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

Comparing OpenCL/OpenCL.xs (file contents):
Revision 1.46 by root, Tue Apr 24 12:23:56 2012 UTC vs.
Revision 1.57 by root, Sat Apr 28 13:31:40 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#define X_STACKSIZE sizeof (void *) * 512 * 1024 // 2-4mb should be enough, really
6#include "xthread.h"
7#include "schmorp.h"
4 8
5#ifdef I_DLFCN 9#ifdef I_DLFCN
6 #include <dlfcn.h> 10 #include <dlfcn.h>
7#endif 11#endif
8 12
189 if (res) \ 193 if (res) \
190 FAIL (name); 194 FAIL (name);
191 195
192/*****************************************************************************/ 196/*****************************************************************************/
193 197
194#define NEW_MORTAL_OBJ(class,ptr) sv_setref_pv (sv_newmortal (), class, ptr)
195#define XPUSH_NEW_OBJ(class,ptr) XPUSHs (NEW_MORTAL_OBJ (class, ptr))
196
197static void *
198SvPTROBJ (const char *func, const char *svname, SV *sv, const char *pkg)
199{
200 if (SvROK (sv) && sv_derived_from (sv, pkg))
201 return (void *)SvIV (SvRV (sv));
202
203 croak ("%s: %s is not of type %s", func, svname, pkg);
204}
205
206/*****************************************************************************/
207
208static cl_context_properties * 198static cl_context_properties *
209SvCONTEXTPROPERTIES (const char *func, const char *svname, SV *sv, cl_context_properties *extra, int extracount) 199SvCONTEXTPROPERTIES (const char *func, const char *svname, SV *sv, cl_context_properties *extra, int extracount)
210{ 200{
211 if (!sv || !SvOK (sv)) 201 if (!sv || !SvOK (sv))
212 if (extra) 202 if (extra)
270 croak ("%s: %s is not a property list (either undef or [type => value, ...])", func, svname); 260 croak ("%s: %s is not a property list (either undef or [type => value, ...])", func, svname);
271} 261}
272 262
273/*****************************************************************************/ 263/*****************************************************************************/
274 264
265#define NEW_CLOBJ(class,ptr) sv_setref_pv (sv_newmortal (), class, ptr)
266#define PUSH_CLOBJ(class,ptr) PUSHs (NEW_CLOBJ (class, ptr))
267#define XPUSH_CLOBJ(class,ptr) XPUSHs (NEW_CLOBJ (class, ptr))
268
269/* cl objects are either \$iv, or [$iv, ...] */
270/* they can be upgraded at runtime to the array form */
271static void *
272SvCLOBJ (const char *func, const char *svname, SV *sv, const char *pkg)
273{
274 if (SvROK (sv) && sv_derived_from (sv, pkg))
275 return (void *)SvIV (SvRV (sv));
276
277 croak ("%s: %s is not of type %s", func, svname, pkg);
278}
279
280/*****************************************************************************/
281/* callback stuff */
282
283/* default context callback, log to stderr */
284static void CL_CALLBACK
285context_default_notify (const char *msg, const void *info, size_t cb, void *data)
286{
287 fprintf (stderr, "OpenCL Context Notify: %s\n", msg);
288}
289
290typedef struct
291{
292 int free_cb;
293 void (*push)(void *data1, void *data2, void *data3);
294} eq_vtbl;
295
296typedef struct eq_item
297{
298 struct eq_item *next;
299 eq_vtbl *vtbl;
300 SV *cb;
301 void *data1, *data2, *data3;
302} eq_item;
303
304static void (*eq_signal_func)(void *signal_arg, int value);
305static void *eq_signal_arg;
306static xmutex_t eq_lock = X_MUTEX_INIT;
307static eq_item *eq_head, *eq_tail;
308
309static void
310eq_enq (eq_vtbl *vtbl, SV *cb, void *data1, void *data2, void *data3)
311{
312 eq_item *item = malloc (sizeof (eq_item));
313
314 item->next = 0;
315 item->vtbl = vtbl;
316 item->cb = cb;
317 item->data1 = data1;
318 item->data2 = data2;
319 item->data3 = data3;
320
321 X_LOCK (eq_lock);
322
323 *(eq_head ? &eq_tail->next : &eq_head) = item;
324 eq_tail = item;
325
326 X_UNLOCK (eq_lock);
327
328 eq_signal_func (eq_signal_arg, 0);
329}
330
331static eq_item *
332eq_dec (void)
333{
334 eq_item *res;
335
336 X_LOCK (eq_lock);
337
338 res = eq_head;
339 if (res)
340 eq_head = res->next;
341
342 X_UNLOCK (eq_lock);
343
344 return res;
345}
346
347static void
348eq_poll (void)
349{
350 eq_item *item;
351
352 while ((item = eq_dec ()))
353 {
354 ENTER;
355 SAVETMPS;
356
357 dSP;
358 PUSHMARK (SP);
359 EXTEND (SP, 2);
360
361 if (item->vtbl->free_cb)
362 sv_2mortal (item->cb);
363
364 PUTBACK;
365 item->vtbl->push (item->data1, item->data2, item->data3);
366
367 SV *cb = item->cb;
368 free (item);
369
370 call_sv (cb, G_DISCARD | G_VOID);
371
372 FREETMPS;
373 LEAVE;
374 }
375}
376
377static void
378eq_poll_interrupt (pTHX_ void *c_arg, int value)
379{
380 eq_poll ();
381}
382
383/*****************************************************************************/
384/* context notify */
385
386static void
387eq_context_push (void *data1, void *data2, void *data3)
388{
389 dSP;
390 PUSHs (sv_2mortal (newSVpv (data1, 0)));
391 PUSHs (sv_2mortal (newSVpvn (data2, (STRLEN)data3)));
392 PUTBACK;
393
394 free (data1);
395 free (data2);
396}
397
398static eq_vtbl eq_context_vtbl = { 0, eq_context_push };
399
400static void CL_CALLBACK
401eq_context_notify (const char *msg, const void *pvt, size_t cb, void *user_data)
402{
403 void *pvt_copy = malloc (cb);
404 memcpy (pvt_copy, pvt, cb);
405 eq_enq (&eq_context_vtbl, user_data, strdup (msg), pvt_copy, (void *)cb);
406}
407
408#define CONTEXT_NOTIFY_CALLBACK \
409 void (CL_CALLBACK *pfn_notify)(const char *, const void *, size_t, void *) = context_default_notify; \
410 void *user_data = 0; \
411 \
412 if (SvOK (notify)) \
413 { \
414 pfn_notify = eq_context_notify; \
415 user_data = s_get_cv (notify); \
416 }
417
418static SV *
419new_clobj_context (cl_context ctx, void *user_data)
420{
421 SV *sv = NEW_CLOBJ ("OpenCL::Context", ctx);
422
423 if (user_data)
424 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0);
425
426 return sv;
427}
428
429#define XPUSH_CLOBJ_CONTEXT XPUSHs (new_clobj_context (ctx, user_data));
430
431/*****************************************************************************/
432/* build/compile/link notify */
433
434static void
435eq_program_push (void *data1, void *data2, void *data3)
436{
437 dSP;
438 PUSH_CLOBJ ("OpenCL::Program", data1);
439 PUTBACK;
440}
441
442static eq_vtbl eq_program_vtbl = { 1, eq_program_push };
443
444static void CL_CALLBACK
445eq_program_notify (cl_program program, void *user_data)
446{
447 eq_enq (&eq_program_vtbl, user_data, (void *)program, 0, 0);
448}
449
450struct build_args
451{
452 cl_program program;
453 char *options;
454 void *user_data;
455 cl_uint num_devices;
456};
457
458X_THREAD_PROC (build_program_thread)
459{
460 struct build_args *arg = thr_arg;
461
462 clBuildProgram (arg->program, arg->num_devices, arg->num_devices ? (void *)(arg + 1) : 0, arg->options, 0, 0);
463
464 if (arg->user_data)
465 eq_program_notify (arg->program, arg->user_data);
466 else
467 clReleaseProgram (arg->program);
468
469 free (arg->options);
470 free (arg);
471}
472
473static void
474build_program_async (cl_program program, cl_uint num_devices, const cl_device_id *device_list, const char *options, void *user_data)
475{
476 struct build_args *arg = malloc (sizeof (struct build_args) + sizeof (*device_list) * num_devices);
477
478 arg->program = program;
479 arg->options = strdup (options);
480 arg->user_data = user_data;
481 arg->num_devices = num_devices;
482 memcpy (arg + 1, device_list, sizeof (*device_list) * num_devices);
483
484 xthread_t id;
485 thread_create (&id, build_program_thread, arg);
486}
487
488/*****************************************************************************/
489/* event objects */
490
491static void
492eq_event_push (void *data1, void *data2, void *data3)
493{
494 dSP;
495 PUSH_CLOBJ ("OpenCL::Event", data1);
496 PUSHs (sv_2mortal (newSViv ((IV)data2)));
497 PUTBACK;
498}
499
500static eq_vtbl eq_event_vtbl = { 1, eq_event_push };
501
502static void CL_CALLBACK
503eq_event_notify (cl_event event, cl_int event_command_exec_status, void *user_data)
504{
505 clRetainEvent (event);
506 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0);
507}
508
509/*****************************************************************************/
510
275static size_t 511static size_t
276img_row_pitch (cl_mem img) 512img_row_pitch (cl_mem img)
277{ 513{
278 size_t res; 514 size_t res;
279 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0); 515 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0);
293 529
294 do 530 do
295 { 531 {
296 --count; 532 --count;
297 if (SvOK (items [count])) 533 if (SvOK (items [count]))
298 list [i++] = SvPTROBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event"); 534 list [i++] = SvCLOBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event");
299 } 535 }
300 while (count); 536 while (count);
301 537
302 *rcount = i; 538 *rcount = i;
303 539
321} 557}
322 558
323MODULE = OpenCL PACKAGE = OpenCL 559MODULE = OpenCL PACKAGE = OpenCL
324 560
325PROTOTYPES: ENABLE 561PROTOTYPES: ENABLE
562
563void
564poll ()
565 CODE:
566 eq_poll ();
567
568void
569_eq_initialise (IV func, IV arg)
570 CODE:
571 eq_signal_func = (void (*)(void *, int))func;
572 eq_signal_arg = (void*)arg;
326 573
327BOOT: 574BOOT:
328{ 575{
329 HV *stash = gv_stashpv ("OpenCL", 1); 576 HV *stash = gv_stashpv ("OpenCL", 1);
330 static const ivstr *civ, const_iv[] = { 577 static const ivstr *civ, const_iv[] = {
339 { sizeof (cl_half ), "SIZEOF_HALF" }, 586 { sizeof (cl_half ), "SIZEOF_HALF" },
340 { sizeof (cl_float ), "SIZEOF_FLOAT" }, 587 { sizeof (cl_float ), "SIZEOF_FLOAT" },
341 { sizeof (cl_double), "SIZEOF_DOUBLE" }, 588 { sizeof (cl_double), "SIZEOF_DOUBLE" },
342#include "constiv.h" 589#include "constiv.h"
343 }; 590 };
591
344 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--) 592 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--)
345 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv)); 593 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv));
594
595 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt);
346} 596}
347 597
348cl_int 598cl_int
349errno () 599errno ()
350 CODE: 600 CODE:
351 RETVAL = res; 601 RETVAL = res;
352 OUTPUT: 602 OUTPUT:
353 RETVAL 603 RETVAL
354 604
355const char * 605const char *
356err2str (cl_int err) 606err2str (cl_int err = res)
357 607
358const char * 608const char *
359enum2str (cl_uint value) 609enum2str (cl_uint value)
360 610
361void 611void
369 list = tmpbuf (sizeof (*list) * count); 619 list = tmpbuf (sizeof (*list) * count);
370 NEED_SUCCESS (GetPlatformIDs, (count, list, 0)); 620 NEED_SUCCESS (GetPlatformIDs, (count, list, 0));
371 621
372 EXTEND (SP, count); 622 EXTEND (SP, count);
373 for (i = 0; i < count; ++i) 623 for (i = 0; i < count; ++i)
374 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Platform", list [i])); 624 PUSH_CLOBJ ("OpenCL::Platform", list [i]);
375 625
376void 626void
377context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, FUTURE notify = 0) 627context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
378 PPCODE: 628 PPCODE:
629 CONTEXT_NOTIFY_CALLBACK;
379 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, 0, 0, &res)); 630 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, 0, 0, &res));
380 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 631 XPUSH_CLOBJ_CONTEXT;
381 632
382#if 0
383
384void 633void
385context (cl_context_properties *properties = 0, FUTURE devices, FUTURE notify = 0) 634context (FUTURE properties, FUTURE devices, FUTURE notify)
386 PPCODE: 635 PPCODE:
387 /* der Gipfel der Kunst */ 636 /* der Gipfel der Kunst */
388
389#endif
390 637
391void 638void
392wait_for_events (...) 639wait_for_events (...)
393 CODE: 640 CODE:
394 EVENT_LIST (0, items); 641 EVENT_LIST (0, items);
400 647
401void 648void
402info (OpenCL::Platform self, cl_platform_info name) 649info (OpenCL::Platform self, cl_platform_info name)
403 PPCODE: 650 PPCODE:
404 INFO (Platform) 651 INFO (Platform)
652
653void
654unload_compiler (OpenCL::Platform self)
655 CODE:
656#if CL_VERSION_1_2
657 clUnloadPlatformCompiler (self);
658#endif
405 659
406#BEGIN:platform 660#BEGIN:platform
407 661
408void 662void
409profile (OpenCL::Platform self) 663profile (OpenCL::Platform self)
438 EXTEND (SP, count); 692 EXTEND (SP, count);
439 for (i = 0; i < count; ++i) 693 for (i = 0; i < count; ++i)
440 PUSHs (sv_setref_pv (sv_newmortal (), "OpenCL::Device", list [i])); 694 PUSHs (sv_setref_pv (sv_newmortal (), "OpenCL::Device", list [i]));
441 695
442void 696void
443context (OpenCL::Platform self, cl_context_properties *properties = 0, SV *devices, FUTURE notify = 0) 697context (OpenCL::Platform self, cl_context_properties *properties, SV *devices, SV *notify = &PL_sv_undef)
444 PPCODE: 698 PPCODE:
445 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV) 699 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV)
446 croak ("OpenCL::Platform::context argument 'device' must be an arrayref with device objects, in call"); 700 croak ("OpenCL::Platform::context argument 'device' must be an arrayref with device objects, in call");
447 701
448 AV *av = (AV *)SvRV (devices); 702 AV *av = (AV *)SvRV (devices);
449 cl_uint num_devices = av_len (av) + 1; 703 cl_uint num_devices = av_len (av) + 1;
450 cl_device_id *device_list = tmpbuf (sizeof (cl_device_id) * num_devices); 704 cl_device_id *device_list = tmpbuf (sizeof (cl_device_id) * num_devices);
705
451 int i; 706 int i;
452
453 for (i = num_devices; i--; ) 707 for (i = num_devices; i--; )
454 device_list [i] = SvPTROBJ ("clCreateContext", "devices", *av_fetch (av, i, 0), "OpenCL::Device"); 708 device_list [i] = SvCLOBJ ("clCreateContext", "devices", *av_fetch (av, i, 0), "OpenCL::Device");
455 709
710 CONTEXT_NOTIFY_CALLBACK;
456 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (properties, num_devices, device_list, 0, 0, &res)); 711 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (properties, num_devices, device_list, pfn_notify, user_data, &res));
457 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 712 XPUSH_CLOBJ_CONTEXT;
458 713
459void 714void
460context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, FUTURE notify = 0) 715context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
461 PPCODE: 716 PPCODE:
462 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self }; 717 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
463 cl_context_properties *props = SvCONTEXTPROPERTIES ("OpenCL::Platform::context_from_type", "properties", properties, extra, 2); 718 cl_context_properties *props = SvCONTEXTPROPERTIES ("OpenCL::Platform::context_from_type", "properties", properties, extra, 2);
719
720 CONTEXT_NOTIFY_CALLBACK;
464 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, 0, 0, &res)); 721 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, 0, 0, &res));
465 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 722 XPUSH_CLOBJ_CONTEXT;
466 723
467MODULE = OpenCL PACKAGE = OpenCL::Device 724MODULE = OpenCL PACKAGE = OpenCL::Device
468 725
469void 726void
470info (OpenCL::Device self, cl_device_info name) 727info (OpenCL::Device self, cl_device_info name)
643 cl_platform_id value [1]; 900 cl_platform_id value [1];
644 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0)); 901 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0));
645 EXTEND (SP, 1); 902 EXTEND (SP, 1);
646 const int i = 0; 903 const int i = 0;
647 { 904 {
648 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Platform", value [i])); 905 PUSH_CLOBJ ("OpenCL::Platform", value [i]);
649 } 906 }
650 907
651void 908void
652name (OpenCL::Device self) 909name (OpenCL::Device self)
653 ALIAS: 910 ALIAS:
672 cl_device_id value [1]; 929 cl_device_id value [1];
673 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0)); 930 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0));
674 EXTEND (SP, 1); 931 EXTEND (SP, 1);
675 const int i = 0; 932 const int i = 0;
676 { 933 {
677 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 934 PUSH_CLOBJ ("OpenCL::Device", value [i]);
678 } 935 }
679 936
680void 937void
681partition_types_ext (OpenCL::Device self) 938partition_types_ext (OpenCL::Device self)
682 ALIAS: 939 ALIAS:
709 966
710void 967void
711queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0) 968queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0)
712 PPCODE: 969 PPCODE:
713 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res)); 970 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res));
714 XPUSH_NEW_OBJ ("OpenCL::Queue", queue); 971 XPUSH_CLOBJ ("OpenCL::Queue", queue);
715 972
716void 973void
717user_event (OpenCL::Context self) 974user_event (OpenCL::Context self)
718 PPCODE: 975 PPCODE:
719 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res)); 976 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res));
720 XPUSH_NEW_OBJ ("OpenCL::UserEvent", ev); 977 XPUSH_CLOBJ ("OpenCL::UserEvent", ev);
721 978
722void 979void
723buffer (OpenCL::Context self, cl_mem_flags flags, size_t len) 980buffer (OpenCL::Context self, cl_mem_flags flags, size_t len)
724 PPCODE: 981 PPCODE:
725 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) 982 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))
726 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?"); 983 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?");
727 984
728 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res)); 985 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res));
729 XPUSH_NEW_OBJ ("OpenCL::BufferObj", mem); 986 XPUSH_CLOBJ ("OpenCL::BufferObj", mem);
730 987
731void 988void
732buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data) 989buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data)
733 PPCODE: 990 PPCODE:
734 STRLEN len; 991 STRLEN len;
735 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0; 992 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
736 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) 993 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)))
737 croak ("OpenCL::Context::buffer_sv: you have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?"); 994 croak ("OpenCL::Context::buffer_sv: you have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?");
738 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res)); 995 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res));
739 XPUSH_NEW_OBJ ("OpenCL::BufferObj", mem); 996 XPUSH_CLOBJ ("OpenCL::BufferObj", mem);
740 997
741#if CL_VERSION_1_2 998#if CL_VERSION_1_2
742 999
743void 1000void
744image (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) 1001image (OpenCL::Context self, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, cl_mem_object_type type, size_t width, size_t height, size_t depth = 0, size_t array_size = 0, size_t row_pitch = 0, size_t slice_pitch = 0, cl_uint num_mip_level = 0, cl_uint num_samples = 0, SV *data = &PL_sv_undef)
745 PPCODE: 1002 PPCODE:
746 STRLEN len; 1003 STRLEN len;
747 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0; 1004 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
748 const cl_image_format format = { channel_order, channel_type }; 1005 const cl_image_format format = { channel_order, channel_type };
749 const cl_image_desc desc = { 1006 const cl_image_desc desc = {
750 type, 1007 type,
751 width, height, depth, 1008 width, height, depth,
752 array_size, row_pitch, slice_pitch, 1009 array_size, row_pitch, slice_pitch,
753 num_mip_level, num_samples, 1010 num_mip_level, num_samples,
754 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvPTROBJ ("OpenCL::Context::Image", "data", data, "OpenCL::Buffer") : 0 1011 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ ("OpenCL::Context::Image", "data", data, "OpenCL::Buffer") : 0
755 }; 1012 };
756 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res)); 1013 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
757 char *klass = "OpenCL::Image"; 1014 char *klass = "OpenCL::Image";
758 switch (type) 1015 switch (type)
759 { 1016 {
762 case CL_MEM_OBJECT_IMAGE1D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1019 case CL_MEM_OBJECT_IMAGE1D_ARRAY: klass = "OpenCL::Image2DArray"; break;
763 case CL_MEM_OBJECT_IMAGE2D: klass = "OpenCL::Image2D"; break; 1020 case CL_MEM_OBJECT_IMAGE2D: klass = "OpenCL::Image2D"; break;
764 case CL_MEM_OBJECT_IMAGE2D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1021 case CL_MEM_OBJECT_IMAGE2D_ARRAY: klass = "OpenCL::Image2DArray"; break;
765 case CL_MEM_OBJECT_IMAGE3D: klass = "OpenCL::Image3D"; break; 1022 case CL_MEM_OBJECT_IMAGE3D: klass = "OpenCL::Image3D"; break;
766 } 1023 }
767 XPUSH_NEW_OBJ (klass, mem); 1024 XPUSH_CLOBJ (klass, mem);
768 1025
769#endif 1026#endif
770 1027
771void 1028void
772image2d (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) 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)
778 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res)); 1035 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res));
779#else 1036#else
780 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, row_pitch, 0, 0, 0, 0 }; 1037 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, row_pitch, 0, 0, 0, 0 };
781 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res)); 1038 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
782#endif 1039#endif
783 XPUSH_NEW_OBJ ("OpenCL::Image2D", mem); 1040 XPUSH_CLOBJ ("OpenCL::Image2D", mem);
784 1041
785void 1042void
786image3d (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) 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)
787 PPCODE: 1044 PPCODE:
788 STRLEN len; 1045 STRLEN len;
792 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res)); 1049 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res));
793#else 1050#else
794 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, row_pitch, slice_pitch, 0, 0, 0 }; 1051 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, row_pitch, slice_pitch, 0, 0, 0 };
795 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res)); 1052 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
796#endif 1053#endif
797 XPUSH_NEW_OBJ ("OpenCL::Image3D", mem); 1054 XPUSH_CLOBJ ("OpenCL::Image3D", mem);
798 1055
799#if cl_apple_gl_sharing || cl_khr_gl_sharing 1056#if cl_apple_gl_sharing || cl_khr_gl_sharing
800 1057
801void 1058void
802gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj) 1059gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj)
803 PPCODE: 1060 PPCODE:
804 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res)); 1061 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res));
805 XPUSH_NEW_OBJ ("OpenCL::BufferObj", mem); 1062 XPUSH_CLOBJ ("OpenCL::BufferObj", mem);
806 1063
807void 1064void
808gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer) 1065gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer)
809 PPCODE: 1066 PPCODE:
810 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res)); 1067 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res));
811 XPUSH_NEW_OBJ ("OpenCL::Image2D", mem); 1068 XPUSH_CLOBJ ("OpenCL::Image2D", mem);
812 1069
813#if CL_VERSION_1_2 1070#if CL_VERSION_1_2
814 1071
815void 1072void
816gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1073gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
827 case CL_GL_OBJECT_TEXTURE1D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1084 case CL_GL_OBJECT_TEXTURE1D_ARRAY: klass = "OpenCL::Image2DArray"; break;
828 case CL_GL_OBJECT_TEXTURE2D: klass = "OpenCL::Image2D"; break; 1085 case CL_GL_OBJECT_TEXTURE2D: klass = "OpenCL::Image2D"; break;
829 case CL_GL_OBJECT_TEXTURE2D_ARRAY: klass = "OpenCL::Image2DArray"; break; 1086 case CL_GL_OBJECT_TEXTURE2D_ARRAY: klass = "OpenCL::Image2DArray"; break;
830 case CL_GL_OBJECT_TEXTURE3D: klass = "OpenCL::Image3D"; break; 1087 case CL_GL_OBJECT_TEXTURE3D: klass = "OpenCL::Image3D"; break;
831 } 1088 }
832 XPUSH_NEW_OBJ (klass, mem); 1089 XPUSH_CLOBJ (klass, mem);
833 1090
834#endif 1091#endif
835 1092
836void 1093void
837gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1094gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
839#if PREFER_1_1 1096#if PREFER_1_1
840 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res)); 1097 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res));
841#else 1098#else
842 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res)); 1099 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
843#endif 1100#endif
844 XPUSH_NEW_OBJ ("OpenCL::Image2D", mem); 1101 XPUSH_CLOBJ ("OpenCL::Image2D", mem);
845 1102
846void 1103void
847gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1104gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
848 PPCODE: 1105 PPCODE:
849#if PREFER_1_1 1106#if PREFER_1_1
850 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res)); 1107 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res));
851#else 1108#else
852 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res)); 1109 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
853#endif 1110#endif
854 XPUSH_NEW_OBJ ("OpenCL::Image3D", mem); 1111 XPUSH_CLOBJ ("OpenCL::Image3D", mem);
855 1112
856#endif 1113#endif
857 1114
858void 1115void
859supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type) 1116supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type)
879 1136
880void 1137void
881sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode) 1138sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode)
882 PPCODE: 1139 PPCODE:
883 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res)); 1140 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res));
884 XPUSH_NEW_OBJ ("OpenCL::Sampler", sampler); 1141 XPUSH_CLOBJ ("OpenCL::Sampler", sampler);
885 1142
886void 1143void
887program_with_source (OpenCL::Context self, SV *program) 1144program_with_source (OpenCL::Context self, SV *program)
888 PPCODE: 1145 PPCODE:
889 STRLEN len; 1146 STRLEN len;
890 size_t len2; 1147 size_t len2;
891 const char *ptr = SvPVbyte (program, len); 1148 const char *ptr = SvPVbyte (program, len);
892 1149
893 len2 = len; 1150 len2 = len;
894 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res)); 1151 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res));
895 XPUSH_NEW_OBJ ("OpenCL::Program", prog); 1152 XPUSH_CLOBJ ("OpenCL::Program", prog);
896 1153
897#BEGIN:context 1154#BEGIN:context
898 1155
899void 1156void
900reference_count (OpenCL::Context self) 1157reference_count (OpenCL::Context self)
917 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0)); 1174 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0));
918 int i, n = size / sizeof (*value); 1175 int i, n = size / sizeof (*value);
919 EXTEND (SP, n); 1176 EXTEND (SP, n);
920 for (i = 0; i < n; ++i) 1177 for (i = 0; i < n; ++i)
921 { 1178 {
922 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 1179 PUSH_CLOBJ ("OpenCL::Device", value [i]);
923 } 1180 }
924 1181
925void 1182void
926properties (OpenCL::Context self) 1183properties (OpenCL::Context self)
927 PPCODE: 1184 PPCODE:
942DESTROY (OpenCL::Queue self) 1199DESTROY (OpenCL::Queue self)
943 CODE: 1200 CODE:
944 clReleaseCommandQueue (self); 1201 clReleaseCommandQueue (self);
945 1202
946void 1203void
947enqueue_read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...) 1204read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...)
1205 ALIAS:
1206 enqueue_read_buffer = 0
948 PPCODE: 1207 PPCODE:
949 cl_event ev = 0; 1208 cl_event ev = 0;
950 EVENT_LIST (6, items - 6); 1209 EVENT_LIST (6, items - 6);
951 1210
952 SvUPGRADE (data, SVt_PV); 1211 SvUPGRADE (data, SVt_PV);
954 SvPOK_only (data); 1213 SvPOK_only (data);
955 SvCUR_set (data, len); 1214 SvCUR_set (data, len);
956 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1215 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
957 1216
958 if (ev) 1217 if (ev)
959 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1218 XPUSH_CLOBJ ("OpenCL::Event", ev);
960 1219
961void 1220void
962enqueue_write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...) 1221write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...)
1222 ALIAS:
1223 enqueue_write_buffer = 0
963 PPCODE: 1224 PPCODE:
964 cl_event ev = 0; 1225 cl_event ev = 0;
965 STRLEN len; 1226 STRLEN len;
966 char *ptr = SvPVbyte (data, len); 1227 char *ptr = SvPVbyte (data, len);
967 EVENT_LIST (5, items - 5); 1228 EVENT_LIST (5, items - 5);
968 1229
969 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1230 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
970 1231
971 if (ev) 1232 if (ev)
972 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1233 XPUSH_CLOBJ ("OpenCL::Event", ev);
973 1234
1235#if CL_VERSION_1_2
1236
974void 1237void
1238fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...)
1239 ALIAS:
1240 enqueue_fill_buffer = 0
1241 PPCODE:
1242 cl_event ev = 0;
1243 STRLEN len;
1244 char *ptr = SvPVbyte (data, len);
1245 EVENT_LIST (5, items - 5);
1246
1247 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1248
1249 if (ev)
1250 XPUSH_CLOBJ ("OpenCL::Event", ev);
1251
1252void
1253fill_image (OpenCL::Queue self, OpenCL::Image img, NV r, NV g, NV b, NV a, size_t x, size_t y, size_t z, size_t width, size_t height, size_t depth, ...)
1254 ALIAS:
1255 enqueue_fill_image = 0
1256 PPCODE:
1257 cl_event ev = 0;
1258 STRLEN len;
1259 const size_t origin [3] = { x, y, z };
1260 const size_t region [3] = { width, height, depth };
1261 EVENT_LIST (12, items - 12);
1262
1263 const cl_float c_f [4] = { r, g, b, a };
1264 const cl_uint c_u [4] = { r, g, b, a };
1265 const cl_int c_s [4] = { r, g, b, a };
1266 const void *c_fus [3] = { &c_f, &c_u, &c_s };
1267 static const char fus [] = { 0, 0, 0, 0, 0, 0, 0, 2, 2, 2, 1, 1, 1, 0, 0 };
1268 cl_image_format format;
1269 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
1270 assert (sizeof (fus) == CL_FLOAT + 1 - CL_SNORM_INT8);
1271 if (format.image_channel_data_type < CL_SNORM_INT8 || CL_FLOAT < format.image_channel_data_type)
1272 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported.");
1273
1274 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type - CL_SNORM_INT8]],
1275 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1276
1277 if (ev)
1278 XPUSH_CLOBJ ("OpenCL::Event", ev);
1279
1280#endif
1281
1282void
975enqueue_copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...) 1283copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...)
1284 ALIAS:
1285 enqueue_copy_buffer = 0
976 PPCODE: 1286 PPCODE:
977 cl_event ev = 0; 1287 cl_event ev = 0;
978 EVENT_LIST (6, items - 6); 1288 EVENT_LIST (6, items - 6);
979 1289
980 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1290 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
981 1291
982 if (ev) 1292 if (ev)
983 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1293 XPUSH_CLOBJ ("OpenCL::Event", ev);
984 1294
985void 1295void
986enqueue_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, ...) 1296read_buffer_rect (OpenCL::Queue self, OpenCL::Memory buf, cl_bool blocking, size_t buf_x, size_t buf_y, size_t buf_z, size_t host_x, size_t host_y, size_t host_z, size_t width, size_t height, size_t depth, size_t buf_row_pitch, size_t buf_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, SV *data, ...)
1297 ALIAS:
1298 enqueue_read_buffer_rect = 0
987 PPCODE: 1299 PPCODE:
988 cl_event ev = 0; 1300 cl_event ev = 0;
989 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1301 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
990 const size_t host_origin[3] = { host_x, host_y, host_z }; 1302 const size_t host_origin[3] = { host_x, host_y, host_z };
991 const size_t region[3] = { width, height, depth }; 1303 const size_t region[3] = { width, height, depth };
1010 SvPOK_only (data); 1322 SvPOK_only (data);
1011 SvCUR_set (data, len); 1323 SvCUR_set (data, len);
1012 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)); 1324 NEED_SUCCESS (EnqueueReadBufferRect, (self, buf, blocking, buf_origin, host_origin, region, buf_row_pitch, buf_slice_pitch, host_row_pitch, host_slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1013 1325
1014 if (ev) 1326 if (ev)
1015 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1327 XPUSH_CLOBJ ("OpenCL::Event", ev);
1016 1328
1017void 1329void
1018enqueue_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, ...) 1330write_buffer_rect (OpenCL::Queue self, OpenCL::Memory buf, cl_bool blocking, size_t buf_x, size_t buf_y, size_t buf_z, size_t host_x, size_t host_y, size_t host_z, size_t width, size_t height, size_t depth, size_t buf_row_pitch, size_t buf_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, SV *data, ...)
1331 ALIAS:
1332 enqueue_write_buffer_rect = 0
1019 PPCODE: 1333 PPCODE:
1020 cl_event ev = 0; 1334 cl_event ev = 0;
1021 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1335 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
1022 const size_t host_origin[3] = { host_x, host_y, host_z }; 1336 const size_t host_origin[3] = { host_x, host_y, host_z };
1023 const size_t region[3] = { width, height, depth }; 1337 const size_t region[3] = { width, height, depth };
1043 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1357 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
1044 1358
1045 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)); 1359 NEED_SUCCESS (EnqueueWriteBufferRect, (self, buf, blocking, buf_origin, host_origin, region, buf_row_pitch, buf_slice_pitch, host_row_pitch, host_slice_pitch, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1046 1360
1047 if (ev) 1361 if (ev)
1048 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1362 XPUSH_CLOBJ ("OpenCL::Event", ev);
1049 1363
1050void 1364void
1051enqueue_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, ...) 1365copy_buffer_rect (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_x, size_t src_y, size_t src_z, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, size_t src_row_pitch, size_t src_slice_pitch, size_t dst_row_pitch, size_t dst_slice_pitch, ...)
1366 ALIAS:
1367 enqueue_copy_buffer_rect = 0
1052 PPCODE: 1368 PPCODE:
1053 cl_event ev = 0; 1369 cl_event ev = 0;
1054 const size_t src_origin[3] = { src_x, src_y, src_z }; 1370 const size_t src_origin[3] = { src_x, src_y, src_z };
1055 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1371 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1056 const size_t region[3] = { width, height, depth }; 1372 const size_t region[3] = { width, height, depth };
1057 EVENT_LIST (16, items - 16); 1373 EVENT_LIST (16, items - 16);
1058 1374
1059 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)); 1375 NEED_SUCCESS (EnqueueCopyBufferRect, (self, src, dst, src_origin, dst_origin, region, src_row_pitch, src_slice_pitch, dst_row_pitch, dst_slice_pitch, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1060 1376
1061 if (ev) 1377 if (ev)
1062 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1378 XPUSH_CLOBJ ("OpenCL::Event", ev);
1063 1379
1064void 1380void
1065enqueue_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, ...) 1381read_image (OpenCL::Queue self, OpenCL::Image src, cl_bool blocking, size_t src_x, size_t src_y, size_t src_z, size_t width, size_t height, size_t depth, size_t row_pitch, size_t slice_pitch, SV *data, ...)
1382 ALIAS:
1383 enqueue_read_image = 0
1066 PPCODE: 1384 PPCODE:
1067 cl_event ev = 0; 1385 cl_event ev = 0;
1068 const size_t src_origin[3] = { src_x, src_y, src_z }; 1386 const size_t src_origin[3] = { src_x, src_y, src_z };
1069 const size_t region[3] = { width, height, depth }; 1387 const size_t region[3] = { width, height, depth };
1070 EVENT_LIST (12, items - 12); 1388 EVENT_LIST (12, items - 12);
1082 SvPOK_only (data); 1400 SvPOK_only (data);
1083 SvCUR_set (data, len); 1401 SvCUR_set (data, len);
1084 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)); 1402 NEED_SUCCESS (EnqueueReadImage, (self, src, blocking, src_origin, region, row_pitch, slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1085 1403
1086 if (ev) 1404 if (ev)
1087 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1405 XPUSH_CLOBJ ("OpenCL::Event", ev);
1088 1406
1089void 1407void
1090enqueue_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, ...) 1408write_image (OpenCL::Queue self, OpenCL::Image dst, cl_bool blocking, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, size_t row_pitch, size_t slice_pitch, SV *data, ...)
1409 ALIAS:
1410 enqueue_write_image = 0
1091 PPCODE: 1411 PPCODE:
1092 cl_event ev = 0; 1412 cl_event ev = 0;
1093 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1413 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1094 const size_t region[3] = { width, height, depth }; 1414 const size_t region[3] = { width, height, depth };
1095 STRLEN len; 1415 STRLEN len;
1108 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1428 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
1109 1429
1110 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)); 1430 NEED_SUCCESS (EnqueueWriteImage, (self, dst, blocking, dst_origin, region, row_pitch, slice_pitch, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1111 1431
1112 if (ev) 1432 if (ev)
1113 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1433 XPUSH_CLOBJ ("OpenCL::Event", ev);
1114 1434
1115void 1435void
1116enqueue_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, ...) 1436copy_image (OpenCL::Queue self, OpenCL::Image src, OpenCL::Image dst, size_t src_x, size_t src_y, size_t src_z, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, ...)
1437 ALIAS:
1438 enqueue_copy_image = 0
1117 PPCODE: 1439 PPCODE:
1118 cl_event ev = 0; 1440 cl_event ev = 0;
1119 const size_t src_origin[3] = { src_x, src_y, src_z }; 1441 const size_t src_origin[3] = { src_x, src_y, src_z };
1120 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1442 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1121 const size_t region[3] = { width, height, depth }; 1443 const size_t region[3] = { width, height, depth };
1122 EVENT_LIST (12, items - 12); 1444 EVENT_LIST (12, items - 12);
1123 1445
1124 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1446 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1125 1447
1126 if (ev) 1448 if (ev)
1127 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1449 XPUSH_CLOBJ ("OpenCL::Event", ev);
1128 1450
1129void 1451void
1130enqueue_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, ...) 1452copy_image_to_buffer (OpenCL::Queue self, OpenCL::Image src, OpenCL::Buffer dst, size_t src_x, size_t src_y, size_t src_z, size_t width, size_t height, size_t depth, size_t dst_offset, ...)
1453 ALIAS:
1454 enqueue_copy_image_to_buffer = 0
1131 PPCODE: 1455 PPCODE:
1132 cl_event ev = 0; 1456 cl_event ev = 0;
1133 const size_t src_origin[3] = { src_x, src_y, src_z }; 1457 const size_t src_origin[3] = { src_x, src_y, src_z };
1134 const size_t region[3] = { width, height, depth }; 1458 const size_t region[3] = { width, height, depth };
1135 EVENT_LIST (10, items - 10); 1459 EVENT_LIST (10, items - 10);
1136 1460
1137 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1461 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1138 1462
1139 if (ev) 1463 if (ev)
1140 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1464 XPUSH_CLOBJ ("OpenCL::Event", ev);
1141 1465
1142void 1466void
1143enqueue_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, ...) 1467copy_buffer_to_image (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Image dst, size_t src_offset, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, ...)
1468 ALIAS:
1469 enqueue_copy_buffer_to_image = 0
1144 PPCODE: 1470 PPCODE:
1145 cl_event ev = 0; 1471 cl_event ev = 0;
1146 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1472 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1147 const size_t region[3] = { width, height, depth }; 1473 const size_t region[3] = { width, height, depth };
1148 EVENT_LIST (10, items - 10); 1474 EVENT_LIST (10, items - 10);
1149 1475
1150 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1476 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1151 1477
1152 if (ev) 1478 if (ev)
1153 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1479 XPUSH_CLOBJ ("OpenCL::Event", ev);
1154 1480
1155void 1481void
1156enqueue_task (OpenCL::Queue self, OpenCL::Kernel kernel, ...) 1482task (OpenCL::Queue self, OpenCL::Kernel kernel, ...)
1483 ALIAS:
1484 enqueue_task = 0
1157 PPCODE: 1485 PPCODE:
1158 cl_event ev = 0; 1486 cl_event ev = 0;
1159 EVENT_LIST (2, items - 2); 1487 EVENT_LIST (2, items - 2);
1160 1488
1161 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1489 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1162 1490
1163 if (ev) 1491 if (ev)
1164 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1492 XPUSH_CLOBJ ("OpenCL::Event", ev);
1165 1493
1166void 1494void
1167enqueue_nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...) 1495nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...)
1496 ALIAS:
1497 enqueue_nd_range_kernel = 0
1168 PPCODE: 1498 PPCODE:
1169 cl_event ev = 0; 1499 cl_event ev = 0;
1170 size_t *gwo = 0, *gws, *lws = 0; 1500 size_t *gwo = 0, *gws, *lws = 0;
1171 int gws_len; 1501 int gws_len;
1172 size_t *lists; 1502 size_t *lists;
1211 } 1541 }
1212 1542
1213 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1543 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1214 1544
1215 if (ev) 1545 if (ev)
1216 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1546 XPUSH_CLOBJ ("OpenCL::Event", ev);
1217 1547
1218#if cl_apple_gl_sharing || cl_khr_gl_sharing 1548#if cl_apple_gl_sharing || cl_khr_gl_sharing
1219 1549
1220void 1550void
1221enqueue_acquire_gl_objects (OpenCL::Queue self, SV *objects, ...) 1551acquire_gl_objects (OpenCL::Queue self, SV *objects, ...)
1552 ALIAS:
1553 enqueue_acquire_gl_objects = 0
1222 ALIAS: 1554 ALIAS:
1223 enqueue_release_gl_objects = 1 1555 enqueue_release_gl_objects = 1
1224 PPCODE: 1556 PPCODE:
1225 if (!SvROK (objects) || SvTYPE (SvRV (objects)) != SVt_PVAV) 1557 if (!SvROK (objects) || SvTYPE (SvRV (objects)) != SVt_PVAV)
1226 croak ("OpenCL::Queue::enqueue_acquire/release_gl_objects argument 'objects' must be an arrayref with memory objects, in call"); 1558 croak ("OpenCL::Queue::enqueue_acquire/release_gl_objects argument 'objects' must be an arrayref with memory objects, in call");
1231 cl_uint num_objects = av_len (av) + 1; 1563 cl_uint num_objects = av_len (av) + 1;
1232 cl_mem *object_list = tmpbuf (sizeof (cl_mem) * num_objects); 1564 cl_mem *object_list = tmpbuf (sizeof (cl_mem) * num_objects);
1233 int i; 1565 int i;
1234 1566
1235 for (i = num_objects; i--; ) 1567 for (i = num_objects; i--; )
1236 object_list [i] = SvPTROBJ ("OpenCL::Queue::enqueue_acquire/release_gl_objects", "objects", *av_fetch (av, i, 0), "OpenCL::Memory"); 1568 object_list [i] = SvCLOBJ ("OpenCL::Queue::enqueue_acquire/release_gl_objects", "objects", *av_fetch (av, i, 0), "OpenCL::Memory");
1237 1569
1238 if (ix) 1570 if (ix)
1239 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1571 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1240 else 1572 else
1241 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1573 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1242 1574
1243 if (ev) 1575 if (ev)
1244 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1576 XPUSH_CLOBJ ("OpenCL::Event", ev);
1245 1577
1246#endif 1578#endif
1247 1579
1248#if !CL_VERSION_1_2 || defined CL_USE_DEPRECATED_OPENCL_1_1_APIS
1249
1250void 1580void
1251enqueue_wait_for_events (OpenCL::Queue self, ...) 1581wait_for_events (OpenCL::Queue self, ...)
1582 ALIAS:
1583 enqueue_wait_for_events = 0
1252 CODE: 1584 CODE:
1253 EVENT_LIST (1, items - 1); 1585 EVENT_LIST (1, items - 1);
1586#if PREFER_1_1
1254 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr)); 1587 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
1255 1588#else
1589 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0));
1256#endif 1590#endif
1257 1591
1258void 1592void
1259enqueue_marker (OpenCL::Queue self, ...) 1593marker (OpenCL::Queue self, ...)
1594 ALIAS:
1595 enqueue_marker = 0
1260 PPCODE: 1596 PPCODE:
1261 cl_event ev = 0; 1597 cl_event ev = 0;
1262 EVENT_LIST (1, items - 1); 1598 EVENT_LIST (1, items - 1);
1263#if PREFER_1_1 1599#if PREFER_1_1
1600 if (!event_list_count)
1601 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
1602 else
1264#if CL_VERSION_1_2 1603#if CL_VERSION_1_2
1265 if (event_list_count)
1266 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1604 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1267 else
1268 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
1269#else 1605#else
1270 if (event_list_count) 1606 {
1271 croak ("OpenCL::Queue->enqueue_marker does not support a wait list in OpenCL 1.1 - upgrade to 1.2"); 1607 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr)); // also a barrier
1272 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0)); 1608 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
1609 }
1273#endif 1610#endif
1274#else 1611#else
1275 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1612 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1276#endif 1613#endif
1277 if (ev) 1614 if (ev)
1278 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1615 XPUSH_CLOBJ ("OpenCL::Event", ev);
1279 1616
1280void 1617void
1281enqueue_barrier (OpenCL::Queue self, ...) 1618barrier (OpenCL::Queue self, ...)
1619 ALIAS:
1620 enqueue_barrier = 0
1282 PPCODE: 1621 PPCODE:
1283 cl_event ev = 0; 1622 cl_event ev = 0;
1284 EVENT_LIST (1, items - 1); 1623 EVENT_LIST (1, items - 1);
1285#if PREFER_1_1 1624#if PREFER_1_1
1625 if (!event_list_count && GIMME_V == G_VOID)
1626 NEED_SUCCESS (EnqueueBarrier, (self));
1627 else
1286#if CL_VERSION_1_2 1628#if CL_VERSION_1_2
1287 if (event_list_count || GIMME_V != G_VOID)
1288 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1629 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1289 else
1290 NEED_SUCCESS (EnqueueBarrier, (self));
1291#else 1630#else
1631 {
1292 if (event_list_count) 1632 if (event_list_count)
1293 croak ("OpenCL::Queue->enqueue_barrier does not support a wait list in OpenCL 1.1 - upgrade to 1.2"); 1633 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
1634
1294 if (GIMME_V != G_VOID) 1635 if (GIMME_V != G_VOID)
1295 croak ("OpenCL::Queue->enqueue_barrier does not return an event object in OpenCL 1.1 - upgrade to 1.2");
1296 NEED_SUCCESS (EnqueueBarrier, (self)); 1636 NEED_SUCCESS (EnqueueMarker, (self, &ev));
1637 }
1297#endif 1638#endif
1298#else 1639#else
1299 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1640 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1300#endif 1641#endif
1301 if (ev) 1642 if (ev)
1302 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1643 XPUSH_CLOBJ ("OpenCL::Event", ev);
1303 1644
1304void 1645void
1305flush (OpenCL::Queue self) 1646flush (OpenCL::Queue self)
1306 CODE: 1647 CODE:
1307 NEED_SUCCESS (Flush, (self)); 1648 NEED_SUCCESS (Flush, (self));
1325 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0)); 1666 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0));
1326 EXTEND (SP, 1); 1667 EXTEND (SP, 1);
1327 const int i = 0; 1668 const int i = 0;
1328 { 1669 {
1329 NEED_SUCCESS (RetainContext, (value [i])); 1670 NEED_SUCCESS (RetainContext, (value [i]));
1330 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 1671 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1331 } 1672 }
1332 1673
1333void 1674void
1334device (OpenCL::Queue self) 1675device (OpenCL::Queue self)
1335 PPCODE: 1676 PPCODE:
1336 cl_device_id value [1]; 1677 cl_device_id value [1];
1337 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0)); 1678 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0));
1338 EXTEND (SP, 1); 1679 EXTEND (SP, 1);
1339 const int i = 0; 1680 const int i = 0;
1340 { 1681 {
1341 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 1682 PUSH_CLOBJ ("OpenCL::Device", value [i]);
1342 } 1683 }
1343 1684
1344void 1685void
1345reference_count (OpenCL::Queue self) 1686reference_count (OpenCL::Queue self)
1346 PPCODE: 1687 PPCODE:
1433 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0)); 1774 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0));
1434 EXTEND (SP, 1); 1775 EXTEND (SP, 1);
1435 const int i = 0; 1776 const int i = 0;
1436 { 1777 {
1437 NEED_SUCCESS (RetainContext, (value [i])); 1778 NEED_SUCCESS (RetainContext, (value [i]));
1438 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 1779 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1439 } 1780 }
1440 1781
1441void 1782void
1442associated_memobject (OpenCL::Memory self) 1783associated_memobject (OpenCL::Memory self)
1443 PPCODE: 1784 PPCODE:
1445 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0)); 1786 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0));
1446 EXTEND (SP, 1); 1787 EXTEND (SP, 1);
1447 const int i = 0; 1788 const int i = 0;
1448 { 1789 {
1449 NEED_SUCCESS (RetainMemObject, (value [i])); 1790 NEED_SUCCESS (RetainMemObject, (value [i]));
1450 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Memory", value [i])); 1791 PUSH_CLOBJ ("OpenCL::Memory", value [i]);
1451 } 1792 }
1452 1793
1453#END:mem 1794#END:mem
1454 1795
1455#if cl_apple_gl_sharing || cl_khr_gl_sharing 1796#if cl_apple_gl_sharing || cl_khr_gl_sharing
1475 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!"); 1816 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!");
1476 1817
1477 cl_buffer_region crdata = { origin, size }; 1818 cl_buffer_region crdata = { origin, size };
1478 1819
1479 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res)); 1820 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res));
1480 XPUSH_NEW_OBJ ("OpenCL::Buffer", mem); 1821 XPUSH_CLOBJ ("OpenCL::Buffer", mem);
1481 1822
1482MODULE = OpenCL PACKAGE = OpenCL::Image 1823MODULE = OpenCL PACKAGE = OpenCL::Image
1483 1824
1484void 1825void
1485image_info (OpenCL::Image self, cl_image_info name) 1826image_info (OpenCL::Image self, cl_image_info name)
1486 PPCODE: 1827 PPCODE:
1487 INFO (Image) 1828 INFO (Image)
1829
1830void
1831format (OpenCL::Image self)
1832 PPCODE:
1833 cl_image_format format;
1834 NEED_SUCCESS (GetImageInfo, (self, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
1835 EXTEND (SP, 2);
1836 PUSHs (sv_2mortal (newSVuv (format.image_channel_order)));
1837 PUSHs (sv_2mortal (newSVuv (format.image_channel_data_type)));
1488 1838
1489#BEGIN:image 1839#BEGIN:image
1490 1840
1491void 1841void
1492element_size (OpenCL::Image self) 1842element_size (OpenCL::Image self)
1562 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0)); 1912 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0));
1563 EXTEND (SP, 1); 1913 EXTEND (SP, 1);
1564 const int i = 0; 1914 const int i = 0;
1565 { 1915 {
1566 NEED_SUCCESS (RetainContext, (value [i])); 1916 NEED_SUCCESS (RetainContext, (value [i]));
1567 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 1917 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1568 } 1918 }
1569 1919
1570void 1920void
1571normalized_coords (OpenCL::Sampler self) 1921normalized_coords (OpenCL::Sampler self)
1572 PPCODE: 1922 PPCODE:
1602DESTROY (OpenCL::Program self) 1952DESTROY (OpenCL::Program self)
1603 CODE: 1953 CODE:
1604 clReleaseProgram (self); 1954 clReleaseProgram (self);
1605 1955
1606void 1956void
1607build (OpenCL::Program self, OpenCL::Device device, SV *options = &PL_sv_undef) 1957build (OpenCL::Program self, SV *devices = &PL_sv_undef, SV *options = &PL_sv_undef, SV *notify = &PL_sv_undef)
1958 ALIAS:
1959 build_async = 1
1608 CODE: 1960 CODE:
1961 void (CL_CALLBACK *pfn_notify)(cl_program program, void *user_data) = 0;
1962 void *user_data = 0;
1963 cl_uint num_devices = 0;
1964 cl_device_id *device_list = 0;
1965
1966 if (SvOK (devices))
1967 {
1968 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV)
1969 croak ("clProgramBuild: devices must be undef or an array of OpenCL::Device objects.");
1970
1971 AV *av = (AV *)SvRV (devices);
1972 num_devices = av_len (av) + 1;
1973
1974 if (num_devices)
1975 {
1976 device_list = tmpbuf (sizeof (*device_list) * num_devices);
1977 int count;
1978 for (count = 0; count < num_devices; ++count)
1979 device_list [count] = SvCLOBJ ("clBuildProgram", "devices", *av_fetch (av, count, 1), "OpenCL::Device");
1980 }
1981 }
1982
1983 if (SvOK (notify))
1984 {
1985 NEED_SUCCESS (RetainProgram, (self));
1986 pfn_notify = eq_program_notify;
1987 user_data = SvREFCNT_inc (s_get_cv (notify));
1988 }
1989
1990 if (ix)
1991 build_program_async (self, num_devices, device_list, SvPVbyte_nolen (options), user_data);
1992 else
1609 NEED_SUCCESS (BuildProgram, (self, 1, &device, SvPVbyte_nolen (options), 0, 0)); 1993 NEED_SUCCESS (BuildProgram, (self, num_devices, device_list, SvPVbyte_nolen (options), pfn_notify, user_data));
1610 1994
1611void 1995void
1612build_info (OpenCL::Program self, OpenCL::Device device, cl_program_build_info name) 1996build_info (OpenCL::Program self, OpenCL::Device device, cl_program_build_info name)
1613 PPCODE: 1997 PPCODE:
1614 size_t size; 1998 size_t size;
1649 2033
1650void 2034void
1651kernel (OpenCL::Program program, SV *function) 2035kernel (OpenCL::Program program, SV *function)
1652 PPCODE: 2036 PPCODE:
1653 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res)); 2037 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res));
1654 XPUSH_NEW_OBJ ("OpenCL::Kernel", kernel); 2038 XPUSH_CLOBJ ("OpenCL::Kernel", kernel);
2039
2040void
2041kernels_in_program (OpenCL::Program program)
2042 PPCODE:
2043 cl_uint num_kernels;
2044 NEED_SUCCESS (CreateKernelsInProgram, (program, 0, 0, &num_kernels));
2045 cl_kernel *kernels = tmpbuf (sizeof (cl_kernel) * num_kernels);
2046 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0));
2047
2048 int i;
2049 EXTEND (SP, num_kernels);
2050 for (i = 0; i < num_kernels; ++i)
2051 PUSH_CLOBJ ("OpenCL::Kernel", kernels [i]);
1655 2052
1656void 2053void
1657info (OpenCL::Program self, cl_program_info name) 2054info (OpenCL::Program self, cl_program_info name)
1658 PPCODE: 2055 PPCODE:
1659 INFO (Program) 2056 INFO (Program)
1707 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0)); 2104 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0));
1708 EXTEND (SP, 1); 2105 EXTEND (SP, 1);
1709 const int i = 0; 2106 const int i = 0;
1710 { 2107 {
1711 NEED_SUCCESS (RetainContext, (value [i])); 2108 NEED_SUCCESS (RetainContext, (value [i]));
1712 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2109 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1713 } 2110 }
1714 2111
1715void 2112void
1716devices (OpenCL::Program self) 2113devices (OpenCL::Program self)
1717 PPCODE: 2114 PPCODE:
1721 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0)); 2118 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0));
1722 int i, n = size / sizeof (*value); 2119 int i, n = size / sizeof (*value);
1723 EXTEND (SP, n); 2120 EXTEND (SP, n);
1724 for (i = 0; i < n; ++i) 2121 for (i = 0; i < n; ++i)
1725 { 2122 {
1726 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 2123 PUSH_CLOBJ ("OpenCL::Device", value [i]);
1727 } 2124 }
1728 2125
1729void 2126void
1730source (OpenCL::Program self) 2127source (OpenCL::Program self)
1731 PPCODE: 2128 PPCODE:
1757DESTROY (OpenCL::Kernel self) 2154DESTROY (OpenCL::Kernel self)
1758 CODE: 2155 CODE:
1759 clReleaseKernel (self); 2156 clReleaseKernel (self);
1760 2157
1761void 2158void
2159setf (OpenCL::Kernel self, const char *format, ...)
2160 CODE:
2161 int i;
2162 for (i = 2; ; ++i)
2163 {
2164 while (*format == ' ')
2165 ++format;
2166
2167 char type = *format++;
2168
2169 if (!type)
2170 break;
2171
2172 if (i >= items)
2173 croak ("OpenCL::Kernel::setf format string too long (not enough arguments)");
2174
2175 SV *sv = ST (i);
2176
2177 union
2178 {
2179 cl_char cc; cl_uchar cC; cl_short cs; cl_ushort cS;
2180 cl_int ci; cl_uint cI; cl_long cl; cl_ulong cL;
2181 cl_half ch; cl_float cf; cl_double cd;
2182 cl_mem cm;
2183 cl_sampler ca;
2184 size_t cz;
2185 cl_event ce;
2186 } arg;
2187 size_t size;
2188
2189 switch (type)
2190 {
2191 case 'c': arg.cc = SvIV (sv); size = sizeof (arg.cc); break;
2192 case 'C': arg.cC = SvUV (sv); size = sizeof (arg.cC); break;
2193 case 's': arg.cs = SvIV (sv); size = sizeof (arg.cs); break;
2194 case 'S': arg.cS = SvUV (sv); size = sizeof (arg.cS); break;
2195 case 'i': arg.ci = SvIV (sv); size = sizeof (arg.ci); break;
2196 case 'I': arg.cI = SvUV (sv); size = sizeof (arg.cI); break;
2197 case 'l': arg.cl = SvIV (sv); size = sizeof (arg.cl); break;
2198 case 'L': arg.cL = SvUV (sv); size = sizeof (arg.cL); break;
2199
2200 case 'h': arg.ch = SvUV (sv); size = sizeof (arg.ch); break;
2201 case 'f': arg.cf = SvNV (sv); size = sizeof (arg.cf); break;
2202 case 'd': arg.cd = SvNV (sv); size = sizeof (arg.cd); break;
2203 case 'z': arg.cz = SvUV (sv); size = sizeof (arg.cz); break;
2204
2205 case 'm': arg.cm = SvCLOBJ ("OpenCL::Kernel::setf", "m", sv, "OpenCL::Memory" ); size = sizeof (arg.cm); break;
2206 case 'a': arg.ca = SvCLOBJ ("OpenCL::Kernel::setf", "a", sv, "OpenCL::Sampler"); size = sizeof (arg.ca); break;
2207 case 'e': arg.ca = SvCLOBJ ("OpenCL::Kernel::setf", "e", sv, "OpenCL::Event" ); size = sizeof (arg.ce); break;
2208
2209 default:
2210 croak ("OpenCL::Kernel::setf format character '%c' not supported", type);
2211 }
2212
2213 clSetKernelArg (self, i - 2, size, &arg);
2214 }
2215
2216 if (i != items)
2217 croak ("OpenCL::Kernel::setf format string too short (too many arguments)");
2218
2219void
1762set_char (OpenCL::Kernel self, cl_uint idx, cl_char value) 2220set_char (OpenCL::Kernel self, cl_uint idx, cl_char value)
1763 CODE: 2221 CODE:
1764 clSetKernelArg (self, idx, sizeof (value), &value); 2222 clSetKernelArg (self, idx, sizeof (value), &value);
1765 2223
1766void 2224void
1822set_buffer (OpenCL::Kernel self, cl_uint idx, OpenCL::Buffer_ornull value) 2280set_buffer (OpenCL::Kernel self, cl_uint idx, OpenCL::Buffer_ornull value)
1823 CODE: 2281 CODE:
1824 clSetKernelArg (self, idx, sizeof (value), &value); 2282 clSetKernelArg (self, idx, sizeof (value), &value);
1825 2283
1826void 2284void
1827set_image2d (OpenCL::Kernel self, cl_uint idx, OpenCL::Image2D_ornull value) 2285set_image (OpenCL::Kernel self, cl_uint idx, OpenCL::Image_ornull value)
1828 CODE: 2286 ALIAS:
1829 clSetKernelArg (self, idx, sizeof (value), &value); 2287 set_image2d = 0
1830 2288 set_image3d = 0
1831void
1832set_image3d (OpenCL::Kernel self, cl_uint idx, OpenCL::Image3D_ornull value)
1833 CODE: 2289 CODE:
1834 clSetKernelArg (self, idx, sizeof (value), &value); 2290 clSetKernelArg (self, idx, sizeof (value), &value);
1835 2291
1836void 2292void
1837set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value) 2293set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value)
1885 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0)); 2341 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0));
1886 EXTEND (SP, 1); 2342 EXTEND (SP, 1);
1887 const int i = 0; 2343 const int i = 0;
1888 { 2344 {
1889 NEED_SUCCESS (RetainContext, (value [i])); 2345 NEED_SUCCESS (RetainContext, (value [i]));
1890 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2346 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1891 } 2347 }
1892 2348
1893void 2349void
1894program (OpenCL::Kernel self) 2350program (OpenCL::Kernel self)
1895 PPCODE: 2351 PPCODE:
1897 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0)); 2353 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0));
1898 EXTEND (SP, 1); 2354 EXTEND (SP, 1);
1899 const int i = 0; 2355 const int i = 0;
1900 { 2356 {
1901 NEED_SUCCESS (RetainProgram, (value [i])); 2357 NEED_SUCCESS (RetainProgram, (value [i]));
1902 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Program", value [i])); 2358 PUSH_CLOBJ ("OpenCL::Program", value [i]);
1903 } 2359 }
1904 2360
1905#END:kernel 2361#END:kernel
1906 2362
1907void 2363void
1967wait (OpenCL::Event self) 2423wait (OpenCL::Event self)
1968 CODE: 2424 CODE:
1969 clWaitForEvents (1, &self); 2425 clWaitForEvents (1, &self);
1970 2426
1971void 2427void
2428cb (OpenCL::Event self, cl_int command_exec_callback_type, SV *cb)
2429 CODE:
2430 clSetEventCallback (self, command_exec_callback_type, eq_event_notify, SvREFCNT_inc (s_get_cv (cb)));
2431
2432void
1972info (OpenCL::Event self, cl_event_info name) 2433info (OpenCL::Event self, cl_event_info name)
1973 PPCODE: 2434 PPCODE:
1974 INFO (Event) 2435 INFO (Event)
1975 2436
1976#BEGIN:event 2437#BEGIN:event
1982 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0)); 2443 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0));
1983 EXTEND (SP, 1); 2444 EXTEND (SP, 1);
1984 const int i = 0; 2445 const int i = 0;
1985 { 2446 {
1986 NEED_SUCCESS (RetainCommandQueue, (value [i])); 2447 NEED_SUCCESS (RetainCommandQueue, (value [i]));
1987 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Queue", value [i])); 2448 PUSH_CLOBJ ("OpenCL::Queue", value [i]);
1988 } 2449 }
1989 2450
1990void 2451void
1991command_type (OpenCL::Event self) 2452command_type (OpenCL::Event self)
1992 PPCODE: 2453 PPCODE:
2015 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0)); 2476 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0));
2016 EXTEND (SP, 1); 2477 EXTEND (SP, 1);
2017 const int i = 0; 2478 const int i = 0;
2018 { 2479 {
2019 NEED_SUCCESS (RetainContext, (value [i])); 2480 NEED_SUCCESS (RetainContext, (value [i]));
2020 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2481 PUSH_CLOBJ ("OpenCL::Context", value [i]);
2021 } 2482 }
2022 2483
2023#END:event 2484#END:event
2024 2485
2025void 2486void

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines