1762task (OpenCL::Queue self, OpenCL::Kernel kernel, ...)
1763 ALIAS:
1764 enqueue_task = 0
1765 PPCODE:
1766 cl_event ev = 0;
1767 EVENT_LIST (2);
1768
1437 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1769 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1438 1770
1439 if (ev) 1771 if (ev)
1440 XPUSH_CLOBJ ("OpenCL::Event", ev); 1772 XPUSH_CLOBJ (stash_event, ev);
1441 1773
1442void 1774void
1443enqueue_copy_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, ...)
1444 PPCODE:
1445 cl_event ev = 0;
1446 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1447 const size_t region[3] = { width, height, depth };
1448 EVENT_LIST (10, items - 10);
1449
1450 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1451
1452 if (ev)
1453 XPUSH_CLOBJ ("OpenCL::Event", ev);
1454
1455void
1456enqueue_task (OpenCL::Queue self, OpenCL::Kernel kernel, ...)
1457 PPCODE:
1458 cl_event ev = 0;
1459 EVENT_LIST (2, items - 2);
1460
1461 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1462
1463 if (ev)
1464 XPUSH_CLOBJ ("OpenCL::Event", ev);
1465
1466void
1467enqueue_nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...) 1775nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...)
1776 ALIAS:
1777 enqueue_nd_range_kernel = 0
1468 PPCODE: 1778 PPCODE:
1469 cl_event ev = 0; 1779 cl_event ev = 0;
1470 size_t *gwo = 0, *gws, *lws = 0; 1780 size_t *gwo = 0, *gws, *lws = 0;
1471 int gws_len; 1781 int gws_len;
1472 size_t *lists; 1782 size_t *lists;
1473 int i; 1783 int i;
1474 EVENT_LIST (5, items - 5); 1784 EVENT_LIST (5);
1475 1785
1476 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV) 1786 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV)
1477 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference"); 1787 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference");
1478 1788
1479 gws_len = AvFILLp (SvRV (global_work_size)) + 1; 1789 gws_len = AvFILLp (SvRV (global_work_size)) + 1;
1480 1790
1481 lists = tmpbuf (sizeof (size_t) * 3 * gws_len); 1791 lists = tmpbuf (sizeof (size_t) * 3 * gws_len);
1482 1792
1483 gws = lists + gws_len * 0; 1793 gws = lists + gws_len * 0;
1484 for (i = 0; i < gws_len; ++i) 1794 for (i = 0; i < gws_len; ++i)
1795 {
1485 gws [i] = SvIV (AvARRAY (SvRV (global_work_size))[i]); 1796 gws [i] = SvIV (AvARRAY (SvRV (global_work_size))[i]);
1797 // at least nvidia crashes for 0-sized work group sizes, work around
1798 if (!gws [i])
1799 croak ("clEnqueueNDRangeKernel: global_work_size[%d] is zero, must be non-zero", i);
1800 }
1486 1801
1487 if (SvOK (global_work_offset)) 1802 if (SvOK (global_work_offset))
1488 { 1803 {
1489 if (!SvROK (global_work_offset) || SvTYPE (SvRV (global_work_offset)) != SVt_PVAV) 1804 if (!SvROK (global_work_offset) || SvTYPE (SvRV (global_work_offset)) != SVt_PVAV)
1490 croak ("clEnqueueNDRangeKernel: global_work_offset must be undef or an array reference"); 1805 croak ("clEnqueueNDRangeKernel: global_work_offset must be undef or an array reference");
1498 } 1813 }
1499 1814
1500 if (SvOK (local_work_size)) 1815 if (SvOK (local_work_size))
1501 { 1816 {
1502 if ((SvOK (local_work_size) && !SvROK (local_work_size)) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV) 1817 if ((SvOK (local_work_size) && !SvROK (local_work_size)) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV)
1503 croak ("clEnqueueNDRangeKernel: global_work_size must be undef or an array reference"); 1818 croak ("clEnqueueNDRangeKernel: local_work_size must be undef or an array reference");
1504 1819
1505 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len) 1820 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len)
1506 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size"); 1821 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size");
1507 1822
1508 lws = lists + gws_len * 2; 1823 lws = lists + gws_len * 2;
1509 for (i = 0; i < gws_len; ++i) 1824 for (i = 0; i < gws_len; ++i)
1825 {
1510 lws [i] = SvIV (AvARRAY (SvRV (local_work_size))[i]); 1826 lws [i] = SvIV (AvARRAY (SvRV (local_work_size))[i]);
1827 // at least nvidia crashes for 0-sized work group sizes, work around
1828 if (!lws [i])
1829 croak ("clEnqueueNDRangeKernel: local_work_size[%d] is zero, must be non-zero", i);
1830 }
1511 } 1831 }
1512 1832
1513 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1833 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1514 1834
1515 if (ev) 1835 if (ev)
1516 XPUSH_CLOBJ ("OpenCL::Event", ev); 1836 XPUSH_CLOBJ (stash_event, ev);
1517 1837
1518#if cl_apple_gl_sharing || cl_khr_gl_sharing 1838#if cl_apple_gl_sharing || cl_khr_gl_sharing
1519 1839
1520void 1840void
1521enqueue_acquire_gl_objects (OpenCL::Queue self, SV *objects, ...) 1841acquire_gl_objects (OpenCL::Queue self, SV *objects, ...)
1522 ALIAS: 1842 ALIAS:
1843 release_gl_objects = 1
1844 enqueue_acquire_gl_objects = 0
1523 enqueue_release_gl_objects = 1 1845 enqueue_release_gl_objects = 1
1524 PPCODE: 1846 PPCODE:
1525 if (!SvROK (objects) || SvTYPE (SvRV (objects)) != SVt_PVAV) 1847 if (!SvROK (objects) || SvTYPE (SvRV (objects)) != SVt_PVAV)
1526 croak ("OpenCL::Queue::enqueue_acquire/release_gl_objects argument 'objects' must be an arrayref with memory objects, in call"); 1848 croak ("OpenCL::Queue::enqueue_acquire/release_gl_objects argument 'objects' must be an arrayref with memory objects, in call");
1527 1849
1528 cl_event ev = 0; 1850 cl_event ev = 0;
1529 EVENT_LIST (2, items - 2); 1851 EVENT_LIST (2);
1530 AV *av = (AV *)SvRV (objects); 1852 AV *av = (AV *)SvRV (objects);
1531 cl_uint num_objects = av_len (av) + 1; 1853 cl_uint num_objects = av_len (av) + 1;
1532 cl_mem *object_list = tmpbuf (sizeof (cl_mem) * num_objects); 1854 cl_mem *object_list = tmpbuf (sizeof (cl_mem) * num_objects);
1533 int i; 1855 int i;
1534 1856
1539 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1861 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1540 else 1862 else
1541 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1863 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1542 1864
1543 if (ev) 1865 if (ev)
1544 XPUSH_CLOBJ ("OpenCL::Event", ev); 1866 XPUSH_CLOBJ (stash_event, ev);
1545 1867
1546#endif 1868#endif
1547 1869
1548void 1870void
1549enqueue_wait_for_events (OpenCL::Queue self, ...) 1871wait_for_events (OpenCL::Queue self, ...)
1872 ALIAS:
1873 enqueue_wait_for_events = 0
1550 CODE: 1874 CODE:
1551 EVENT_LIST (1, items - 1); 1875 EVENT_LIST (1);
1552#if PREFER_1_1 1876#if PREFER_1_1
1553 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr)); 1877 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
1554#else 1878#else
1555 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0)); 1879 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0));
1556#endif 1880#endif
1557 1881
1558void 1882void
1559enqueue_marker (OpenCL::Queue self, ...) 1883marker (OpenCL::Queue self, ...)
1884 ALIAS:
1885 enqueue_marker = 0
1560 PPCODE: 1886 PPCODE:
1561 cl_event ev = 0; 1887 cl_event ev = 0;
1562 EVENT_LIST (1, items - 1); 1888 EVENT_LIST (1);
1563#if PREFER_1_1 1889#if PREFER_1_1
1564 if (!event_list_count) 1890 if (!event_list_count)
1565 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0)); 1891 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
1566 else 1892 else
1567#if CL_VERSION_1_2 1893#if CL_VERSION_1_2
1574#endif 1900#endif
1575#else 1901#else
1576 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1902 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1577#endif 1903#endif
1578 if (ev) 1904 if (ev)
1579 XPUSH_CLOBJ ("OpenCL::Event", ev); 1905 XPUSH_CLOBJ (stash_event, ev);
1580 1906
1581void 1907void
1582enqueue_barrier (OpenCL::Queue self, ...) 1908barrier (OpenCL::Queue self, ...)
1909 ALIAS:
1910 enqueue_barrier = 0
1583 PPCODE: 1911 PPCODE:
1584 cl_event ev = 0; 1912 cl_event ev = 0;
1585 EVENT_LIST (1, items - 1); 1913 EVENT_LIST (1);
1586#if PREFER_1_1 1914#if PREFER_1_1
1587 if (!event_list_count && GIMME_V == G_VOID) 1915 if (!event_list_count && GIMME_V == G_VOID)
1588 NEED_SUCCESS (EnqueueBarrier, (self)); 1916 NEED_SUCCESS (EnqueueBarrier, (self));
1589 else 1917 else
1590#if CL_VERSION_1_2 1918#if CL_VERSION_1_2
1600#endif 1928#endif
1601#else 1929#else
1602 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1930 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1603#endif 1931#endif
1604 if (ev) 1932 if (ev)
1605 XPUSH_CLOBJ ("OpenCL::Event", ev); 1933 XPUSH_CLOBJ (stash_event, ev);
1606 1934
1607void 1935void
1608flush (OpenCL::Queue self) 1936flush (OpenCL::Queue self)
1609 CODE: 1937 CODE:
1610 NEED_SUCCESS (Flush, (self)); 1938 NEED_SUCCESS (Flush, (self));
1628 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0)); 1956 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0));
1629 EXTEND (SP, 1); 1957 EXTEND (SP, 1);
1630 const int i = 0; 1958 const int i = 0;
1631 { 1959 {
1632 NEED_SUCCESS (RetainContext, (value [i])); 1960 NEED_SUCCESS (RetainContext, (value [i]));
1633 PUSH_CLOBJ ("OpenCL::Context", value [i]); 1961 PUSH_CLOBJ (stash_context, value [i]);
1634 } 1962 }
1635 1963
1636void 1964void
1637device (OpenCL::Queue self) 1965device (OpenCL::Queue self)
1638 PPCODE: 1966 PPCODE:
1639 cl_device_id value [1]; 1967 cl_device_id value [1];
1640 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0)); 1968 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0));
1641 EXTEND (SP, 1); 1969 EXTEND (SP, 1);
1642 const int i = 0; 1970 const int i = 0;
1643 { 1971 {
1644 PUSH_CLOBJ ("OpenCL::Device", value [i]); 1972 PUSH_CLOBJ (stash_device, value [i]);
1645 } 1973 }
1646 1974
1647void 1975void
1648reference_count (OpenCL::Queue self) 1976reference_count (OpenCL::Queue self)
1649 PPCODE: 1977 PPCODE:
1658 PPCODE: 1986 PPCODE:
1659 cl_command_queue_properties value [1]; 1987 cl_command_queue_properties value [1];
1660 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_PROPERTIES, sizeof (value), value, 0)); 1988 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_PROPERTIES, sizeof (value), value, 0));
1661 EXTEND (SP, 1); 1989 EXTEND (SP, 1);
1662 const int i = 0; 1990 const int i = 0;
1663 PUSHs (sv_2mortal (newSViv (value [i]))); 1991 PUSHs (sv_2mortal (newSVuv (value [i])));
1664 1992
1665#END:command_queue 1993#END:command_queue
1666 1994
1667MODULE = OpenCL PACKAGE = OpenCL::Memory 1995MODULE = OpenCL PACKAGE = OpenCL::Memory
1668 1996
1683 PPCODE: 2011 PPCODE:
1684 cl_mem_object_type value [1]; 2012 cl_mem_object_type value [1];
1685 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_TYPE, sizeof (value), value, 0)); 2013 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_TYPE, sizeof (value), value, 0));
1686 EXTEND (SP, 1); 2014 EXTEND (SP, 1);
1687 const int i = 0; 2015 const int i = 0;
1688 PUSHs (sv_2mortal (newSViv (value [i]))); 2016 PUSHs (sv_2mortal (newSVuv (value [i])));
1689 2017
1690void 2018void
1691flags (OpenCL::Memory self) 2019flags (OpenCL::Memory self)
1692 PPCODE: 2020 PPCODE:
1693 cl_mem_flags value [1]; 2021 cl_mem_flags value [1];
1694 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_FLAGS, sizeof (value), value, 0)); 2022 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_FLAGS, sizeof (value), value, 0));
1695 EXTEND (SP, 1); 2023 EXTEND (SP, 1);
1696 const int i = 0; 2024 const int i = 0;
1697 PUSHs (sv_2mortal (newSViv (value [i]))); 2025 PUSHs (sv_2mortal (newSVuv (value [i])));
1698 2026
1699void 2027void
1700size (OpenCL::Memory self) 2028size (OpenCL::Memory self)
1701 ALIAS: 2029 ALIAS:
1702 size = CL_MEM_SIZE 2030 size = CL_MEM_SIZE
1736 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0)); 2064 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0));
1737 EXTEND (SP, 1); 2065 EXTEND (SP, 1);
1738 const int i = 0; 2066 const int i = 0;
1739 { 2067 {
1740 NEED_SUCCESS (RetainContext, (value [i])); 2068 NEED_SUCCESS (RetainContext, (value [i]));
1741 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2069 PUSH_CLOBJ (stash_context, value [i]);
1742 } 2070 }
1743 2071
1744void 2072void
1745associated_memobject (OpenCL::Memory self) 2073associated_memobject (OpenCL::Memory self)
1746 PPCODE: 2074 PPCODE:
1748 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0)); 2076 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0));
1749 EXTEND (SP, 1); 2077 EXTEND (SP, 1);
1750 const int i = 0; 2078 const int i = 0;
1751 { 2079 {
1752 NEED_SUCCESS (RetainMemObject, (value [i])); 2080 NEED_SUCCESS (RetainMemObject, (value [i]));
1753 PUSH_CLOBJ ("OpenCL::Memory", value [i]); 2081 PUSH_CLOBJ (stash_memory, value [i]);
1754 } 2082 }
1755 2083
1756#END:mem 2084#END:mem
1757 2085
1758#if cl_apple_gl_sharing || cl_khr_gl_sharing 2086#if cl_apple_gl_sharing || cl_khr_gl_sharing
1778 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!"); 2106 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!");
1779 2107
1780 cl_buffer_region crdata = { origin, size }; 2108 cl_buffer_region crdata = { origin, size };
1781 2109
1782 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res)); 2110 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res));
1783 XPUSH_CLOBJ ("OpenCL::Buffer", mem); 2111 XPUSH_CLOBJ (stash_buffer, mem);
1784 2112
1785MODULE = OpenCL PACKAGE = OpenCL::Image 2113MODULE = OpenCL PACKAGE = OpenCL::Image
1786 2114
1787void 2115void
1788image_info (OpenCL::Image self, cl_image_info name) 2116image_info (OpenCL::Image self, cl_image_info name)
1874 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0)); 2202 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0));
1875 EXTEND (SP, 1); 2203 EXTEND (SP, 1);
1876 const int i = 0; 2204 const int i = 0;
1877 { 2205 {
1878 NEED_SUCCESS (RetainContext, (value [i])); 2206 NEED_SUCCESS (RetainContext, (value [i]));
1879 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2207 PUSH_CLOBJ (stash_context, value [i]);
1880 } 2208 }
1881 2209
1882void 2210void
1883normalized_coords (OpenCL::Sampler self) 2211normalized_coords (OpenCL::Sampler self)
1884 PPCODE: 2212 PPCODE:
1885 cl_addressing_mode value [1]; 2213 cl_addressing_mode value [1];
1886 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0)); 2214 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0));
1887 EXTEND (SP, 1); 2215 EXTEND (SP, 1);
1888 const int i = 0; 2216 const int i = 0;
1889 PUSHs (sv_2mortal (newSViv (value [i]))); 2217 PUSHs (sv_2mortal (newSVuv (value [i])));
1890 2218
1891void 2219void
1892addressing_mode (OpenCL::Sampler self) 2220addressing_mode (OpenCL::Sampler self)
1893 PPCODE: 2221 PPCODE:
1894 cl_filter_mode value [1]; 2222 cl_filter_mode value [1];
1895 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0)); 2223 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0));
1896 EXTEND (SP, 1); 2224 EXTEND (SP, 1);
1897 const int i = 0; 2225 const int i = 0;
1898 PUSHs (sv_2mortal (newSViv (value [i]))); 2226 PUSHs (sv_2mortal (newSVuv (value [i])));
1899 2227
1900void 2228void
1901filter_mode (OpenCL::Sampler self) 2229filter_mode (OpenCL::Sampler self)
1902 PPCODE: 2230 PPCODE:
1903 cl_bool value [1]; 2231 cl_bool value [1];
1995 2323
1996void 2324void
1997kernel (OpenCL::Program program, SV *function) 2325kernel (OpenCL::Program program, SV *function)
1998 PPCODE: 2326 PPCODE:
1999 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res)); 2327 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res));
2000 XPUSH_CLOBJ ("OpenCL::Kernel", kernel); 2328 XPUSH_CLOBJ (stash_kernel, kernel);
2001 2329
2002void 2330void
2003kernels_in_program (OpenCL::Program program) 2331kernels_in_program (OpenCL::Program program)
2004 PPCODE: 2332 PPCODE:
2005 cl_uint num_kernels; 2333 cl_uint num_kernels;
2008 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0)); 2336 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0));
2009 2337
2010 int i; 2338 int i;
2011 EXTEND (SP, num_kernels); 2339 EXTEND (SP, num_kernels);
2012 for (i = 0; i < num_kernels; ++i) 2340 for (i = 0; i < num_kernels; ++i)
2013 PUSH_CLOBJ ("OpenCL::Kernel", kernels [i]); 2341 PUSH_CLOBJ (stash_kernel, kernels [i]);
2014 2342
2015void 2343void
2016info (OpenCL::Program self, cl_program_info name) 2344info (OpenCL::Program self, cl_program_info name)
2017 PPCODE: 2345 PPCODE:
2018 INFO (Program) 2346 INFO (Program)
2066 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0)); 2394 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0));
2067 EXTEND (SP, 1); 2395 EXTEND (SP, 1);
2068 const int i = 0; 2396 const int i = 0;
2069 { 2397 {
2070 NEED_SUCCESS (RetainContext, (value [i])); 2398 NEED_SUCCESS (RetainContext, (value [i]));
2071 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2399 PUSH_CLOBJ (stash_context, value [i]);
2072 } 2400 }
2073 2401
2074void 2402void
2075devices (OpenCL::Program self) 2403devices (OpenCL::Program self)
2076 PPCODE: 2404 PPCODE:
2080 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0)); 2408 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0));
2081 int i, n = size / sizeof (*value); 2409 int i, n = size / sizeof (*value);
2082 EXTEND (SP, n); 2410 EXTEND (SP, n);
2083 for (i = 0; i < n; ++i) 2411 for (i = 0; i < n; ++i)
2084 { 2412 {
2085 PUSH_CLOBJ ("OpenCL::Device", value [i]); 2413 PUSH_CLOBJ (stash_device, value [i]);
2086 } 2414 }
2087 2415
2088void 2416void
2089source (OpenCL::Program self) 2417source (OpenCL::Program self)
2090 PPCODE: 2418 PPCODE:
2116DESTROY (OpenCL::Kernel self) 2444DESTROY (OpenCL::Kernel self)
2117 CODE: 2445 CODE:
2118 clReleaseKernel (self); 2446 clReleaseKernel (self);
2119 2447
2120void 2448void
2449setf (OpenCL::Kernel self, const char *format, ...)
2450 CODE:
2451 int i;
2452 for (i = 2; ; ++i)
2453 {
2454 while (*format == ' ')
2455 ++format;
2456
2457 char type = *format++;
2458
2459 if (!type)
2460 break;
2461
2462 if (i >= items)
2463 croak ("OpenCL::Kernel::setf format string too long (not enough arguments)");
2464
2465 SV *sv = ST (i);
2466
2467 union
2468 {
2469 cl_char cc; cl_uchar cC; cl_short cs; cl_ushort cS;
2470 cl_int ci; cl_uint cI; cl_long cl; cl_ulong cL;
2471 cl_half ch; cl_float cf; cl_double cd;
2472 cl_mem cm;
2473 cl_sampler ca;
2474 size_t cz;
2475 cl_event ce;
2476 } arg;
2477 size_t size;
2478 int nullarg = 0;
2479
2480 switch (type)
2481 {
2482 case 'c': arg.cc = SvIV (sv); size = sizeof (arg.cc); break;
2483 case 'C': arg.cC = SvUV (sv); size = sizeof (arg.cC); break;
2484 case 's': arg.cs = SvIV (sv); size = sizeof (arg.cs); break;
2485 case 'S': arg.cS = SvUV (sv); size = sizeof (arg.cS); break;
2486 case 'i': arg.ci = SvIV (sv); size = sizeof (arg.ci); break;
2487 case 'I': arg.cI = SvUV (sv); size = sizeof (arg.cI); break;
2488 case 'l': arg.cl = SvIV (sv); size = sizeof (arg.cl); break;
2489 case 'L': arg.cL = SvUV (sv); size = sizeof (arg.cL); break;
2490
2491 case 'h': arg.ch = SvUV (sv); size = sizeof (arg.ch); break;
2492 case 'f': arg.cf = SvNV (sv); size = sizeof (arg.cf); break;
2493 case 'd': arg.cd = SvNV (sv); size = sizeof (arg.cd); break;
2494
2495 case 'z': nullarg = 1; size = SvIV (sv); break;
2496
2497 case 'm': nullarg = !SvOK (sv); arg.cm = SvCLOBJ ("OpenCL::Kernel::setf", "m", sv, "OpenCL::Memory" ); size = sizeof (arg.cm); break;
2498 case 'a': nullarg = !SvOK (sv); arg.ca = SvCLOBJ ("OpenCL::Kernel::setf", "a", sv, "OpenCL::Sampler"); size = sizeof (arg.ca); break;
2499 case 'e': nullarg = !SvOK (sv); arg.ca = SvCLOBJ ("OpenCL::Kernel::setf", "e", sv, "OpenCL::Event" ); size = sizeof (arg.ce); break;
2500
2501 default:
2502 croak ("OpenCL::Kernel::setf format character '%c' not supported", type);
2503 }
2504
2505 res = clSetKernelArg (self, i - 2, size, nullarg ? 0 : &arg);
2506 if (res)
2507 croak ("OpenCL::Kernel::setf kernel parameter '%c' (#%d): %s", type, i - 2, err2str (res));
2508 }
2509
2510 if (i != items)
2511 croak ("OpenCL::Kernel::setf format string too short (too many arguments)");
2512
2513void
2121set_char (OpenCL::Kernel self, cl_uint idx, cl_char value) 2514set_char (OpenCL::Kernel self, cl_uint idx, cl_char value)
2122 CODE: 2515 CODE:
2123 clSetKernelArg (self, idx, sizeof (value), &value); 2516 clSetKernelArg (self, idx, sizeof (value), &value);
2124 2517
2125void 2518void
2173 clSetKernelArg (self, idx, sizeof (value), &value); 2566 clSetKernelArg (self, idx, sizeof (value), &value);
2174 2567
2175void 2568void
2176set_memory (OpenCL::Kernel self, cl_uint idx, OpenCL::Memory_ornull value) 2569set_memory (OpenCL::Kernel self, cl_uint idx, OpenCL::Memory_ornull value)
2177 CODE: 2570 CODE:
2178 clSetKernelArg (self, idx, sizeof (value), &value); 2571 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
2179 2572
2180void 2573void
2181set_buffer (OpenCL::Kernel self, cl_uint idx, OpenCL::Buffer_ornull value) 2574set_buffer (OpenCL::Kernel self, cl_uint idx, OpenCL::Buffer_ornull value)
2182 CODE: 2575 CODE:
2183 clSetKernelArg (self, idx, sizeof (value), &value); 2576 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
2184 2577
2185void 2578void
2186set_image2d (OpenCL::Kernel self, cl_uint idx, OpenCL::Image2D_ornull value) 2579set_image (OpenCL::Kernel self, cl_uint idx, OpenCL::Image_ornull value)
2580 ALIAS:
2581 set_image2d = 0
2582 set_image3d = 0
2187 CODE: 2583 CODE:
2188 clSetKernelArg (self, idx, sizeof (value), &value); 2584 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
2189
2190void
2191set_image3d (OpenCL::Kernel self, cl_uint idx, OpenCL::Image3D_ornull value)
2192 CODE:
2193 clSetKernelArg (self, idx, sizeof (value), &value);
2194 2585
2195void 2586void
2196set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value) 2587set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value)
2197 CODE: 2588 CODE:
2198 clSetKernelArg (self, idx, sizeof (value), &value); 2589 clSetKernelArg (self, idx, sizeof (value), &value);
2244 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0)); 2635 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0));
2245 EXTEND (SP, 1); 2636 EXTEND (SP, 1);
2246 const int i = 0; 2637 const int i = 0;
2247 { 2638 {
2248 NEED_SUCCESS (RetainContext, (value [i])); 2639 NEED_SUCCESS (RetainContext, (value [i]));
2249 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2640 PUSH_CLOBJ (stash_context, value [i]);
2250 } 2641 }
2251 2642
2252void 2643void
2253program (OpenCL::Kernel self) 2644program (OpenCL::Kernel self)
2254 PPCODE: 2645 PPCODE:
2256 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0)); 2647 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0));
2257 EXTEND (SP, 1); 2648 EXTEND (SP, 1);
2258 const int i = 0; 2649 const int i = 0;
2259 { 2650 {
2260 NEED_SUCCESS (RetainProgram, (value [i])); 2651 NEED_SUCCESS (RetainProgram, (value [i]));
2261 PUSH_CLOBJ ("OpenCL::Program", value [i]); 2652 PUSH_CLOBJ (stash_program, value [i]);
2262 } 2653 }
2263 2654
2264#END:kernel 2655#END:kernel
2265 2656
2266void 2657void
2346 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0)); 2737 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0));
2347 EXTEND (SP, 1); 2738 EXTEND (SP, 1);
2348 const int i = 0; 2739 const int i = 0;
2349 { 2740 {
2350 NEED_SUCCESS (RetainCommandQueue, (value [i])); 2741 NEED_SUCCESS (RetainCommandQueue, (value [i]));
2351 PUSH_CLOBJ ("OpenCL::Queue", value [i]); 2742 PUSH_CLOBJ (stash_queue, value [i]);
2352 } 2743 }
2353 2744
2354void 2745void
2355command_type (OpenCL::Event self) 2746command_type (OpenCL::Event self)
2356 PPCODE: 2747 PPCODE:
2379 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0)); 2770 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0));
2380 EXTEND (SP, 1); 2771 EXTEND (SP, 1);
2381 const int i = 0; 2772 const int i = 0;
2382 { 2773 {
2383 NEED_SUCCESS (RetainContext, (value [i])); 2774 NEED_SUCCESS (RetainContext, (value [i]));
2384 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2775 PUSH_CLOBJ (stash_context, value [i]);
2385 } 2776 }
2386 2777
2387#END:event 2778#END:event
2388 2779
2389void 2780void
2414void 2805void
2415set_status (OpenCL::UserEvent self, cl_int execution_status) 2806set_status (OpenCL::UserEvent self, cl_int execution_status)
2416 CODE: 2807 CODE:
2417 clSetUserEventStatus (self, execution_status); 2808 clSetUserEventStatus (self, execution_status);
2418 2809
2810MODULE = OpenCL PACKAGE = OpenCL::Mapped
2811
2812void
2813DESTROY (SV *self)
2814 CODE:
2815 OpenCL__Mapped mapped = SvMAPPED (self);
2816
2817 clEnqueueUnmapMemObject (mapped->queue, mapped->memobj, mapped->ptr, 1, &mapped->event, 0);
2818 mapped_detach (self, mapped);
2819
2820 clReleaseCommandQueue (mapped->queue);
2821 clReleaseEvent (mapped->event);
2822 Safefree (mapped);
2823
2824void
2825unmap (OpenCL::Mapped self, ...)
2826 CODE:
2827 mapped_unmap (ST (0), self, self->queue, &ST (1), items - 1);
2828
2829bool
2830mapped (OpenCL::Mapped self)
2831 CODE:
2832 RETVAL = !!self->ptr;
2833 OUTPUT:
2834 RETVAL
2835
2836void
2837wait (OpenCL::Mapped self)
2838 PPCODE:
2839 if (self->event)
2840 NEED_SUCCESS (WaitForEvents, (1, &self->event));
2841
2842void
2843event (OpenCL::Mapped self)
2844 PPCODE:
2845 if (!self->event)
2846 XSRETURN_UNDEF;
2847
2848 clRetainEvent (self->event);
2849 XPUSH_CLOBJ (stash_event, self->event);
2850
2851size_t
2852size (OpenCL::Mapped self)
2853 CODE:
2854 RETVAL = self->cb;
2855 OUTPUT:
2856 RETVAL
2857
2858IV
2859ptr (OpenCL::Mapped self)
2860 CODE:
2861 RETVAL = PTR2IV (self->ptr);
2862 OUTPUT:
2863 RETVAL
2864
2865void
2866set (OpenCL::Mapped self, size_t offset, SV *data)
2867 CODE:
2868 STRLEN len;
2869 const char *ptr = SvPVbyte (data, len);
2870
2871 if (offset + len > self->cb)
2872 croak ("OpenCL::Mapped::set out of bound condition detected");
2873
2874 memcpy (offset + (char *)self->ptr, ptr, len);
2875
2876MODULE = OpenCL PACKAGE = OpenCL::MappedBuffer
2877
2878MODULE = OpenCL PACKAGE = OpenCL::MappedImage
2879
2880IV
2881row_pitch (OpenCL::Mapped self)
2882 ALIAS:
2883 slice_pitch = 1
2884 CODE:
2885 RETVAL = ix ? self->slice_pitch : self->row_pitch;
2886 OUTPUT:
2887 RETVAL
2888
2889
2890

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines