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

Comparing OpenCL/OpenCL.xs (file contents):
Revision 1.52 by root, Tue Apr 24 23:53:12 2012 UTC vs.
Revision 1.61 by root, Tue May 1 16:37:23 2012 UTC

57typedef cl_program OpenCL__Program; 57typedef cl_program OpenCL__Program;
58typedef cl_kernel OpenCL__Kernel; 58typedef cl_kernel OpenCL__Kernel;
59typedef cl_event OpenCL__Event; 59typedef cl_event OpenCL__Event;
60typedef cl_event OpenCL__UserEvent; 60typedef cl_event OpenCL__UserEvent;
61 61
62typedef struct mapped * OpenCL__Mapped;
63
62typedef SV *FUTURE; 64typedef SV *FUTURE;
65
66static HV
67 *stash_platform,
68 *stash_device,
69 *stash_context,
70 *stash_queue,
71 *stash_program,
72 *stash_kernel,
73 *stash_sampler,
74 *stash_event,
75 *stash_userevent,
76 *stash_memory,
77 *stash_buffer,
78 *stash_bufferobj,
79 *stash_image,
80 *stash_image1d,
81 *stash_image1darray,
82 *stash_image1dbuffer,
83 *stash_image2d,
84 *stash_image2darray,
85 *stash_image3d,
86 *stash_mapped,
87 *stash_mappedbuffer,
88 *stash_mappedimage;
63 89
64/*****************************************************************************/ 90/*****************************************************************************/
65 91
66// name must include a leading underscore 92// name must include a leading underscore
67// all of this horrors would be unneceesary if somebody wrote a proper OpenGL module 93// all of this horrors would be unneceesary if somebody wrote a proper OpenGL module
260 croak ("%s: %s is not a property list (either undef or [type => value, ...])", func, svname); 286 croak ("%s: %s is not a property list (either undef or [type => value, ...])", func, svname);
261} 287}
262 288
263/*****************************************************************************/ 289/*****************************************************************************/
264 290
265#define NEW_CLOBJ(class,ptr) sv_setref_pv (sv_newmortal (), class, ptr) 291static SV *
266#define PUSH_CLOBJ(class,ptr) PUSHs (NEW_CLOBJ (class, ptr)) 292new_clobj (HV *stash, IV id)
267#define XPUSH_CLOBJ(class,ptr) XPUSHs (NEW_CLOBJ (class, ptr)) 293{
294 return sv_2mortal (sv_bless (newRV_noinc (newSViv (id)), stash));
295}
296
297#define PUSH_CLOBJ(stash,id) PUSHs (new_clobj ((stash), (IV)(id)))
298#define XPUSH_CLOBJ(stash,id) XPUSHs (new_clobj ((stash), (IV)(id)))
268 299
269/* cl objects are either \$iv, or [$iv, ...] */ 300/* cl objects are either \$iv, or [$iv, ...] */
270/* they can be upgraded at runtime to the array form */ 301/* they can be upgraded at runtime to the array form */
271static void * 302static void *
272SvCLOBJ (const char *func, const char *svname, SV *sv, const char *pkg) 303SvCLOBJ (const char *func, const char *svname, SV *sv, const char *pkg)
273{ 304{
274 if (SvROK (sv) && sv_derived_from (sv, pkg)) 305 if (SvROK (sv) && sv_derived_from (sv, pkg))
275 {
276 SV *rv = SvRV (sv);
277
278#if CLOBJ_PUSH
279 if (SvTYPE (rv) == SVt_PVAV)
280 rv = AvARRAY (rv)[0];
281#endif
282
283 return (void *)SvIV (rv); 306 return (void *)SvIV (SvRV (sv));
284 }
285 307
286 croak ("%s: %s is not of type %s", func, svname, pkg); 308 croak ("%s: %s is not of type %s", func, svname, pkg);
287} 309}
288
289#if CLOBJ_PUSH
290
291static void
292CLOBJ_push (SV *self, SV *data)
293{
294 SV *rv = SvRV (self);
295
296 if (SvTYPE (rv) != SVt_PVAV)
297 {
298 AV *av = newAV ();
299 av_push (av, rv);
300 rv = (SV *)av;
301 SvRV_set (self, rv);
302 }
303
304 av_push ((AV *)rv, data);
305}
306
307static SV *
308sv_struct (STRLEN size)
309{
310 SV *sv = newSV (size);
311 SvPOK_only (sv);
312 return sv;
313}
314
315static void *
316CLOBJ_push_struct (SV *self, STRLEN size)
317{
318 SV *sv = sv_struct (size);
319 CLOBJ_push (self, sv);
320 return SvPVX (sv);
321}
322
323#endif
324 310
325/*****************************************************************************/ 311/*****************************************************************************/
326/* callback stuff */ 312/* callback stuff */
327 313
328/* default context callback, log to stderr */ 314/* default context callback, log to stderr */
386 372
387 X_UNLOCK (eq_lock); 373 X_UNLOCK (eq_lock);
388 374
389 return res; 375 return res;
390} 376}
391
392#if 0
393static void
394mem_free (pTHX_ void *p)
395{
396 free (p);
397}
398//SAVEDESTRUCTOR_X (mem_free, item);
399#endif
400 377
401static void 378static void
402eq_poll (void) 379eq_poll (void)
403{ 380{
404 eq_item *item; 381 eq_item *item;
470 } 447 }
471 448
472static SV * 449static SV *
473new_clobj_context (cl_context ctx, void *user_data) 450new_clobj_context (cl_context ctx, void *user_data)
474{ 451{
475 SV *sv = NEW_CLOBJ ("OpenCL::Context", ctx); 452 SV *sv = new_clobj (stash_context, (IV)ctx);
476 453
477 if (user_data) 454 if (user_data)
478 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0); 455 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0);
479 456
480 return sv; 457 return sv;
487 464
488static void 465static void
489eq_program_push (void *data1, void *data2, void *data3) 466eq_program_push (void *data1, void *data2, void *data3)
490{ 467{
491 dSP; 468 dSP;
492 PUSH_CLOBJ ("OpenCL::Program", data1); 469 PUSH_CLOBJ (stash_program, data1);
493 PUTBACK; 470 PUTBACK;
494} 471}
495 472
496static eq_vtbl eq_program_vtbl = { 1, eq_program_push }; 473static eq_vtbl eq_program_vtbl = { 1, eq_program_push };
497 474
544 521
545static void 522static void
546eq_event_push (void *data1, void *data2, void *data3) 523eq_event_push (void *data1, void *data2, void *data3)
547{ 524{
548 dSP; 525 dSP;
549 PUSH_CLOBJ ("OpenCL::Event", data1); 526 PUSH_CLOBJ (stash_event, data1);
550 PUSHs (sv_2mortal (newSViv ((IV)data2))); 527 PUSHs (sv_2mortal (newSViv ((IV)data2)));
551 PUTBACK; 528 PUTBACK;
552} 529}
553 530
554static eq_vtbl eq_event_vtbl = { 1, eq_event_push }; 531static eq_vtbl eq_event_vtbl = { 1, eq_event_push };
556static void CL_CALLBACK 533static void CL_CALLBACK
557eq_event_notify (cl_event event, cl_int event_command_exec_status, void *user_data) 534eq_event_notify (cl_event event, cl_int event_command_exec_status, void *user_data)
558{ 535{
559 clRetainEvent (event); 536 clRetainEvent (event);
560 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0); 537 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0);
538}
539
540/*****************************************************************************/
541/* mapped_xxx */
542
543static OpenCL__Mapped
544SvMAPPED (SV *sv)
545{
546 // no typechecking atm., keep your fingers crossed
547 return (OpenCL__Mapped)SvMAGIC (SvRV (sv))->mg_ptr;
548}
549
550struct mapped
551{
552 cl_command_queue queue;
553 cl_mem memobj;
554 void *ptr;
555 size_t cb;
556 cl_event event;
557 size_t row_pitch;
558 size_t slice_pitch;
559};
560
561static SV *
562mapped_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)
563{
564 SV *data = newSV (0);
565 SvUPGRADE (data, SVt_PVMG);
566
567 OpenCL__Mapped mapped;
568 New (0, mapped, 1, struct mapped);
569
570 clRetainCommandQueue (queue);
571
572 mapped->queue = queue;
573 mapped->memobj = memobj;
574 mapped->ptr = ptr;
575 mapped->cb = cb;
576 mapped->event = ev;
577 mapped->row_pitch = row_pitch;
578 mapped->slice_pitch = slice_pitch;
579
580 sv_magicext (data, 0, PERL_MAGIC_ext, 0, (char *)mapped, 0);
581
582 if (!(flags & CL_MAP_WRITE))
583 SvREADONLY_on (data);
584
585 if (SvLEN (data))
586 Safefree (data);
587
588 SvPVX (data) = (char *)ptr;
589 SvCUR_set (data, cb);
590 SvLEN_set (data, 0);
591 SvPOK_only (data);
592
593 return sv_2mortal (sv_bless (newRV_noinc (data), stash));
594}
595
596static void
597mapped_detach (SV *sv, OpenCL__Mapped mapped)
598{
599 SV *data = SvRV (sv);
600
601 if (SvPVX (data) != (char *)mapped->ptr)
602 warn ("FATAL: OpenCL memory mapped scalar changed location, detected");
603 else
604 {
605 SvREADONLY_off (data);
606 SvCUR_set (data, 0);
607 SvPVX (data) = 0;
608 SvOK_off (data);
609 }
610
611 mapped->ptr = 0;
561} 612}
562 613
563/*****************************************************************************/ 614/*****************************************************************************/
564 615
565static size_t 616static size_t
569 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0); 620 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0);
570 return res; 621 return res;
571} 622}
572 623
573static cl_event * 624static cl_event *
574event_list (SV **items, cl_uint *rcount) 625event_list (SV **items, cl_uint *rcount, cl_event extra)
575{ 626{
576 cl_uint count = *rcount; 627 cl_uint count = *rcount;
577 628
578 if (!count) 629 if (!count && !extra)
579 return 0; 630 return 0;
580 631
581 cl_event *list = tmpbuf (sizeof (cl_event) * count); 632 cl_event *list = tmpbuf (sizeof (cl_event) * (count + 1));
582 int i = 0; 633 int i = 0;
583 634
584 do 635 while (count)
585 { 636 {
586 --count; 637 --count;
587 if (SvOK (items [count])) 638 if (SvOK (items [count]))
588 list [i++] = SvCLOBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event"); 639 list [i++] = SvCLOBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event");
589 } 640 }
590 while (count); 641
642 if (extra)
643 list [i++] = extra;
591 644
592 *rcount = i; 645 *rcount = i;
593 646
594 return i ? list : 0; 647 return i ? list : 0;
595} 648}
596 649
597#define EVENT_LIST(items,count) \ 650#define EVENT_LIST(skip) \
598 cl_uint event_list_count = (count); \ 651 cl_uint event_list_count = items - skip; \
599 cl_event *event_list_ptr = event_list (&ST (items), &event_list_count) 652 cl_event *event_list_ptr = event_list (&ST (skip), &event_list_count, 0)
600 653
601#define INFO(class) \ 654#define INFO(class) \
602{ \ 655{ \
603 size_t size; \ 656 size_t size; \
604 NEED_SUCCESS (Get ## class ## Info, (self, name, 0, 0, &size)); \ 657 NEED_SUCCESS (Get ## class ## Info, (self, name, 0, 0, &size)); \
644 }; 697 };
645 698
646 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--) 699 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--)
647 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv)); 700 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv));
648 701
702 stash_platform = gv_stashpv ("OpenCL::Platform", GV_ADD);
703 stash_device = gv_stashpv ("OpenCL::Device", GV_ADD);
704 stash_context = gv_stashpv ("OpenCL::Context", GV_ADD);
705 stash_queue = gv_stashpv ("OpenCL::Queue", GV_ADD);
706 stash_program = gv_stashpv ("OpenCL::Program", GV_ADD);
707 stash_kernel = gv_stashpv ("OpenCL::Kernel", GV_ADD);
708 stash_sampler = gv_stashpv ("OpenCL::Sampler", GV_ADD);
709 stash_event = gv_stashpv ("OpenCL::Event", GV_ADD);
710 stash_userevent = gv_stashpv ("OpenCL::UserEvent", GV_ADD);
711 stash_memory = gv_stashpv ("OpenCL::Memory", GV_ADD);
712 stash_buffer = gv_stashpv ("OpenCL::Buffer", GV_ADD);
713 stash_bufferobj = gv_stashpv ("OpenCL::BufferObj", GV_ADD);
714 stash_image = gv_stashpv ("OpenCL::Image", GV_ADD);
715 stash_image1d = gv_stashpv ("OpenCL::Image1D", GV_ADD);
716 stash_image1darray = gv_stashpv ("OpenCL::Image1DArray", GV_ADD);
717 stash_image1dbuffer = gv_stashpv ("OpenCL::Image1DBuffer", GV_ADD);
718 stash_image2d = gv_stashpv ("OpenCL::Image2D", GV_ADD);
719 stash_image2darray = gv_stashpv ("OpenCL::Image2DArray", GV_ADD);
720 stash_image3d = gv_stashpv ("OpenCL::Image3D", GV_ADD);
721 stash_mapped = gv_stashpv ("OpenCL::Mapped", GV_ADD);
722 stash_mappedbuffer = gv_stashpv ("OpenCL::MappedBuffer", GV_ADD);
723 stash_mappedimage = gv_stashpv ("OpenCL::MappedImage", GV_ADD);
724
649 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt); 725 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt);
650} 726}
651 727
652cl_int 728cl_int
653errno () 729errno ()
655 RETVAL = res; 731 RETVAL = res;
656 OUTPUT: 732 OUTPUT:
657 RETVAL 733 RETVAL
658 734
659const char * 735const char *
660err2str (cl_int err) 736err2str (cl_int err = res)
661 737
662const char * 738const char *
663enum2str (cl_uint value) 739enum2str (cl_uint value)
664 740
665void 741void
673 list = tmpbuf (sizeof (*list) * count); 749 list = tmpbuf (sizeof (*list) * count);
674 NEED_SUCCESS (GetPlatformIDs, (count, list, 0)); 750 NEED_SUCCESS (GetPlatformIDs, (count, list, 0));
675 751
676 EXTEND (SP, count); 752 EXTEND (SP, count);
677 for (i = 0; i < count; ++i) 753 for (i = 0; i < count; ++i)
678 PUSH_CLOBJ ("OpenCL::Platform", list [i]); 754 PUSH_CLOBJ (stash_platform, (IV)list [i]);
679 755
680void 756void
681context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef) 757context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
682 PPCODE: 758 PPCODE:
683 CONTEXT_NOTIFY_CALLBACK; 759 CONTEXT_NOTIFY_CALLBACK;
690 /* der Gipfel der Kunst */ 766 /* der Gipfel der Kunst */
691 767
692void 768void
693wait_for_events (...) 769wait_for_events (...)
694 CODE: 770 CODE:
695 EVENT_LIST (0, items); 771 EVENT_LIST (0);
696 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr)); 772 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr));
697 773
698PROTOTYPES: DISABLE 774PROTOTYPES: DISABLE
699 775
700MODULE = OpenCL PACKAGE = OpenCL::Platform 776MODULE = OpenCL PACKAGE = OpenCL::Platform
743 list = tmpbuf (sizeof (*list) * count); 819 list = tmpbuf (sizeof (*list) * count);
744 NEED_SUCCESS (GetDeviceIDs, (self, type, count, list, 0)); 820 NEED_SUCCESS (GetDeviceIDs, (self, type, count, list, 0));
745 821
746 EXTEND (SP, count); 822 EXTEND (SP, count);
747 for (i = 0; i < count; ++i) 823 for (i = 0; i < count; ++i)
748 PUSHs (sv_setref_pv (sv_newmortal (), "OpenCL::Device", list [i])); 824 PUSH_CLOBJ (stash_device, list [i]);
749 825
750void 826void
751context (OpenCL::Platform self, cl_context_properties *properties, SV *devices, SV *notify = &PL_sv_undef) 827context (OpenCL::Platform self, cl_context_properties *properties, SV *devices, SV *notify = &PL_sv_undef)
752 PPCODE: 828 PPCODE:
753 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV) 829 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV)
789 PPCODE: 865 PPCODE:
790 cl_device_type value [1]; 866 cl_device_type value [1];
791 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_TYPE, sizeof (value), value, 0)); 867 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_TYPE, sizeof (value), value, 0));
792 EXTEND (SP, 1); 868 EXTEND (SP, 1);
793 const int i = 0; 869 const int i = 0;
794 PUSHs (sv_2mortal (newSViv (value [i]))); 870 PUSHs (sv_2mortal (newSVuv (value [i])));
795 871
796void 872void
797vendor_id (OpenCL::Device self) 873vendor_id (OpenCL::Device self)
798 ALIAS: 874 ALIAS:
799 vendor_id = CL_DEVICE_VENDOR_ID 875 vendor_id = CL_DEVICE_VENDOR_ID
944 PPCODE: 1020 PPCODE:
945 cl_command_queue_properties value [1]; 1021 cl_command_queue_properties value [1];
946 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0)); 1022 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0));
947 EXTEND (SP, 1); 1023 EXTEND (SP, 1);
948 const int i = 0; 1024 const int i = 0;
949 PUSHs (sv_2mortal (newSViv (value [i]))); 1025 PUSHs (sv_2mortal (newSVuv (value [i])));
950 1026
951void 1027void
952platform (OpenCL::Device self) 1028platform (OpenCL::Device self)
953 PPCODE: 1029 PPCODE:
954 cl_platform_id value [1]; 1030 cl_platform_id value [1];
955 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0)); 1031 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0));
956 EXTEND (SP, 1); 1032 EXTEND (SP, 1);
957 const int i = 0; 1033 const int i = 0;
958 { 1034 {
959 PUSH_CLOBJ ("OpenCL::Platform", value [i]); 1035 PUSH_CLOBJ (stash_platform, value [i]);
960 } 1036 }
961 1037
962void 1038void
963name (OpenCL::Device self) 1039name (OpenCL::Device self)
964 ALIAS: 1040 ALIAS:
983 cl_device_id value [1]; 1059 cl_device_id value [1];
984 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0)); 1060 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0));
985 EXTEND (SP, 1); 1061 EXTEND (SP, 1);
986 const int i = 0; 1062 const int i = 0;
987 { 1063 {
988 PUSH_CLOBJ ("OpenCL::Device", value [i]); 1064 PUSH_CLOBJ (stash_device, value [i]);
989 } 1065 }
990 1066
991void 1067void
992partition_types_ext (OpenCL::Device self) 1068partition_types_ext (OpenCL::Device self)
993 ALIAS: 1069 ALIAS:
1020 1096
1021void 1097void
1022queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0) 1098queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0)
1023 PPCODE: 1099 PPCODE:
1024 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res)); 1100 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res));
1025 XPUSH_CLOBJ ("OpenCL::Queue", queue); 1101 XPUSH_CLOBJ (stash_queue, queue);
1026 1102
1027void 1103void
1028user_event (OpenCL::Context self) 1104user_event (OpenCL::Context self)
1029 PPCODE: 1105 PPCODE:
1030 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res)); 1106 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res));
1031 XPUSH_CLOBJ ("OpenCL::UserEvent", ev); 1107 XPUSH_CLOBJ (stash_userevent, ev);
1032 1108
1033void 1109void
1034buffer (OpenCL::Context self, cl_mem_flags flags, size_t len) 1110buffer (OpenCL::Context self, cl_mem_flags flags, size_t len)
1035 PPCODE: 1111 PPCODE:
1036 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) 1112 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))
1037 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?"); 1113 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?");
1038 1114
1039 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res)); 1115 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res));
1040 XPUSH_CLOBJ ("OpenCL::BufferObj", mem); 1116 XPUSH_CLOBJ (stash_bufferobj, mem);
1041 1117
1042void 1118void
1043buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data) 1119buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data)
1044 PPCODE: 1120 PPCODE:
1045 STRLEN len; 1121 STRLEN len;
1046 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0; 1122 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
1047 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) 1123 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)))
1048 croak ("OpenCL::Context::buffer_sv: you have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?"); 1124 croak ("OpenCL::Context::buffer_sv: you have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?");
1049 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res)); 1125 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res));
1050 XPUSH_CLOBJ ("OpenCL::BufferObj", mem); 1126 XPUSH_CLOBJ (stash_bufferobj, mem);
1051 1127
1052#if CL_VERSION_1_2 1128#if CL_VERSION_1_2
1053 1129
1054void 1130void
1055image (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) 1131image (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)
1056 PPCODE: 1132 PPCODE:
1057 STRLEN len; 1133 STRLEN len;
1058 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0; 1134 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
1059 const cl_image_format format = { channel_order, channel_type }; 1135 const cl_image_format format = { channel_order, channel_type };
1060 const cl_image_desc desc = { 1136 const cl_image_desc desc = {
1063 array_size, row_pitch, slice_pitch, 1139 array_size, row_pitch, slice_pitch,
1064 num_mip_level, num_samples, 1140 num_mip_level, num_samples,
1065 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ ("OpenCL::Context::Image", "data", data, "OpenCL::Buffer") : 0 1141 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ ("OpenCL::Context::Image", "data", data, "OpenCL::Buffer") : 0
1066 }; 1142 };
1067 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res)); 1143 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1068 char *klass = "OpenCL::Image"; 1144 HV *stash = stash_image;
1069 switch (type) 1145 switch (type)
1070 { 1146 {
1071 case CL_MEM_OBJECT_IMAGE1D_BUFFER: klass = "OpenCL::Image1DBuffer"; break; 1147 case CL_MEM_OBJECT_IMAGE1D_BUFFER: stash = stash_image1dbuffer; break;
1072 case CL_MEM_OBJECT_IMAGE1D: klass = "OpenCL::Image1D"; break; 1148 case CL_MEM_OBJECT_IMAGE1D: stash = stash_image1d; break;
1073 case CL_MEM_OBJECT_IMAGE1D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1149 case CL_MEM_OBJECT_IMAGE1D_ARRAY: stash = stash_image2darray; break;
1074 case CL_MEM_OBJECT_IMAGE2D: klass = "OpenCL::Image2D"; break; 1150 case CL_MEM_OBJECT_IMAGE2D: stash = stash_image2d; break;
1075 case CL_MEM_OBJECT_IMAGE2D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1151 case CL_MEM_OBJECT_IMAGE2D_ARRAY: stash = stash_image2darray; break;
1076 case CL_MEM_OBJECT_IMAGE3D: klass = "OpenCL::Image3D"; break; 1152 case CL_MEM_OBJECT_IMAGE3D: stash = stash_image3d; break;
1077 } 1153 }
1078 XPUSH_CLOBJ (klass, mem); 1154 XPUSH_CLOBJ (stash, mem);
1079 1155
1080#endif 1156#endif
1081 1157
1082void 1158void
1083image2d (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) 1159image2d (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)
1089 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res)); 1165 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res));
1090#else 1166#else
1091 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, row_pitch, 0, 0, 0, 0 }; 1167 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, row_pitch, 0, 0, 0, 0 };
1092 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res)); 1168 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1093#endif 1169#endif
1094 XPUSH_CLOBJ ("OpenCL::Image2D", mem); 1170 XPUSH_CLOBJ (stash_image2d, mem);
1095 1171
1096void 1172void
1097image3d (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) 1173image3d (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)
1098 PPCODE: 1174 PPCODE:
1099 STRLEN len; 1175 STRLEN len;
1103 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res)); 1179 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res));
1104#else 1180#else
1105 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, row_pitch, slice_pitch, 0, 0, 0 }; 1181 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, row_pitch, slice_pitch, 0, 0, 0 };
1106 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res)); 1182 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1107#endif 1183#endif
1108 XPUSH_CLOBJ ("OpenCL::Image3D", mem); 1184 XPUSH_CLOBJ (stash_image3d, mem);
1109 1185
1110#if cl_apple_gl_sharing || cl_khr_gl_sharing 1186#if cl_apple_gl_sharing || cl_khr_gl_sharing
1111 1187
1112void 1188void
1113gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj) 1189gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj)
1114 PPCODE: 1190 PPCODE:
1115 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res)); 1191 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res));
1116 XPUSH_CLOBJ ("OpenCL::BufferObj", mem); 1192 XPUSH_CLOBJ (stash_bufferobj, mem);
1117 1193
1118void 1194void
1119gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer) 1195gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer)
1120 PPCODE: 1196 PPCODE:
1121 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res)); 1197 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res));
1122 XPUSH_CLOBJ ("OpenCL::Image2D", mem); 1198 XPUSH_CLOBJ (stash_image2d, mem);
1123 1199
1124#if CL_VERSION_1_2 1200#if CL_VERSION_1_2
1125 1201
1126void 1202void
1127gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1203gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1128 ALIAS: 1204 ALIAS:
1129 PPCODE: 1205 PPCODE:
1130 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture, (self, flags, target, miplevel, texture, &res)); 1206 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture, (self, flags, target, miplevel, texture, &res));
1131 cl_gl_object_type type; 1207 cl_gl_object_type type;
1132 NEED_SUCCESS (GetGLObjectInfo, (mem, &type, 0)); // TODO: use target instead? 1208 NEED_SUCCESS (GetGLObjectInfo, (mem, &type, 0)); // TODO: use target instead?
1133 char *klass = "OpenCL::Memory"; 1209 HV *stash = stash_memory;
1134 switch (type) 1210 switch (type)
1135 { 1211 {
1136 case CL_GL_OBJECT_TEXTURE_BUFFER: klass = "OpenCL::Image1DBuffer"; break; 1212 case CL_GL_OBJECT_TEXTURE_BUFFER: stash = stash_image1dbuffer; break;
1137 case CL_GL_OBJECT_TEXTURE1D: klass = "OpenCL::Image1D"; break; 1213 case CL_GL_OBJECT_TEXTURE1D: stash = stash_image1d; break;
1138 case CL_GL_OBJECT_TEXTURE1D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1214 case CL_GL_OBJECT_TEXTURE1D_ARRAY: stash = stash_image2darray; break;
1139 case CL_GL_OBJECT_TEXTURE2D: klass = "OpenCL::Image2D"; break; 1215 case CL_GL_OBJECT_TEXTURE2D: stash = stash_image2d; break;
1140 case CL_GL_OBJECT_TEXTURE2D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1216 case CL_GL_OBJECT_TEXTURE2D_ARRAY: stash = stash_image2darray; break;
1141 case CL_GL_OBJECT_TEXTURE3D: klass = "OpenCL::Image3D"; break; 1217 case CL_GL_OBJECT_TEXTURE3D: stash = stash_image3d; break;
1142 } 1218 }
1143 XPUSH_CLOBJ (klass, mem); 1219 XPUSH_CLOBJ (stash, mem);
1144 1220
1145#endif 1221#endif
1146 1222
1147void 1223void
1148gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1224gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1150#if PREFER_1_1 1226#if PREFER_1_1
1151 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res)); 1227 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res));
1152#else 1228#else
1153 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res)); 1229 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1154#endif 1230#endif
1155 XPUSH_CLOBJ ("OpenCL::Image2D", mem); 1231 XPUSH_CLOBJ (stash_image2d, mem);
1156 1232
1157void 1233void
1158gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1234gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1159 PPCODE: 1235 PPCODE:
1160#if PREFER_1_1 1236#if PREFER_1_1
1161 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res)); 1237 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res));
1162#else 1238#else
1163 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res)); 1239 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1164#endif 1240#endif
1165 XPUSH_CLOBJ ("OpenCL::Image3D", mem); 1241 XPUSH_CLOBJ (stash_image3d, mem);
1166 1242
1167#endif 1243#endif
1168 1244
1169void 1245void
1170supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type) 1246supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type)
1190 1266
1191void 1267void
1192sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode) 1268sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode)
1193 PPCODE: 1269 PPCODE:
1194 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res)); 1270 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res));
1195 XPUSH_CLOBJ ("OpenCL::Sampler", sampler); 1271 XPUSH_CLOBJ (stash_sampler, sampler);
1196 1272
1197void 1273void
1198program_with_source (OpenCL::Context self, SV *program) 1274program_with_source (OpenCL::Context self, SV *program)
1199 PPCODE: 1275 PPCODE:
1200 STRLEN len; 1276 STRLEN len;
1201 size_t len2; 1277 size_t len2;
1202 const char *ptr = SvPVbyte (program, len); 1278 const char *ptr = SvPVbyte (program, len);
1203 1279
1204 len2 = len; 1280 len2 = len;
1205 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res)); 1281 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res));
1206 XPUSH_CLOBJ ("OpenCL::Program", prog); 1282 XPUSH_CLOBJ (stash_program, prog);
1207 1283
1208#BEGIN:context 1284#BEGIN:context
1209 1285
1210void 1286void
1211reference_count (OpenCL::Context self) 1287reference_count (OpenCL::Context self)
1228 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0)); 1304 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0));
1229 int i, n = size / sizeof (*value); 1305 int i, n = size / sizeof (*value);
1230 EXTEND (SP, n); 1306 EXTEND (SP, n);
1231 for (i = 0; i < n; ++i) 1307 for (i = 0; i < n; ++i)
1232 { 1308 {
1233 PUSH_CLOBJ ("OpenCL::Device", value [i]); 1309 PUSH_CLOBJ (stash_device, value [i]);
1234 } 1310 }
1235 1311
1236void 1312void
1237properties (OpenCL::Context self) 1313properties (OpenCL::Context self)
1238 PPCODE: 1314 PPCODE:
1253DESTROY (OpenCL::Queue self) 1329DESTROY (OpenCL::Queue self)
1254 CODE: 1330 CODE:
1255 clReleaseCommandQueue (self); 1331 clReleaseCommandQueue (self);
1256 1332
1257void 1333void
1258enqueue_read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...) 1334read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...)
1335 ALIAS:
1336 enqueue_read_buffer = 0
1259 PPCODE: 1337 PPCODE:
1260 cl_event ev = 0; 1338 cl_event ev = 0;
1261 EVENT_LIST (6, items - 6); 1339 EVENT_LIST (6);
1262 1340
1263 SvUPGRADE (data, SVt_PV); 1341 SvUPGRADE (data, SVt_PV);
1264 SvGROW (data, len); 1342 SvGROW (data, len);
1265 SvPOK_only (data); 1343 SvPOK_only (data);
1266 SvCUR_set (data, len); 1344 SvCUR_set (data, len);
1267 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1345 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1268 1346
1269 if (ev) 1347 if (ev)
1270 XPUSH_CLOBJ ("OpenCL::Event", ev); 1348 XPUSH_CLOBJ (stash_event, ev);
1271 1349
1272void 1350void
1273enqueue_write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...) 1351write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...)
1352 ALIAS:
1353 enqueue_write_buffer = 0
1274 PPCODE: 1354 PPCODE:
1275 cl_event ev = 0; 1355 cl_event ev = 0;
1276 STRLEN len; 1356 STRLEN len;
1277 char *ptr = SvPVbyte (data, len); 1357 char *ptr = SvPVbyte (data, len);
1278 EVENT_LIST (5, items - 5); 1358 EVENT_LIST (5);
1279 1359
1280 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1360 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1281 1361
1282 if (ev) 1362 if (ev)
1283 XPUSH_CLOBJ ("OpenCL::Event", ev); 1363 XPUSH_CLOBJ (stash_event, ev);
1284 1364
1285#if CL_VERSION_1_2 1365#if CL_VERSION_1_2
1286 1366
1287void 1367void
1288enqueue_fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...) 1368fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...)
1369 ALIAS:
1370 enqueue_fill_buffer = 0
1289 PPCODE: 1371 PPCODE:
1290 cl_event ev = 0; 1372 cl_event ev = 0;
1291 STRLEN len; 1373 STRLEN len;
1292 char *ptr = SvPVbyte (data, len); 1374 char *ptr = SvPVbyte (data, len);
1293 EVENT_LIST (5, items - 5); 1375 EVENT_LIST (5);
1294 1376
1295 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1377 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1296 1378
1297 if (ev) 1379 if (ev)
1298 XPUSH_CLOBJ ("OpenCL::Event", ev); 1380 XPUSH_CLOBJ (stash_event, ev);
1299 1381
1300void 1382void
1301enqueue_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, ...) 1383fill_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, ...)
1384 ALIAS:
1385 enqueue_fill_image = 0
1302 PPCODE: 1386 PPCODE:
1303 cl_event ev = 0; 1387 cl_event ev = 0;
1304 STRLEN len; 1388 STRLEN len;
1305 const size_t origin [3] = { x, y, z }; 1389 const size_t origin [3] = { x, y, z };
1306 const size_t region [3] = { width, height, depth }; 1390 const size_t region [3] = { width, height, depth };
1307 EVENT_LIST (12, items - 12); 1391 EVENT_LIST (12);
1308 1392
1309 const cl_float c_f [4] = { r, g, b, a }; 1393 const cl_float c_f [4] = { r, g, b, a };
1310 const cl_uint c_u [4] = { r, g, b, a }; 1394 const cl_uint c_u [4] = { r, g, b, a };
1311 const cl_int c_s [4] = { r, g, b, a }; 1395 const cl_int c_s [4] = { r, g, b, a };
1312 const void *c_fus [3] = { &c_f, &c_u, &c_s }; 1396 const void *c_fus [3] = { &c_f, &c_u, &c_s };
1315 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0)); 1399 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
1316 assert (sizeof (fus) == CL_FLOAT + 1 - CL_SNORM_INT8); 1400 assert (sizeof (fus) == CL_FLOAT + 1 - CL_SNORM_INT8);
1317 if (format.image_channel_data_type < CL_SNORM_INT8 || CL_FLOAT < format.image_channel_data_type) 1401 if (format.image_channel_data_type < CL_SNORM_INT8 || CL_FLOAT < format.image_channel_data_type)
1318 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported."); 1402 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported.");
1319 1403
1320 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type]], 1404 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type - CL_SNORM_INT8]],
1321 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1405 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1322 1406
1323 if (ev) 1407 if (ev)
1324 XPUSH_CLOBJ ("OpenCL::Event", ev); 1408 XPUSH_CLOBJ (stash_event, ev);
1325 1409
1326#endif 1410#endif
1327 1411
1328void 1412void
1329enqueue_copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...) 1413copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...)
1414 ALIAS:
1415 enqueue_copy_buffer = 0
1330 PPCODE: 1416 PPCODE:
1331 cl_event ev = 0; 1417 cl_event ev = 0;
1332 EVENT_LIST (6, items - 6); 1418 EVENT_LIST (6);
1333 1419
1334 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1420 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1335 1421
1336 if (ev) 1422 if (ev)
1337 XPUSH_CLOBJ ("OpenCL::Event", ev); 1423 XPUSH_CLOBJ (stash_event, ev);
1338 1424
1339void 1425void
1340enqueue_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, ...) 1426read_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, ...)
1427 ALIAS:
1428 enqueue_read_buffer_rect = 0
1341 PPCODE: 1429 PPCODE:
1342 cl_event ev = 0; 1430 cl_event ev = 0;
1343 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1431 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
1344 const size_t host_origin[3] = { host_x, host_y, host_z }; 1432 const size_t host_origin[3] = { host_x, host_y, host_z };
1345 const size_t region[3] = { width, height, depth }; 1433 const size_t region[3] = { width, height, depth };
1346 EVENT_LIST (17, items - 17); 1434 EVENT_LIST (17);
1347 1435
1348 if (!buf_row_pitch) 1436 if (!buf_row_pitch)
1349 buf_row_pitch = region [0]; 1437 buf_row_pitch = region [0];
1350 1438
1351 if (!buf_slice_pitch) 1439 if (!buf_slice_pitch)
1364 SvPOK_only (data); 1452 SvPOK_only (data);
1365 SvCUR_set (data, len); 1453 SvCUR_set (data, len);
1366 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)); 1454 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));
1367 1455
1368 if (ev) 1456 if (ev)
1369 XPUSH_CLOBJ ("OpenCL::Event", ev); 1457 XPUSH_CLOBJ (stash_event, ev);
1370 1458
1371void 1459void
1372enqueue_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, ...) 1460write_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, ...)
1461 ALIAS:
1462 enqueue_write_buffer_rect = 0
1373 PPCODE: 1463 PPCODE:
1374 cl_event ev = 0; 1464 cl_event ev = 0;
1375 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1465 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
1376 const size_t host_origin[3] = { host_x, host_y, host_z }; 1466 const size_t host_origin[3] = { host_x, host_y, host_z };
1377 const size_t region[3] = { width, height, depth }; 1467 const size_t region[3] = { width, height, depth };
1378 STRLEN len; 1468 STRLEN len;
1379 char *ptr = SvPVbyte (data, len); 1469 char *ptr = SvPVbyte (data, len);
1380 EVENT_LIST (17, items - 17); 1470 EVENT_LIST (17);
1381 1471
1382 if (!buf_row_pitch) 1472 if (!buf_row_pitch)
1383 buf_row_pitch = region [0]; 1473 buf_row_pitch = region [0];
1384 1474
1385 if (!buf_slice_pitch) 1475 if (!buf_slice_pitch)
1397 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1487 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
1398 1488
1399 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)); 1489 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));
1400 1490
1401 if (ev) 1491 if (ev)
1402 XPUSH_CLOBJ ("OpenCL::Event", ev); 1492 XPUSH_CLOBJ (stash_event, ev);
1403 1493
1404void 1494void
1405enqueue_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, ...) 1495copy_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, ...)
1496 ALIAS:
1497 enqueue_copy_buffer_rect = 0
1406 PPCODE: 1498 PPCODE:
1407 cl_event ev = 0; 1499 cl_event ev = 0;
1408 const size_t src_origin[3] = { src_x, src_y, src_z }; 1500 const size_t src_origin[3] = { src_x, src_y, src_z };
1409 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1501 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1410 const size_t region[3] = { width, height, depth }; 1502 const size_t region[3] = { width, height, depth };
1411 EVENT_LIST (16, items - 16); 1503 EVENT_LIST (16);
1412 1504
1413 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)); 1505 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));
1414 1506
1415 if (ev) 1507 if (ev)
1416 XPUSH_CLOBJ ("OpenCL::Event", ev); 1508 XPUSH_CLOBJ (stash_event, ev);
1417 1509
1418void 1510void
1419enqueue_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, ...) 1511read_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, ...)
1512 ALIAS:
1513 enqueue_read_image = 0
1420 PPCODE: 1514 PPCODE:
1421 cl_event ev = 0; 1515 cl_event ev = 0;
1422 const size_t src_origin[3] = { src_x, src_y, src_z }; 1516 const size_t src_origin[3] = { src_x, src_y, src_z };
1423 const size_t region[3] = { width, height, depth }; 1517 const size_t region[3] = { width, height, depth };
1424 EVENT_LIST (12, items - 12); 1518 EVENT_LIST (12);
1425 1519
1426 if (!row_pitch) 1520 if (!row_pitch)
1427 row_pitch = img_row_pitch (src); 1521 row_pitch = img_row_pitch (src);
1428 1522
1429 if (depth > 1 && !slice_pitch) 1523 if (depth > 1 && !slice_pitch)
1436 SvPOK_only (data); 1530 SvPOK_only (data);
1437 SvCUR_set (data, len); 1531 SvCUR_set (data, len);
1438 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)); 1532 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));
1439 1533
1440 if (ev) 1534 if (ev)
1441 XPUSH_CLOBJ ("OpenCL::Event", ev); 1535 XPUSH_CLOBJ (stash_event, ev);
1442 1536
1443void 1537void
1444enqueue_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, ...) 1538write_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, ...)
1539 ALIAS:
1540 enqueue_write_image = 0
1445 PPCODE: 1541 PPCODE:
1446 cl_event ev = 0; 1542 cl_event ev = 0;
1447 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1543 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1448 const size_t region[3] = { width, height, depth }; 1544 const size_t region[3] = { width, height, depth };
1449 STRLEN len; 1545 STRLEN len;
1450 char *ptr = SvPVbyte (data, len); 1546 char *ptr = SvPVbyte (data, len);
1451 EVENT_LIST (12, items - 12); 1547 EVENT_LIST (12);
1452 1548
1453 if (!row_pitch) 1549 if (!row_pitch)
1454 row_pitch = img_row_pitch (dst); 1550 row_pitch = img_row_pitch (dst);
1455 1551
1456 if (depth > 1 && !slice_pitch) 1552 if (depth > 1 && !slice_pitch)
1462 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1558 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
1463 1559
1464 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)); 1560 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));
1465 1561
1466 if (ev) 1562 if (ev)
1467 XPUSH_CLOBJ ("OpenCL::Event", ev); 1563 XPUSH_CLOBJ (stash_event, ev);
1468 1564
1469void 1565void
1470enqueue_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, ...) 1566copy_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, ...)
1567 ALIAS:
1568 enqueue_copy_image = 0
1471 PPCODE: 1569 PPCODE:
1472 cl_event ev = 0; 1570 cl_event ev = 0;
1473 const size_t src_origin[3] = { src_x, src_y, src_z }; 1571 const size_t src_origin[3] = { src_x, src_y, src_z };
1474 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1572 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1475 const size_t region[3] = { width, height, depth }; 1573 const size_t region[3] = { width, height, depth };
1476 EVENT_LIST (12, items - 12); 1574 EVENT_LIST (12);
1477 1575
1478 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1576 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1479 1577
1480 if (ev) 1578 if (ev)
1481 XPUSH_CLOBJ ("OpenCL::Event", ev); 1579 XPUSH_CLOBJ (stash_event, ev);
1482 1580
1483void 1581void
1484enqueue_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, ...) 1582copy_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, ...)
1583 ALIAS:
1584 enqueue_copy_image_to_buffer = 0
1485 PPCODE: 1585 PPCODE:
1486 cl_event ev = 0; 1586 cl_event ev = 0;
1487 const size_t src_origin[3] = { src_x, src_y, src_z }; 1587 const size_t src_origin[3] = { src_x, src_y, src_z };
1588 const size_t region [3] = { width, height, depth };
1589 EVENT_LIST (10);
1590
1591 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1592
1593 if (ev)
1594 XPUSH_CLOBJ (stash_event, ev);
1595
1596void
1597copy_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, ...)
1598 ALIAS:
1599 enqueue_copy_buffer_to_image = 0
1600 PPCODE:
1601 cl_event ev = 0;
1602 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1603 const size_t region [3] = { width, height, depth };
1604 EVENT_LIST (10);
1605
1606 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1607
1608 if (ev)
1609 XPUSH_CLOBJ (stash_event, ev);
1610
1611void
1612map_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, size_t cb = 0, ...)
1613 ALIAS:
1614 enqueue_map_buffer = 0
1615 PPCODE:
1616 cl_event ev;
1617 EVENT_LIST (7);
1618
1619 NEED_SUCCESS_ARG (void *ptr, EnqueueMapBuffer, (self, buf, blocking, map_flags, offset, cb, event_list_count, event_list_ptr, &ev, &res));
1620 XPUSHs (mapped_new (stash_mappedbuffer, self, buf, map_flags, ptr, cb, ev, 0, 0));
1621
1622void
1623map_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, size_t width = 0, size_t height = 0, size_t depth = 0, ...)
1624 ALIAS:
1625 enqueue_map_image = 0
1626 PPCODE:
1627 cl_event ev;
1628 const size_t origin[3] = { x, y, z };
1488 const size_t region[3] = { width, height, depth }; 1629 const size_t region[3] = { width, height, depth };
1630 size_t row_pitch, slice_pitch;
1489 EVENT_LIST (10, items - 10); 1631 EVENT_LIST (11);
1490 1632
1633 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));
1634
1635 size_t cb = slice_pitch ? slice_pitch * region [2]
1636 : row_pitch ? row_pitch * region [1]
1637 : region [0];
1638
1639 XPUSHs (mapped_new (stash_mappedimage, self, img, map_flags, ptr, cb, ev, row_pitch, slice_pitch));
1640
1641void
1642unmap (OpenCL::Queue self, OpenCL::Mapped mapped, ...)
1643 PPCODE:
1644 cl_uint event_list_count = items - 2;
1645 cl_event *event_list_ptr = event_list (&ST (2), &event_list_count, mapped->event);
1646
1647 NEED_SUCCESS (EnqueueUnmapMemObject, (self, mapped->memobj, mapped->ptr, event_list_count, event_list_ptr, &mapped->event));
1648 mapped_detach (ST (1), mapped);
1649
1650 if (GIMME_V != G_VOID)
1651 XPUSH_CLOBJ (stash_event, mapped->event);
1652
1653void
1654task (OpenCL::Queue self, OpenCL::Kernel kernel, ...)
1655 ALIAS:
1656 enqueue_task = 0
1657 PPCODE:
1658 cl_event ev = 0;
1659 EVENT_LIST (2);
1660
1491 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1661 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1492 1662
1493 if (ev) 1663 if (ev)
1494 XPUSH_CLOBJ ("OpenCL::Event", ev); 1664 XPUSH_CLOBJ (stash_event, ev);
1495 1665
1496void 1666void
1497enqueue_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, ...)
1498 PPCODE:
1499 cl_event ev = 0;
1500 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1501 const size_t region[3] = { width, height, depth };
1502 EVENT_LIST (10, items - 10);
1503
1504 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1505
1506 if (ev)
1507 XPUSH_CLOBJ ("OpenCL::Event", ev);
1508
1509void
1510enqueue_task (OpenCL::Queue self, OpenCL::Kernel kernel, ...)
1511 PPCODE:
1512 cl_event ev = 0;
1513 EVENT_LIST (2, items - 2);
1514
1515 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1516
1517 if (ev)
1518 XPUSH_CLOBJ ("OpenCL::Event", ev);
1519
1520void
1521enqueue_nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...) 1667nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...)
1668 ALIAS:
1669 enqueue_nd_range_kernel = 0
1522 PPCODE: 1670 PPCODE:
1523 cl_event ev = 0; 1671 cl_event ev = 0;
1524 size_t *gwo = 0, *gws, *lws = 0; 1672 size_t *gwo = 0, *gws, *lws = 0;
1525 int gws_len; 1673 int gws_len;
1526 size_t *lists; 1674 size_t *lists;
1527 int i; 1675 int i;
1528 EVENT_LIST (5, items - 5); 1676 EVENT_LIST (5);
1529 1677
1530 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV) 1678 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV)
1531 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference"); 1679 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference");
1532 1680
1533 gws_len = AvFILLp (SvRV (global_work_size)) + 1; 1681 gws_len = AvFILLp (SvRV (global_work_size)) + 1;
1534 1682
1535 lists = tmpbuf (sizeof (size_t) * 3 * gws_len); 1683 lists = tmpbuf (sizeof (size_t) * 3 * gws_len);
1536 1684
1537 gws = lists + gws_len * 0; 1685 gws = lists + gws_len * 0;
1538 for (i = 0; i < gws_len; ++i) 1686 for (i = 0; i < gws_len; ++i)
1687 {
1539 gws [i] = SvIV (AvARRAY (SvRV (global_work_size))[i]); 1688 gws [i] = SvIV (AvARRAY (SvRV (global_work_size))[i]);
1689 // at least nvidia crashes for 0-sized work group sizes, work around
1690 if (!gws [i])
1691 croak ("clEnqueueNDRangeKernel: global_work_size[%d] is zero, must be non-zero", i);
1692 }
1540 1693
1541 if (SvOK (global_work_offset)) 1694 if (SvOK (global_work_offset))
1542 { 1695 {
1543 if (!SvROK (global_work_offset) || SvTYPE (SvRV (global_work_offset)) != SVt_PVAV) 1696 if (!SvROK (global_work_offset) || SvTYPE (SvRV (global_work_offset)) != SVt_PVAV)
1544 croak ("clEnqueueNDRangeKernel: global_work_offset must be undef or an array reference"); 1697 croak ("clEnqueueNDRangeKernel: global_work_offset must be undef or an array reference");
1552 } 1705 }
1553 1706
1554 if (SvOK (local_work_size)) 1707 if (SvOK (local_work_size))
1555 { 1708 {
1556 if ((SvOK (local_work_size) && !SvROK (local_work_size)) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV) 1709 if ((SvOK (local_work_size) && !SvROK (local_work_size)) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV)
1557 croak ("clEnqueueNDRangeKernel: global_work_size must be undef or an array reference"); 1710 croak ("clEnqueueNDRangeKernel: local_work_size must be undef or an array reference");
1558 1711
1559 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len) 1712 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len)
1560 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size"); 1713 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size");
1561 1714
1562 lws = lists + gws_len * 2; 1715 lws = lists + gws_len * 2;
1563 for (i = 0; i < gws_len; ++i) 1716 for (i = 0; i < gws_len; ++i)
1717 {
1564 lws [i] = SvIV (AvARRAY (SvRV (local_work_size))[i]); 1718 lws [i] = SvIV (AvARRAY (SvRV (local_work_size))[i]);
1719 // at least nvidia crashes for 0-sized work group sizes, work around
1720 if (!lws [i])
1721 croak ("clEnqueueNDRangeKernel: local_work_size[%d] is zero, must be non-zero", i);
1722 }
1565 } 1723 }
1566 1724
1567 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1725 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1568 1726
1569 if (ev) 1727 if (ev)
1570 XPUSH_CLOBJ ("OpenCL::Event", ev); 1728 XPUSH_CLOBJ (stash_event, ev);
1571 1729
1572#if cl_apple_gl_sharing || cl_khr_gl_sharing 1730#if cl_apple_gl_sharing || cl_khr_gl_sharing
1573 1731
1574void 1732void
1575enqueue_acquire_gl_objects (OpenCL::Queue self, SV *objects, ...) 1733acquire_gl_objects (OpenCL::Queue self, SV *objects, ...)
1576 ALIAS: 1734 ALIAS:
1735 release_gl_objects = 1
1736 enqueue_acquire_gl_objects = 0
1577 enqueue_release_gl_objects = 1 1737 enqueue_release_gl_objects = 1
1578 PPCODE: 1738 PPCODE:
1579 if (!SvROK (objects) || SvTYPE (SvRV (objects)) != SVt_PVAV) 1739 if (!SvROK (objects) || SvTYPE (SvRV (objects)) != SVt_PVAV)
1580 croak ("OpenCL::Queue::enqueue_acquire/release_gl_objects argument 'objects' must be an arrayref with memory objects, in call"); 1740 croak ("OpenCL::Queue::enqueue_acquire/release_gl_objects argument 'objects' must be an arrayref with memory objects, in call");
1581 1741
1582 cl_event ev = 0; 1742 cl_event ev = 0;
1583 EVENT_LIST (2, items - 2); 1743 EVENT_LIST (2);
1584 AV *av = (AV *)SvRV (objects); 1744 AV *av = (AV *)SvRV (objects);
1585 cl_uint num_objects = av_len (av) + 1; 1745 cl_uint num_objects = av_len (av) + 1;
1586 cl_mem *object_list = tmpbuf (sizeof (cl_mem) * num_objects); 1746 cl_mem *object_list = tmpbuf (sizeof (cl_mem) * num_objects);
1587 int i; 1747 int i;
1588 1748
1593 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1753 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1594 else 1754 else
1595 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1755 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1596 1756
1597 if (ev) 1757 if (ev)
1598 XPUSH_CLOBJ ("OpenCL::Event", ev); 1758 XPUSH_CLOBJ (stash_event, ev);
1599 1759
1600#endif 1760#endif
1601 1761
1602void 1762void
1603enqueue_wait_for_events (OpenCL::Queue self, ...) 1763wait_for_events (OpenCL::Queue self, ...)
1764 ALIAS:
1765 enqueue_wait_for_events = 0
1604 CODE: 1766 CODE:
1605 EVENT_LIST (1, items - 1); 1767 EVENT_LIST (1);
1606#if PREFER_1_1 1768#if PREFER_1_1
1607 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr)); 1769 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
1608#else 1770#else
1609 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0)); 1771 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0));
1610#endif 1772#endif
1611 1773
1612void 1774void
1613enqueue_marker (OpenCL::Queue self, ...) 1775marker (OpenCL::Queue self, ...)
1776 ALIAS:
1777 enqueue_marker = 0
1614 PPCODE: 1778 PPCODE:
1615 cl_event ev = 0; 1779 cl_event ev = 0;
1616 EVENT_LIST (1, items - 1); 1780 EVENT_LIST (1);
1617#if PREFER_1_1 1781#if PREFER_1_1
1618 if (!event_list_count) 1782 if (!event_list_count)
1619 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0)); 1783 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
1620 else 1784 else
1621#if CL_VERSION_1_2 1785#if CL_VERSION_1_2
1628#endif 1792#endif
1629#else 1793#else
1630 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1794 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1631#endif 1795#endif
1632 if (ev) 1796 if (ev)
1633 XPUSH_CLOBJ ("OpenCL::Event", ev); 1797 XPUSH_CLOBJ (stash_event, ev);
1634 1798
1635void 1799void
1636enqueue_barrier (OpenCL::Queue self, ...) 1800barrier (OpenCL::Queue self, ...)
1801 ALIAS:
1802 enqueue_barrier = 0
1637 PPCODE: 1803 PPCODE:
1638 cl_event ev = 0; 1804 cl_event ev = 0;
1639 EVENT_LIST (1, items - 1); 1805 EVENT_LIST (1);
1640#if PREFER_1_1 1806#if PREFER_1_1
1641 if (!event_list_count && GIMME_V == G_VOID) 1807 if (!event_list_count && GIMME_V == G_VOID)
1642 NEED_SUCCESS (EnqueueBarrier, (self)); 1808 NEED_SUCCESS (EnqueueBarrier, (self));
1643 else 1809 else
1644#if CL_VERSION_1_2 1810#if CL_VERSION_1_2
1654#endif 1820#endif
1655#else 1821#else
1656 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1822 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1657#endif 1823#endif
1658 if (ev) 1824 if (ev)
1659 XPUSH_CLOBJ ("OpenCL::Event", ev); 1825 XPUSH_CLOBJ (stash_event, ev);
1660 1826
1661void 1827void
1662flush (OpenCL::Queue self) 1828flush (OpenCL::Queue self)
1663 CODE: 1829 CODE:
1664 NEED_SUCCESS (Flush, (self)); 1830 NEED_SUCCESS (Flush, (self));
1682 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0)); 1848 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0));
1683 EXTEND (SP, 1); 1849 EXTEND (SP, 1);
1684 const int i = 0; 1850 const int i = 0;
1685 { 1851 {
1686 NEED_SUCCESS (RetainContext, (value [i])); 1852 NEED_SUCCESS (RetainContext, (value [i]));
1687 PUSH_CLOBJ ("OpenCL::Context", value [i]); 1853 PUSH_CLOBJ (stash_context, value [i]);
1688 } 1854 }
1689 1855
1690void 1856void
1691device (OpenCL::Queue self) 1857device (OpenCL::Queue self)
1692 PPCODE: 1858 PPCODE:
1693 cl_device_id value [1]; 1859 cl_device_id value [1];
1694 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0)); 1860 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0));
1695 EXTEND (SP, 1); 1861 EXTEND (SP, 1);
1696 const int i = 0; 1862 const int i = 0;
1697 { 1863 {
1698 PUSH_CLOBJ ("OpenCL::Device", value [i]); 1864 PUSH_CLOBJ (stash_device, value [i]);
1699 } 1865 }
1700 1866
1701void 1867void
1702reference_count (OpenCL::Queue self) 1868reference_count (OpenCL::Queue self)
1703 PPCODE: 1869 PPCODE:
1712 PPCODE: 1878 PPCODE:
1713 cl_command_queue_properties value [1]; 1879 cl_command_queue_properties value [1];
1714 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_PROPERTIES, sizeof (value), value, 0)); 1880 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_PROPERTIES, sizeof (value), value, 0));
1715 EXTEND (SP, 1); 1881 EXTEND (SP, 1);
1716 const int i = 0; 1882 const int i = 0;
1717 PUSHs (sv_2mortal (newSViv (value [i]))); 1883 PUSHs (sv_2mortal (newSVuv (value [i])));
1718 1884
1719#END:command_queue 1885#END:command_queue
1720 1886
1721MODULE = OpenCL PACKAGE = OpenCL::Memory 1887MODULE = OpenCL PACKAGE = OpenCL::Memory
1722 1888
1737 PPCODE: 1903 PPCODE:
1738 cl_mem_object_type value [1]; 1904 cl_mem_object_type value [1];
1739 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_TYPE, sizeof (value), value, 0)); 1905 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_TYPE, sizeof (value), value, 0));
1740 EXTEND (SP, 1); 1906 EXTEND (SP, 1);
1741 const int i = 0; 1907 const int i = 0;
1742 PUSHs (sv_2mortal (newSViv (value [i]))); 1908 PUSHs (sv_2mortal (newSVuv (value [i])));
1743 1909
1744void 1910void
1745flags (OpenCL::Memory self) 1911flags (OpenCL::Memory self)
1746 PPCODE: 1912 PPCODE:
1747 cl_mem_flags value [1]; 1913 cl_mem_flags value [1];
1748 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_FLAGS, sizeof (value), value, 0)); 1914 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_FLAGS, sizeof (value), value, 0));
1749 EXTEND (SP, 1); 1915 EXTEND (SP, 1);
1750 const int i = 0; 1916 const int i = 0;
1751 PUSHs (sv_2mortal (newSViv (value [i]))); 1917 PUSHs (sv_2mortal (newSVuv (value [i])));
1752 1918
1753void 1919void
1754size (OpenCL::Memory self) 1920size (OpenCL::Memory self)
1755 ALIAS: 1921 ALIAS:
1756 size = CL_MEM_SIZE 1922 size = CL_MEM_SIZE
1790 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0)); 1956 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0));
1791 EXTEND (SP, 1); 1957 EXTEND (SP, 1);
1792 const int i = 0; 1958 const int i = 0;
1793 { 1959 {
1794 NEED_SUCCESS (RetainContext, (value [i])); 1960 NEED_SUCCESS (RetainContext, (value [i]));
1795 PUSH_CLOBJ ("OpenCL::Context", value [i]); 1961 PUSH_CLOBJ (stash_context, value [i]);
1796 } 1962 }
1797 1963
1798void 1964void
1799associated_memobject (OpenCL::Memory self) 1965associated_memobject (OpenCL::Memory self)
1800 PPCODE: 1966 PPCODE:
1802 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0)); 1968 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0));
1803 EXTEND (SP, 1); 1969 EXTEND (SP, 1);
1804 const int i = 0; 1970 const int i = 0;
1805 { 1971 {
1806 NEED_SUCCESS (RetainMemObject, (value [i])); 1972 NEED_SUCCESS (RetainMemObject, (value [i]));
1807 PUSH_CLOBJ ("OpenCL::Memory", value [i]); 1973 PUSH_CLOBJ (stash_memory, value [i]);
1808 } 1974 }
1809 1975
1810#END:mem 1976#END:mem
1811 1977
1812#if cl_apple_gl_sharing || cl_khr_gl_sharing 1978#if cl_apple_gl_sharing || cl_khr_gl_sharing
1832 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!"); 1998 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!");
1833 1999
1834 cl_buffer_region crdata = { origin, size }; 2000 cl_buffer_region crdata = { origin, size };
1835 2001
1836 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res)); 2002 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res));
1837 XPUSH_CLOBJ ("OpenCL::Buffer", mem); 2003 XPUSH_CLOBJ (stash_buffer, mem);
1838 2004
1839MODULE = OpenCL PACKAGE = OpenCL::Image 2005MODULE = OpenCL PACKAGE = OpenCL::Image
1840 2006
1841void 2007void
1842image_info (OpenCL::Image self, cl_image_info name) 2008image_info (OpenCL::Image self, cl_image_info name)
1928 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0)); 2094 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0));
1929 EXTEND (SP, 1); 2095 EXTEND (SP, 1);
1930 const int i = 0; 2096 const int i = 0;
1931 { 2097 {
1932 NEED_SUCCESS (RetainContext, (value [i])); 2098 NEED_SUCCESS (RetainContext, (value [i]));
1933 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2099 PUSH_CLOBJ (stash_context, value [i]);
1934 } 2100 }
1935 2101
1936void 2102void
1937normalized_coords (OpenCL::Sampler self) 2103normalized_coords (OpenCL::Sampler self)
1938 PPCODE: 2104 PPCODE:
1939 cl_addressing_mode value [1]; 2105 cl_addressing_mode value [1];
1940 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0)); 2106 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0));
1941 EXTEND (SP, 1); 2107 EXTEND (SP, 1);
1942 const int i = 0; 2108 const int i = 0;
1943 PUSHs (sv_2mortal (newSViv (value [i]))); 2109 PUSHs (sv_2mortal (newSVuv (value [i])));
1944 2110
1945void 2111void
1946addressing_mode (OpenCL::Sampler self) 2112addressing_mode (OpenCL::Sampler self)
1947 PPCODE: 2113 PPCODE:
1948 cl_filter_mode value [1]; 2114 cl_filter_mode value [1];
1949 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0)); 2115 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0));
1950 EXTEND (SP, 1); 2116 EXTEND (SP, 1);
1951 const int i = 0; 2117 const int i = 0;
1952 PUSHs (sv_2mortal (newSViv (value [i]))); 2118 PUSHs (sv_2mortal (newSVuv (value [i])));
1953 2119
1954void 2120void
1955filter_mode (OpenCL::Sampler self) 2121filter_mode (OpenCL::Sampler self)
1956 PPCODE: 2122 PPCODE:
1957 cl_bool value [1]; 2123 cl_bool value [1];
2049 2215
2050void 2216void
2051kernel (OpenCL::Program program, SV *function) 2217kernel (OpenCL::Program program, SV *function)
2052 PPCODE: 2218 PPCODE:
2053 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res)); 2219 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res));
2054 XPUSH_CLOBJ ("OpenCL::Kernel", kernel); 2220 XPUSH_CLOBJ (stash_kernel, kernel);
2055 2221
2056void 2222void
2057kernels_in_program (OpenCL::Program program) 2223kernels_in_program (OpenCL::Program program)
2058 PPCODE: 2224 PPCODE:
2059 cl_uint num_kernels; 2225 cl_uint num_kernels;
2062 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0)); 2228 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0));
2063 2229
2064 int i; 2230 int i;
2065 EXTEND (SP, num_kernels); 2231 EXTEND (SP, num_kernels);
2066 for (i = 0; i < num_kernels; ++i) 2232 for (i = 0; i < num_kernels; ++i)
2067 PUSH_CLOBJ ("OpenCL::Kernel", kernels [i]); 2233 PUSH_CLOBJ (stash_kernel, kernels [i]);
2068 2234
2069void 2235void
2070info (OpenCL::Program self, cl_program_info name) 2236info (OpenCL::Program self, cl_program_info name)
2071 PPCODE: 2237 PPCODE:
2072 INFO (Program) 2238 INFO (Program)
2120 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0)); 2286 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0));
2121 EXTEND (SP, 1); 2287 EXTEND (SP, 1);
2122 const int i = 0; 2288 const int i = 0;
2123 { 2289 {
2124 NEED_SUCCESS (RetainContext, (value [i])); 2290 NEED_SUCCESS (RetainContext, (value [i]));
2125 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2291 PUSH_CLOBJ (stash_context, value [i]);
2126 } 2292 }
2127 2293
2128void 2294void
2129devices (OpenCL::Program self) 2295devices (OpenCL::Program self)
2130 PPCODE: 2296 PPCODE:
2134 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0)); 2300 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0));
2135 int i, n = size / sizeof (*value); 2301 int i, n = size / sizeof (*value);
2136 EXTEND (SP, n); 2302 EXTEND (SP, n);
2137 for (i = 0; i < n; ++i) 2303 for (i = 0; i < n; ++i)
2138 { 2304 {
2139 PUSH_CLOBJ ("OpenCL::Device", value [i]); 2305 PUSH_CLOBJ (stash_device, value [i]);
2140 } 2306 }
2141 2307
2142void 2308void
2143source (OpenCL::Program self) 2309source (OpenCL::Program self)
2144 PPCODE: 2310 PPCODE:
2170DESTROY (OpenCL::Kernel self) 2336DESTROY (OpenCL::Kernel self)
2171 CODE: 2337 CODE:
2172 clReleaseKernel (self); 2338 clReleaseKernel (self);
2173 2339
2174void 2340void
2341setf (OpenCL::Kernel self, const char *format, ...)
2342 CODE:
2343 int i;
2344 for (i = 2; ; ++i)
2345 {
2346 while (*format == ' ')
2347 ++format;
2348
2349 char type = *format++;
2350
2351 if (!type)
2352 break;
2353
2354 if (i >= items)
2355 croak ("OpenCL::Kernel::setf format string too long (not enough arguments)");
2356
2357 SV *sv = ST (i);
2358
2359 union
2360 {
2361 cl_char cc; cl_uchar cC; cl_short cs; cl_ushort cS;
2362 cl_int ci; cl_uint cI; cl_long cl; cl_ulong cL;
2363 cl_half ch; cl_float cf; cl_double cd;
2364 cl_mem cm;
2365 cl_sampler ca;
2366 size_t cz;
2367 cl_event ce;
2368 } arg;
2369 size_t size;
2370 int nullarg = 0;
2371
2372 switch (type)
2373 {
2374 case 'c': arg.cc = SvIV (sv); size = sizeof (arg.cc); break;
2375 case 'C': arg.cC = SvUV (sv); size = sizeof (arg.cC); break;
2376 case 's': arg.cs = SvIV (sv); size = sizeof (arg.cs); break;
2377 case 'S': arg.cS = SvUV (sv); size = sizeof (arg.cS); break;
2378 case 'i': arg.ci = SvIV (sv); size = sizeof (arg.ci); break;
2379 case 'I': arg.cI = SvUV (sv); size = sizeof (arg.cI); break;
2380 case 'l': arg.cl = SvIV (sv); size = sizeof (arg.cl); break;
2381 case 'L': arg.cL = SvUV (sv); size = sizeof (arg.cL); break;
2382
2383 case 'h': arg.ch = SvUV (sv); size = sizeof (arg.ch); break;
2384 case 'f': arg.cf = SvNV (sv); size = sizeof (arg.cf); break;
2385 case 'd': arg.cd = SvNV (sv); size = sizeof (arg.cd); break;
2386
2387 case 'z': nullarg = 1; size = SvIV (sv); break;
2388
2389 case 'm': nullarg = !SvOK (sv); arg.cm = SvCLOBJ ("OpenCL::Kernel::setf", "m", sv, "OpenCL::Memory" ); size = sizeof (arg.cm); break;
2390 case 'a': nullarg = !SvOK (sv); arg.ca = SvCLOBJ ("OpenCL::Kernel::setf", "a", sv, "OpenCL::Sampler"); size = sizeof (arg.ca); break;
2391 case 'e': nullarg = !SvOK (sv); arg.ca = SvCLOBJ ("OpenCL::Kernel::setf", "e", sv, "OpenCL::Event" ); size = sizeof (arg.ce); break;
2392
2393 default:
2394 croak ("OpenCL::Kernel::setf format character '%c' not supported", type);
2395 }
2396
2397 res = clSetKernelArg (self, i - 2, size, nullarg ? 0 : &arg);
2398 if (res)
2399 croak ("OpenCL::Kernel::setf kernel parameter '%c' (#%d): %s", type, i - 2, err2str (res));
2400 }
2401
2402 if (i != items)
2403 croak ("OpenCL::Kernel::setf format string too short (too many arguments)");
2404
2405void
2175set_char (OpenCL::Kernel self, cl_uint idx, cl_char value) 2406set_char (OpenCL::Kernel self, cl_uint idx, cl_char value)
2176 CODE: 2407 CODE:
2177 clSetKernelArg (self, idx, sizeof (value), &value); 2408 clSetKernelArg (self, idx, sizeof (value), &value);
2178 2409
2179void 2410void
2227 clSetKernelArg (self, idx, sizeof (value), &value); 2458 clSetKernelArg (self, idx, sizeof (value), &value);
2228 2459
2229void 2460void
2230set_memory (OpenCL::Kernel self, cl_uint idx, OpenCL::Memory_ornull value) 2461set_memory (OpenCL::Kernel self, cl_uint idx, OpenCL::Memory_ornull value)
2231 CODE: 2462 CODE:
2232 clSetKernelArg (self, idx, sizeof (value), &value); 2463 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
2233 2464
2234void 2465void
2235set_buffer (OpenCL::Kernel self, cl_uint idx, OpenCL::Buffer_ornull value) 2466set_buffer (OpenCL::Kernel self, cl_uint idx, OpenCL::Buffer_ornull value)
2236 CODE: 2467 CODE:
2237 clSetKernelArg (self, idx, sizeof (value), &value); 2468 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
2238 2469
2239void 2470void
2240set_image2d (OpenCL::Kernel self, cl_uint idx, OpenCL::Image2D_ornull value) 2471set_image (OpenCL::Kernel self, cl_uint idx, OpenCL::Image_ornull value)
2472 ALIAS:
2473 set_image2d = 0
2474 set_image3d = 0
2241 CODE: 2475 CODE:
2242 clSetKernelArg (self, idx, sizeof (value), &value); 2476 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
2243
2244void
2245set_image3d (OpenCL::Kernel self, cl_uint idx, OpenCL::Image3D_ornull value)
2246 CODE:
2247 clSetKernelArg (self, idx, sizeof (value), &value);
2248 2477
2249void 2478void
2250set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value) 2479set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value)
2251 CODE: 2480 CODE:
2252 clSetKernelArg (self, idx, sizeof (value), &value); 2481 clSetKernelArg (self, idx, sizeof (value), &value);
2298 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0)); 2527 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0));
2299 EXTEND (SP, 1); 2528 EXTEND (SP, 1);
2300 const int i = 0; 2529 const int i = 0;
2301 { 2530 {
2302 NEED_SUCCESS (RetainContext, (value [i])); 2531 NEED_SUCCESS (RetainContext, (value [i]));
2303 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2532 PUSH_CLOBJ (stash_context, value [i]);
2304 } 2533 }
2305 2534
2306void 2535void
2307program (OpenCL::Kernel self) 2536program (OpenCL::Kernel self)
2308 PPCODE: 2537 PPCODE:
2310 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0)); 2539 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0));
2311 EXTEND (SP, 1); 2540 EXTEND (SP, 1);
2312 const int i = 0; 2541 const int i = 0;
2313 { 2542 {
2314 NEED_SUCCESS (RetainProgram, (value [i])); 2543 NEED_SUCCESS (RetainProgram, (value [i]));
2315 PUSH_CLOBJ ("OpenCL::Program", value [i]); 2544 PUSH_CLOBJ (stash_program, value [i]);
2316 } 2545 }
2317 2546
2318#END:kernel 2547#END:kernel
2319 2548
2320void 2549void
2400 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0)); 2629 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0));
2401 EXTEND (SP, 1); 2630 EXTEND (SP, 1);
2402 const int i = 0; 2631 const int i = 0;
2403 { 2632 {
2404 NEED_SUCCESS (RetainCommandQueue, (value [i])); 2633 NEED_SUCCESS (RetainCommandQueue, (value [i]));
2405 PUSH_CLOBJ ("OpenCL::Queue", value [i]); 2634 PUSH_CLOBJ (stash_queue, value [i]);
2406 } 2635 }
2407 2636
2408void 2637void
2409command_type (OpenCL::Event self) 2638command_type (OpenCL::Event self)
2410 PPCODE: 2639 PPCODE:
2433 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0)); 2662 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0));
2434 EXTEND (SP, 1); 2663 EXTEND (SP, 1);
2435 const int i = 0; 2664 const int i = 0;
2436 { 2665 {
2437 NEED_SUCCESS (RetainContext, (value [i])); 2666 NEED_SUCCESS (RetainContext, (value [i]));
2438 PUSH_CLOBJ ("OpenCL::Context", value [i]); 2667 PUSH_CLOBJ (stash_context, value [i]);
2439 } 2668 }
2440 2669
2441#END:event 2670#END:event
2442 2671
2443void 2672void
2468void 2697void
2469set_status (OpenCL::UserEvent self, cl_int execution_status) 2698set_status (OpenCL::UserEvent self, cl_int execution_status)
2470 CODE: 2699 CODE:
2471 clSetUserEventStatus (self, execution_status); 2700 clSetUserEventStatus (self, execution_status);
2472 2701
2702MODULE = OpenCL PACKAGE = OpenCL::Mapped
2703
2704void
2705DESTROY (SV *self)
2706 CODE:
2707 OpenCL__Mapped mapped = SvMAPPED (self);
2708
2709 clEnqueueUnmapMemObject (mapped->queue, mapped->memobj, mapped->ptr, 1, &mapped->event, 0);
2710 mapped_detach (self, mapped);
2711
2712 clReleaseCommandQueue (mapped->queue);
2713
2714 if (mapped->event)
2715 clReleaseEvent (mapped->event);
2716
2717 Safefree (mapped);
2718
2719bool
2720mapped (OpenCL::Mapped self)
2721 CODE:
2722 RETVAL = !!self->ptr;
2723 OUTPUT:
2724 RETVAL
2725
2726void
2727wait (OpenCL::Mapped self)
2728 PPCODE:
2729 if (self->event)
2730 NEED_SUCCESS (WaitForEvents, (1, &self->event));
2731
2732void
2733event (OpenCL::Mapped self)
2734 PPCODE:
2735 if (!self->event)
2736 XSRETURN_UNDEF;
2737
2738 clRetainEvent (self->event);
2739 XPUSH_CLOBJ (stash_event, self->event);
2740
2741size_t
2742size (OpenCL::Mapped self)
2743 CODE:
2744 RETVAL = self->cb;
2745 OUTPUT:
2746 RETVAL
2747
2748IV
2749ptr (OpenCL::Mapped self)
2750 CODE:
2751 RETVAL = PTR2IV (self->ptr);
2752 OUTPUT:
2753 RETVAL
2754
2755MODULE = OpenCL PACKAGE = OpenCL::MappedBuffer
2756
2757MODULE = OpenCL PACKAGE = OpenCL::MappedImage
2758
2759IV
2760row_pitch (OpenCL::Mapped self)
2761 ALIAS:
2762 slice_pitch = 1
2763 CODE:
2764 RETVAL = ix ? self->slice_pitch : self->row_pitch;
2765 OUTPUT:
2766 RETVAL
2767
2768
2769

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines