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

Comparing OpenCL/OpenCL.xs (file contents):
Revision 1.35 by root, Sat Apr 21 17:47:23 2012 UTC vs.
Revision 1.52 by root, Tue Apr 24 23:53:12 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 4
5#define X_STACKSIZE sizeof (void *) * 512 * 1024 // 2-4mb should be enough, really
6#include "xthread.h"
7#include "schmorp.h"
8
5#ifdef I_DLFCN 9#ifdef I_DLFCN
6 #include <dlfcn.h> 10 #include <dlfcn.h>
11#endif
12
13// how stupid is that, the 1.2 header files define CL_VERSION_1_1,
14// but then fail to define the api functions unless you ALSO define
15// this. This breaks 100% of the opencl 1.1 apps, for what reason?
16// after all, the functions are deprecated, not removed.
17// in addition, you cannot test for this in any future-proof way.
18// each time a new opencl version comes out, you need to make a new
19// release.
20#define CL_USE_DEPRECATED_OPENCL_1_2_APIS /* just guessing, you stupid idiots */
21
22#ifndef PREFER_1_1
23 #define PREFER_1_1 1
24#endif
25
26#if PREFER_1_1
27 #define CL_USE_DEPRECATED_OPENCL_1_1_APIS
7#endif 28#endif
8 29
9#ifdef __APPLE__ 30#ifdef __APPLE__
10 #include <OpenCL/opencl.h> 31 #include <OpenCL/opencl.h>
11#else 32#else
12 #include <CL/opencl.h> 33 #include <CL/opencl.h>
34#endif
35
36#ifndef CL_VERSION_1_2
37 #undef PREFER_1_1
38 #define PREFER_1_1 1
13#endif 39#endif
14 40
15typedef cl_platform_id OpenCL__Platform; 41typedef cl_platform_id OpenCL__Platform;
16typedef cl_device_id OpenCL__Device; 42typedef cl_device_id OpenCL__Device;
17typedef cl_context OpenCL__Context; 43typedef cl_context OpenCL__Context;
89 enum { buffers = 3 }; 115 enum { buffers = 3 };
90 static int idx; 116 static int idx;
91 static void *buf [buffers]; 117 static void *buf [buffers];
92 static size_t len [buffers]; 118 static size_t len [buffers];
93 119
94 idx = ++idx % buffers; 120 idx = (idx + 1) % buffers;
95 121
96 if (len [idx] < size) 122 if (len [idx] < size)
97 { 123 {
98 free (buf [idx]); 124 free (buf [idx]);
99 len [idx] = ((size + 31) & ~4095) + 4096 - 32; 125 len [idx] = ((size + 31) & ~4095) + 4096 - 32;
164 190
165#define NEED_SUCCESS_ARG(retdecl, name, args) \ 191#define NEED_SUCCESS_ARG(retdecl, name, args) \
166 retdecl = cl ## name args; \ 192 retdecl = cl ## name args; \
167 if (res) \ 193 if (res) \
168 FAIL (name); 194 FAIL (name);
169
170/*****************************************************************************/
171
172#define NEW_MORTAL_OBJ(class,ptr) sv_setref_pv (sv_newmortal (), class, ptr)
173#define XPUSH_NEW_OBJ(class,ptr) XPUSHs (NEW_MORTAL_OBJ (class, ptr))
174
175static void *
176SvPTROBJ (const char *func, const char *svname, SV *sv, const char *pkg)
177{
178 if (SvROK (sv) && sv_derived_from (sv, pkg))
179 return (void *)SvIV (SvRV (sv));
180
181 croak ("%s: %s is not of type %s", func, svname, pkg);
182}
183 195
184/*****************************************************************************/ 196/*****************************************************************************/
185 197
186static cl_context_properties * 198static cl_context_properties *
187SvCONTEXTPROPERTIES (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)
248 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);
249} 261}
250 262
251/*****************************************************************************/ 263/*****************************************************************************/
252 264
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 {
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);
284 }
285
286 croak ("%s: %s is not of type %s", func, svname, pkg);
287}
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
325/*****************************************************************************/
326/* callback stuff */
327
328/* default context callback, log to stderr */
329static void CL_CALLBACK
330context_default_notify (const char *msg, const void *info, size_t cb, void *data)
331{
332 fprintf (stderr, "OpenCL Context Notify: %s\n", msg);
333}
334
335typedef struct
336{
337 int free_cb;
338 void (*push)(void *data1, void *data2, void *data3);
339} eq_vtbl;
340
341typedef struct eq_item
342{
343 struct eq_item *next;
344 eq_vtbl *vtbl;
345 SV *cb;
346 void *data1, *data2, *data3;
347} eq_item;
348
349static void (*eq_signal_func)(void *signal_arg, int value);
350static void *eq_signal_arg;
351static xmutex_t eq_lock = X_MUTEX_INIT;
352static eq_item *eq_head, *eq_tail;
353
354static void
355eq_enq (eq_vtbl *vtbl, SV *cb, void *data1, void *data2, void *data3)
356{
357 eq_item *item = malloc (sizeof (eq_item));
358
359 item->next = 0;
360 item->vtbl = vtbl;
361 item->cb = cb;
362 item->data1 = data1;
363 item->data2 = data2;
364 item->data3 = data3;
365
366 X_LOCK (eq_lock);
367
368 *(eq_head ? &eq_tail->next : &eq_head) = item;
369 eq_tail = item;
370
371 X_UNLOCK (eq_lock);
372
373 eq_signal_func (eq_signal_arg, 0);
374}
375
376static eq_item *
377eq_dec (void)
378{
379 eq_item *res;
380
381 X_LOCK (eq_lock);
382
383 res = eq_head;
384 if (res)
385 eq_head = res->next;
386
387 X_UNLOCK (eq_lock);
388
389 return res;
390}
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
401static void
402eq_poll (void)
403{
404 eq_item *item;
405
406 while ((item = eq_dec ()))
407 {
408 ENTER;
409 SAVETMPS;
410
411 dSP;
412 PUSHMARK (SP);
413 EXTEND (SP, 2);
414
415 if (item->vtbl->free_cb)
416 sv_2mortal (item->cb);
417
418 PUTBACK;
419 item->vtbl->push (item->data1, item->data2, item->data3);
420
421 SV *cb = item->cb;
422 free (item);
423
424 call_sv (cb, G_DISCARD | G_VOID);
425
426 FREETMPS;
427 LEAVE;
428 }
429}
430
431static void
432eq_poll_interrupt (pTHX_ void *c_arg, int value)
433{
434 eq_poll ();
435}
436
437/*****************************************************************************/
438/* context notify */
439
440static void
441eq_context_push (void *data1, void *data2, void *data3)
442{
443 dSP;
444 PUSHs (sv_2mortal (newSVpv (data1, 0)));
445 PUSHs (sv_2mortal (newSVpvn (data2, (STRLEN)data3)));
446 PUTBACK;
447
448 free (data1);
449 free (data2);
450}
451
452static eq_vtbl eq_context_vtbl = { 0, eq_context_push };
453
454static void CL_CALLBACK
455eq_context_notify (const char *msg, const void *pvt, size_t cb, void *user_data)
456{
457 void *pvt_copy = malloc (cb);
458 memcpy (pvt_copy, pvt, cb);
459 eq_enq (&eq_context_vtbl, user_data, strdup (msg), pvt_copy, (void *)cb);
460}
461
462#define CONTEXT_NOTIFY_CALLBACK \
463 void (CL_CALLBACK *pfn_notify)(const char *, const void *, size_t, void *) = context_default_notify; \
464 void *user_data = 0; \
465 \
466 if (SvOK (notify)) \
467 { \
468 pfn_notify = eq_context_notify; \
469 user_data = s_get_cv (notify); \
470 }
471
472static SV *
473new_clobj_context (cl_context ctx, void *user_data)
474{
475 SV *sv = NEW_CLOBJ ("OpenCL::Context", ctx);
476
477 if (user_data)
478 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0);
479
480 return sv;
481}
482
483#define XPUSH_CLOBJ_CONTEXT XPUSHs (new_clobj_context (ctx, user_data));
484
485/*****************************************************************************/
486/* build/compile/link notify */
487
488static void
489eq_program_push (void *data1, void *data2, void *data3)
490{
491 dSP;
492 PUSH_CLOBJ ("OpenCL::Program", data1);
493 PUTBACK;
494}
495
496static eq_vtbl eq_program_vtbl = { 1, eq_program_push };
497
498static void CL_CALLBACK
499eq_program_notify (cl_program program, void *user_data)
500{
501 eq_enq (&eq_program_vtbl, user_data, (void *)program, 0, 0);
502}
503
504struct build_args
505{
506 cl_program program;
507 char *options;
508 void *user_data;
509 cl_uint num_devices;
510};
511
512X_THREAD_PROC (build_program_thread)
513{
514 struct build_args *arg = thr_arg;
515
516 clBuildProgram (arg->program, arg->num_devices, arg->num_devices ? (void *)(arg + 1) : 0, arg->options, 0, 0);
517
518 if (arg->user_data)
519 eq_program_notify (arg->program, arg->user_data);
520 else
521 clReleaseProgram (arg->program);
522
523 free (arg->options);
524 free (arg);
525}
526
527static void
528build_program_async (cl_program program, cl_uint num_devices, const cl_device_id *device_list, const char *options, void *user_data)
529{
530 struct build_args *arg = malloc (sizeof (struct build_args) + sizeof (*device_list) * num_devices);
531
532 arg->program = program;
533 arg->options = strdup (options);
534 arg->user_data = user_data;
535 arg->num_devices = num_devices;
536 memcpy (arg + 1, device_list, sizeof (*device_list) * num_devices);
537
538 xthread_t id;
539 thread_create (&id, build_program_thread, arg);
540}
541
542/*****************************************************************************/
543/* event objects */
544
545static void
546eq_event_push (void *data1, void *data2, void *data3)
547{
548 dSP;
549 PUSH_CLOBJ ("OpenCL::Event", data1);
550 PUSHs (sv_2mortal (newSViv ((IV)data2)));
551 PUTBACK;
552}
553
554static eq_vtbl eq_event_vtbl = { 1, eq_event_push };
555
556static void CL_CALLBACK
557eq_event_notify (cl_event event, cl_int event_command_exec_status, void *user_data)
558{
559 clRetainEvent (event);
560 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0);
561}
562
563/*****************************************************************************/
564
253static size_t 565static size_t
254img_row_pitch (cl_mem img) 566img_row_pitch (cl_mem img)
255{ 567{
256 size_t res; 568 size_t res;
257 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0); 569 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0);
271 583
272 do 584 do
273 { 585 {
274 --count; 586 --count;
275 if (SvOK (items [count])) 587 if (SvOK (items [count]))
276 list [i++] = SvPTROBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event"); 588 list [i++] = SvCLOBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event");
277 } 589 }
278 while (count); 590 while (count);
279 591
280 *rcount = i; 592 *rcount = i;
281 593
299} 611}
300 612
301MODULE = OpenCL PACKAGE = OpenCL 613MODULE = OpenCL PACKAGE = OpenCL
302 614
303PROTOTYPES: ENABLE 615PROTOTYPES: ENABLE
616
617void
618poll ()
619 CODE:
620 eq_poll ();
621
622void
623_eq_initialise (IV func, IV arg)
624 CODE:
625 eq_signal_func = (void (*)(void *, int))func;
626 eq_signal_arg = (void*)arg;
304 627
305BOOT: 628BOOT:
306{ 629{
307 HV *stash = gv_stashpv ("OpenCL", 1); 630 HV *stash = gv_stashpv ("OpenCL", 1);
308 static const ivstr *civ, const_iv[] = { 631 static const ivstr *civ, const_iv[] = {
317 { sizeof (cl_half ), "SIZEOF_HALF" }, 640 { sizeof (cl_half ), "SIZEOF_HALF" },
318 { sizeof (cl_float ), "SIZEOF_FLOAT" }, 641 { sizeof (cl_float ), "SIZEOF_FLOAT" },
319 { sizeof (cl_double), "SIZEOF_DOUBLE" }, 642 { sizeof (cl_double), "SIZEOF_DOUBLE" },
320#include "constiv.h" 643#include "constiv.h"
321 }; 644 };
645
322 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--) 646 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--)
323 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv)); 647 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv));
648
649 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt);
324} 650}
325 651
326cl_int 652cl_int
327errno () 653errno ()
328 CODE: 654 CODE:
329 errno = res; 655 RETVAL = res;
656 OUTPUT:
657 RETVAL
330 658
331const char * 659const char *
332err2str (cl_int err) 660err2str (cl_int err)
333 661
334const char * 662const char *
345 list = tmpbuf (sizeof (*list) * count); 673 list = tmpbuf (sizeof (*list) * count);
346 NEED_SUCCESS (GetPlatformIDs, (count, list, 0)); 674 NEED_SUCCESS (GetPlatformIDs, (count, list, 0));
347 675
348 EXTEND (SP, count); 676 EXTEND (SP, count);
349 for (i = 0; i < count; ++i) 677 for (i = 0; i < count; ++i)
350 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Platform", list [i])); 678 PUSH_CLOBJ ("OpenCL::Platform", list [i]);
351 679
352void 680void
353context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, FUTURE notify = 0) 681context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
354 PPCODE: 682 PPCODE:
683 CONTEXT_NOTIFY_CALLBACK;
355 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, 0, 0, &res)); 684 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, 0, 0, &res));
356 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 685 XPUSH_CLOBJ_CONTEXT;
357 686
358void 687void
359context (cl_context_properties *properties = 0, FUTURE devices, FUTURE notify = 0) 688context (FUTURE properties, FUTURE devices, FUTURE notify)
360 PPCODE: 689 PPCODE:
361 /* der Gipfel der Kunst */ 690 /* der Gipfel der Kunst */
362 691
363void 692void
364wait_for_events (...) 693wait_for_events (...)
372 701
373void 702void
374info (OpenCL::Platform self, cl_platform_info name) 703info (OpenCL::Platform self, cl_platform_info name)
375 PPCODE: 704 PPCODE:
376 INFO (Platform) 705 INFO (Platform)
706
707void
708unload_compiler (OpenCL::Platform self)
709 CODE:
710#if CL_VERSION_1_2
711 clUnloadPlatformCompiler (self);
712#endif
377 713
378#BEGIN:platform 714#BEGIN:platform
379 715
380void 716void
381profile (OpenCL::Platform self) 717profile (OpenCL::Platform self)
410 EXTEND (SP, count); 746 EXTEND (SP, count);
411 for (i = 0; i < count; ++i) 747 for (i = 0; i < count; ++i)
412 PUSHs (sv_setref_pv (sv_newmortal (), "OpenCL::Device", list [i])); 748 PUSHs (sv_setref_pv (sv_newmortal (), "OpenCL::Device", list [i]));
413 749
414void 750void
415context (OpenCL::Platform self, cl_context_properties *properties = 0, SV *devices, FUTURE notify = 0) 751context (OpenCL::Platform self, cl_context_properties *properties, SV *devices, SV *notify = &PL_sv_undef)
416 PPCODE: 752 PPCODE:
417 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV) 753 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV)
418 croak ("OpenCL::Platform::context argument 'device' must be an arrayref with device objects, in call"); 754 croak ("OpenCL::Platform::context argument 'device' must be an arrayref with device objects, in call");
419 755
420 AV *av = (AV *)SvRV (devices); 756 AV *av = (AV *)SvRV (devices);
421 cl_uint num_devices = av_len (av) + 1; 757 cl_uint num_devices = av_len (av) + 1;
422 cl_device_id *device_list = tmpbuf (sizeof (cl_device_id) * num_devices); 758 cl_device_id *device_list = tmpbuf (sizeof (cl_device_id) * num_devices);
759
423 int i; 760 int i;
424
425 for (i = num_devices; i--; ) 761 for (i = num_devices; i--; )
426 device_list [i] = SvPTROBJ ("clCreateContext", "devices", *av_fetch (av, i, 0), "OpenCL::Device"); 762 device_list [i] = SvCLOBJ ("clCreateContext", "devices", *av_fetch (av, i, 0), "OpenCL::Device");
427 763
764 CONTEXT_NOTIFY_CALLBACK;
428 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (properties, num_devices, device_list, 0, 0, &res)); 765 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (properties, num_devices, device_list, pfn_notify, user_data, &res));
429 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 766 XPUSH_CLOBJ_CONTEXT;
430 767
431void 768void
432context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, FUTURE notify = 0) 769context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
433 PPCODE: 770 PPCODE:
434 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self }; 771 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
435 cl_context_properties *props = SvCONTEXTPROPERTIES ("OpenCL::Platform::context_from_type", "properties", properties, extra, 2); 772 cl_context_properties *props = SvCONTEXTPROPERTIES ("OpenCL::Platform::context_from_type", "properties", properties, extra, 2);
773
774 CONTEXT_NOTIFY_CALLBACK;
436 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, 0, 0, &res)); 775 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, 0, 0, &res));
437 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 776 XPUSH_CLOBJ_CONTEXT;
438 777
439MODULE = OpenCL PACKAGE = OpenCL::Device 778MODULE = OpenCL PACKAGE = OpenCL::Device
440 779
441void 780void
442info (OpenCL::Device self, cl_device_info name) 781info (OpenCL::Device self, cl_device_info name)
481 native_vector_width_int = CL_DEVICE_NATIVE_VECTOR_WIDTH_INT 820 native_vector_width_int = CL_DEVICE_NATIVE_VECTOR_WIDTH_INT
482 native_vector_width_long = CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG 821 native_vector_width_long = CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG
483 native_vector_width_float = CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT 822 native_vector_width_float = CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT
484 native_vector_width_double = CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE 823 native_vector_width_double = CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE
485 native_vector_width_half = CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF 824 native_vector_width_half = CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF
486 reference_count_ext = CL_DEVICE_REFERENCE_COUNT_EXT 825 reference_count_ext = CL_DEVICE_REFERENCE_COUNT_EXT
487 PPCODE: 826 PPCODE:
488 cl_uint value [1]; 827 cl_uint value [1];
489 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0)); 828 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
490 EXTEND (SP, 1); 829 EXTEND (SP, 1);
491 const int i = 0; 830 const int i = 0;
615 cl_platform_id value [1]; 954 cl_platform_id value [1];
616 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0)); 955 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0));
617 EXTEND (SP, 1); 956 EXTEND (SP, 1);
618 const int i = 0; 957 const int i = 0;
619 { 958 {
620 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Platform", value [i])); 959 PUSH_CLOBJ ("OpenCL::Platform", value [i]);
621 } 960 }
622 961
623void 962void
624name (OpenCL::Device self) 963name (OpenCL::Device self)
625 ALIAS: 964 ALIAS:
644 cl_device_id value [1]; 983 cl_device_id value [1];
645 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0)); 984 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0));
646 EXTEND (SP, 1); 985 EXTEND (SP, 1);
647 const int i = 0; 986 const int i = 0;
648 { 987 {
649 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 988 PUSH_CLOBJ ("OpenCL::Device", value [i]);
650 } 989 }
651 990
652void 991void
653partition_types_ext (OpenCL::Device self) 992partition_types_ext (OpenCL::Device self)
654 ALIAS: 993 ALIAS:
681 1020
682void 1021void
683queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0) 1022queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0)
684 PPCODE: 1023 PPCODE:
685 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res)); 1024 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res));
686 XPUSH_NEW_OBJ ("OpenCL::Queue", queue); 1025 XPUSH_CLOBJ ("OpenCL::Queue", queue);
687 1026
688void 1027void
689user_event (OpenCL::Context self) 1028user_event (OpenCL::Context self)
690 PPCODE: 1029 PPCODE:
691 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res)); 1030 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res));
692 XPUSH_NEW_OBJ ("OpenCL::UserEvent", ev); 1031 XPUSH_CLOBJ ("OpenCL::UserEvent", ev);
693 1032
694void 1033void
695buffer (OpenCL::Context self, cl_mem_flags flags, size_t len) 1034buffer (OpenCL::Context self, cl_mem_flags flags, size_t len)
696 PPCODE: 1035 PPCODE:
697 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) 1036 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))
698 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?"); 1037 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?");
699 1038
700 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res)); 1039 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res));
701 XPUSH_NEW_OBJ ("OpenCL::BufferObj", mem); 1040 XPUSH_CLOBJ ("OpenCL::BufferObj", mem);
702 1041
703void 1042void
704buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data) 1043buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data)
705 PPCODE: 1044 PPCODE:
706 STRLEN len; 1045 STRLEN len;
707 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0; 1046 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
708 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) 1047 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)))
709 croak ("OpenCL::Context::buffer_sv: you have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?"); 1048 croak ("OpenCL::Context::buffer_sv: you have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?");
710 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res)); 1049 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res));
711 XPUSH_NEW_OBJ ("OpenCL::BufferObj", mem); 1050 XPUSH_CLOBJ ("OpenCL::BufferObj", mem);
712 1051
1052#if CL_VERSION_1_2
1053
713void 1054void
714image2d (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) 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)
715 PPCODE: 1056 PPCODE:
716 STRLEN len; 1057 STRLEN len;
717 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0; 1058 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
718 const cl_image_format format = { channel_order, channel_type }; 1059 const cl_image_format format = { channel_order, channel_type };
1060 const cl_image_desc desc = {
1061 type,
1062 width, height, depth,
1063 array_size, row_pitch, slice_pitch,
1064 num_mip_level, num_samples,
1065 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ ("OpenCL::Context::Image", "data", data, "OpenCL::Buffer") : 0
1066 };
719 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res)); 1067 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
720 XPUSH_NEW_OBJ ("OpenCL::Image2D", mem); 1068 char *klass = "OpenCL::Image";
1069 switch (type)
1070 {
1071 case CL_MEM_OBJECT_IMAGE1D_BUFFER: klass = "OpenCL::Image1DBuffer"; break;
1072 case CL_MEM_OBJECT_IMAGE1D: klass = "OpenCL::Image1D"; break;
1073 case CL_MEM_OBJECT_IMAGE1D_ARRAY: klass = "OpenCL::Image2DArray"; break;
1074 case CL_MEM_OBJECT_IMAGE2D: klass = "OpenCL::Image2D"; break;
1075 case CL_MEM_OBJECT_IMAGE2D_ARRAY: klass = "OpenCL::Image2DArray"; break;
1076 case CL_MEM_OBJECT_IMAGE3D: klass = "OpenCL::Image3D"; break;
1077 }
1078 XPUSH_CLOBJ (klass, mem);
721 1079
1080#endif
1081
722void 1082void
723image3d (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) 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)
724 PPCODE: 1084 PPCODE:
725 STRLEN len; 1085 STRLEN len;
726 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0; 1086 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
727 const cl_image_format format = { channel_order, channel_type }; 1087 const cl_image_format format = { channel_order, channel_type };
1088#if PREFER_1_1
1089 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res));
1090#else
1091 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));
1093#endif
1094 XPUSH_CLOBJ ("OpenCL::Image2D", mem);
1095
1096void
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)
1098 PPCODE:
1099 STRLEN len;
1100 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
1101 const cl_image_format format = { channel_order, channel_type };
1102#if PREFER_1_1
728 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res)); 1103 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res));
1104#else
1105 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));
1107#endif
729 XPUSH_NEW_OBJ ("OpenCL::Image3D", mem); 1108 XPUSH_CLOBJ ("OpenCL::Image3D", mem);
730 1109
731#if cl_apple_gl_sharing || cl_khr_gl_sharing 1110#if cl_apple_gl_sharing || cl_khr_gl_sharing
732 1111
733void 1112void
734gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj) 1113gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj)
735 PPCODE: 1114 PPCODE:
736 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res)); 1115 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res));
737 XPUSH_NEW_OBJ ("OpenCL::BufferObj", mem); 1116 XPUSH_CLOBJ ("OpenCL::BufferObj", mem);
1117
1118void
1119gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer)
1120 PPCODE:
1121 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res));
1122 XPUSH_CLOBJ ("OpenCL::Image2D", mem);
1123
1124#if CL_VERSION_1_2
1125
1126void
1127gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1128 ALIAS:
1129 PPCODE:
1130 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture, (self, flags, target, miplevel, texture, &res));
1131 cl_gl_object_type type;
1132 NEED_SUCCESS (GetGLObjectInfo, (mem, &type, 0)); // TODO: use target instead?
1133 char *klass = "OpenCL::Memory";
1134 switch (type)
1135 {
1136 case CL_GL_OBJECT_TEXTURE_BUFFER: klass = "OpenCL::Image1DBuffer"; break;
1137 case CL_GL_OBJECT_TEXTURE1D: klass = "OpenCL::Image1D"; break;
1138 case CL_GL_OBJECT_TEXTURE1D_ARRAY: klass = "OpenCL::Image2DArray"; break;
1139 case CL_GL_OBJECT_TEXTURE2D: klass = "OpenCL::Image2D"; break;
1140 case CL_GL_OBJECT_TEXTURE2D_ARRAY: klass = "OpenCL::Image2DArray"; break;
1141 case CL_GL_OBJECT_TEXTURE3D: klass = "OpenCL::Image3D"; break;
1142 }
1143 XPUSH_CLOBJ (klass, mem);
1144
1145#endif
738 1146
739void 1147void
740gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1148gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
741 PPCODE: 1149 PPCODE:
1150#if PREFER_1_1
742 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res)); 1151 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res));
1152#else
1153 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1154#endif
743 XPUSH_NEW_OBJ ("OpenCL::Image2D", mem); 1155 XPUSH_CLOBJ ("OpenCL::Image2D", mem);
744 1156
745void 1157void
746gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture) 1158gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
747 PPCODE: 1159 PPCODE:
1160#if PREFER_1_1
748 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res)); 1161 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res));
1162#else
1163 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1164#endif
749 XPUSH_NEW_OBJ ("OpenCL::Image3D", mem); 1165 XPUSH_CLOBJ ("OpenCL::Image3D", mem);
750
751void
752gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer)
753 PPCODE:
754 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res));
755 XPUSH_NEW_OBJ ("OpenCL::Image2D", mem);
756 1166
757#endif 1167#endif
758 1168
759void 1169void
760supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type) 1170supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type)
780 1190
781void 1191void
782sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode) 1192sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode)
783 PPCODE: 1193 PPCODE:
784 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res)); 1194 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res));
785 XPUSH_NEW_OBJ ("OpenCL::Sampler", sampler); 1195 XPUSH_CLOBJ ("OpenCL::Sampler", sampler);
786 1196
787void 1197void
788program_with_source (OpenCL::Context self, SV *program) 1198program_with_source (OpenCL::Context self, SV *program)
789 PPCODE: 1199 PPCODE:
790 STRLEN len; 1200 STRLEN len;
791 size_t len2; 1201 size_t len2;
792 const char *ptr = SvPVbyte (program, len); 1202 const char *ptr = SvPVbyte (program, len);
793 1203
794 len2 = len; 1204 len2 = len;
795 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res)); 1205 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res));
796 XPUSH_NEW_OBJ ("OpenCL::Program", prog); 1206 XPUSH_CLOBJ ("OpenCL::Program", prog);
797 1207
798#BEGIN:context 1208#BEGIN:context
799 1209
800void 1210void
801reference_count (OpenCL::Context self) 1211reference_count (OpenCL::Context self)
818 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0)); 1228 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0));
819 int i, n = size / sizeof (*value); 1229 int i, n = size / sizeof (*value);
820 EXTEND (SP, n); 1230 EXTEND (SP, n);
821 for (i = 0; i < n; ++i) 1231 for (i = 0; i < n; ++i)
822 { 1232 {
823 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 1233 PUSH_CLOBJ ("OpenCL::Device", value [i]);
824 } 1234 }
825 1235
826void 1236void
827properties (OpenCL::Context self) 1237properties (OpenCL::Context self)
828 PPCODE: 1238 PPCODE:
855 SvPOK_only (data); 1265 SvPOK_only (data);
856 SvCUR_set (data, len); 1266 SvCUR_set (data, len);
857 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1267 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
858 1268
859 if (ev) 1269 if (ev)
860 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1270 XPUSH_CLOBJ ("OpenCL::Event", ev);
861 1271
862void 1272void
863enqueue_write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...) 1273enqueue_write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...)
864 PPCODE: 1274 PPCODE:
865 cl_event ev = 0; 1275 cl_event ev = 0;
868 EVENT_LIST (5, items - 5); 1278 EVENT_LIST (5, items - 5);
869 1279
870 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1280 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
871 1281
872 if (ev) 1282 if (ev)
873 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1283 XPUSH_CLOBJ ("OpenCL::Event", ev);
1284
1285#if CL_VERSION_1_2
1286
1287void
1288enqueue_fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...)
1289 PPCODE:
1290 cl_event ev = 0;
1291 STRLEN len;
1292 char *ptr = SvPVbyte (data, len);
1293 EVENT_LIST (5, items - 5);
1294
1295 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1296
1297 if (ev)
1298 XPUSH_CLOBJ ("OpenCL::Event", ev);
1299
1300void
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, ...)
1302 PPCODE:
1303 cl_event ev = 0;
1304 STRLEN len;
1305 const size_t origin [3] = { x, y, z };
1306 const size_t region [3] = { width, height, depth };
1307 EVENT_LIST (12, items - 12);
1308
1309 const cl_float c_f [4] = { r, g, b, a };
1310 const cl_uint c_u [4] = { r, g, b, a };
1311 const cl_int c_s [4] = { r, g, b, a };
1312 const void *c_fus [3] = { &c_f, &c_u, &c_s };
1313 static const char fus [] = { 0, 0, 0, 0, 0, 0, 0, 2, 2, 2, 1, 1, 1, 0, 0 };
1314 cl_image_format format;
1315 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
1316 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)
1318 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported.");
1319
1320 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type]],
1321 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1322
1323 if (ev)
1324 XPUSH_CLOBJ ("OpenCL::Event", ev);
1325
1326#endif
874 1327
875void 1328void
876enqueue_copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...) 1329enqueue_copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...)
877 PPCODE: 1330 PPCODE:
878 cl_event ev = 0; 1331 cl_event ev = 0;
879 EVENT_LIST (6, items - 6); 1332 EVENT_LIST (6, items - 6);
880 1333
881 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1334 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
882 1335
883 if (ev) 1336 if (ev)
884 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1337 XPUSH_CLOBJ ("OpenCL::Event", ev);
885 1338
886void 1339void
887enqueue_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, ...) 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, ...)
888 PPCODE: 1341 PPCODE:
889 cl_event ev = 0; 1342 cl_event ev = 0;
911 SvPOK_only (data); 1364 SvPOK_only (data);
912 SvCUR_set (data, len); 1365 SvCUR_set (data, len);
913 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)); 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));
914 1367
915 if (ev) 1368 if (ev)
916 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1369 XPUSH_CLOBJ ("OpenCL::Event", ev);
917 1370
918void 1371void
919enqueue_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, ...) 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, ...)
920 PPCODE: 1373 PPCODE:
921 cl_event ev = 0; 1374 cl_event ev = 0;
941 size_t min_len = host_row_pitch * host_slice_pitch * region [2]; 1394 size_t min_len = host_row_pitch * host_slice_pitch * region [2];
942 1395
943 if (len < min_len) 1396 if (len < min_len)
944 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1397 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
945 1398
946 NEED_SUCCESS (EnqueueWriteBufferRect, (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)); 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));
947 1400
948 if (ev) 1401 if (ev)
949 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1402 XPUSH_CLOBJ ("OpenCL::Event", ev);
950 1403
951void 1404void
952enqueue_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, ...) 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, ...)
953 PPCODE: 1406 PPCODE:
954 cl_event ev = 0; 1407 cl_event ev = 0;
958 EVENT_LIST (16, items - 16); 1411 EVENT_LIST (16, items - 16);
959 1412
960 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)); 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));
961 1414
962 if (ev) 1415 if (ev)
963 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1416 XPUSH_CLOBJ ("OpenCL::Event", ev);
964 1417
965void 1418void
966enqueue_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, ...) 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, ...)
967 PPCODE: 1420 PPCODE:
968 cl_event ev = 0; 1421 cl_event ev = 0;
983 SvPOK_only (data); 1436 SvPOK_only (data);
984 SvCUR_set (data, len); 1437 SvCUR_set (data, len);
985 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)); 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));
986 1439
987 if (ev) 1440 if (ev)
988 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1441 XPUSH_CLOBJ ("OpenCL::Event", ev);
989 1442
990void 1443void
991enqueue_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, ...) 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, ...)
992 PPCODE: 1445 PPCODE:
993 cl_event ev = 0; 1446 cl_event ev = 0;
1006 size_t min_len = slice_pitch ? slice_pitch * depth : row_pitch * height; 1459 size_t min_len = slice_pitch ? slice_pitch * depth : row_pitch * height;
1007 1460
1008 if (len < min_len) 1461 if (len < min_len)
1009 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1462 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
1010 1463
1011 NEED_SUCCESS (EnqueueWriteImage, (self, dst, blocking, dst_origin, region, row_pitch, slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 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));
1012 1465
1013 if (ev) 1466 if (ev)
1014 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1467 XPUSH_CLOBJ ("OpenCL::Event", ev);
1015 1468
1016void 1469void
1017enqueue_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, ...) 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, ...)
1018 PPCODE: 1471 PPCODE:
1019 cl_event ev = 0; 1472 cl_event ev = 0;
1023 EVENT_LIST (12, items - 12); 1476 EVENT_LIST (12, items - 12);
1024 1477
1025 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1478 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1026 1479
1027 if (ev) 1480 if (ev)
1028 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1481 XPUSH_CLOBJ ("OpenCL::Event", ev);
1029 1482
1030void 1483void
1031enqueue_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, ...) 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, ...)
1032 PPCODE: 1485 PPCODE:
1033 cl_event ev = 0; 1486 cl_event ev = 0;
1036 EVENT_LIST (10, items - 10); 1489 EVENT_LIST (10, items - 10);
1037 1490
1038 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1491 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1039 1492
1040 if (ev) 1493 if (ev)
1041 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1494 XPUSH_CLOBJ ("OpenCL::Event", ev);
1042 1495
1043void 1496void
1044enqueue_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, ...) 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, ...)
1045 PPCODE: 1498 PPCODE:
1046 cl_event ev = 0; 1499 cl_event ev = 0;
1049 EVENT_LIST (10, items - 10); 1502 EVENT_LIST (10, items - 10);
1050 1503
1051 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1504 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1052 1505
1053 if (ev) 1506 if (ev)
1054 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1507 XPUSH_CLOBJ ("OpenCL::Event", ev);
1055 1508
1056void 1509void
1057enqueue_task (OpenCL::Queue self, OpenCL::Kernel kernel, ...) 1510enqueue_task (OpenCL::Queue self, OpenCL::Kernel kernel, ...)
1058 PPCODE: 1511 PPCODE:
1059 cl_event ev = 0; 1512 cl_event ev = 0;
1060 EVENT_LIST (2, items - 2); 1513 EVENT_LIST (2, items - 2);
1061 1514
1062 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1515 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1063 1516
1064 if (ev) 1517 if (ev)
1065 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1518 XPUSH_CLOBJ ("OpenCL::Event", ev);
1066 1519
1067void 1520void
1068enqueue_nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...) 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, ...)
1069 PPCODE: 1522 PPCODE:
1070 cl_event ev = 0; 1523 cl_event ev = 0;
1098 gwo [i] = SvIV (AvARRAY (SvRV (global_work_offset))[i]); 1551 gwo [i] = SvIV (AvARRAY (SvRV (global_work_offset))[i]);
1099 } 1552 }
1100 1553
1101 if (SvOK (local_work_size)) 1554 if (SvOK (local_work_size))
1102 { 1555 {
1103 if (SvOK (local_work_size) && !SvROK (local_work_size) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV) 1556 if ((SvOK (local_work_size) && !SvROK (local_work_size)) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV)
1104 croak ("clEnqueueNDRangeKernel: global_work_size must be undef or an array reference"); 1557 croak ("clEnqueueNDRangeKernel: global_work_size must be undef or an array reference");
1105 1558
1106 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len) 1559 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len)
1107 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size"); 1560 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size");
1108 1561
1112 } 1565 }
1113 1566
1114 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1567 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1115 1568
1116 if (ev) 1569 if (ev)
1117 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1570 XPUSH_CLOBJ ("OpenCL::Event", ev);
1118 1571
1119#if cl_apple_gl_sharing || cl_khr_gl_sharing 1572#if cl_apple_gl_sharing || cl_khr_gl_sharing
1120 1573
1121void 1574void
1122enqueue_acquire_gl_objects (OpenCL::Queue self, SV *objects, ...) 1575enqueue_acquire_gl_objects (OpenCL::Queue self, SV *objects, ...)
1123 ALIAS: 1576 ALIAS:
1124 enqueue_release_gl_objects = 1 1577 enqueue_release_gl_objects = 1
1125 CODE: 1578 PPCODE:
1126 if (!SvROK (objects) || SvTYPE (SvRV (objects)) != SVt_PVAV) 1579 if (!SvROK (objects) || SvTYPE (SvRV (objects)) != SVt_PVAV)
1127 croak ("OpenCL::Queue::enqueue_acquire/release_gl_objects argument 'objects' must be an arrayref with memory objects, in call"); 1580 croak ("OpenCL::Queue::enqueue_acquire/release_gl_objects argument 'objects' must be an arrayref with memory objects, in call");
1128 1581
1129 cl_event ev = 0; 1582 cl_event ev = 0;
1130 EVENT_LIST (2, items - 2); 1583 EVENT_LIST (2, items - 2);
1132 cl_uint num_objects = av_len (av) + 1; 1585 cl_uint num_objects = av_len (av) + 1;
1133 cl_mem *object_list = tmpbuf (sizeof (cl_mem) * num_objects); 1586 cl_mem *object_list = tmpbuf (sizeof (cl_mem) * num_objects);
1134 int i; 1587 int i;
1135 1588
1136 for (i = num_objects; i--; ) 1589 for (i = num_objects; i--; )
1137 object_list [i] = SvPTROBJ ("OpenCL::Queue::enqueue_acquire/release_gl_objects", "objects", *av_fetch (av, i, 0), "OpenCL::Memory"); 1590 object_list [i] = SvCLOBJ ("OpenCL::Queue::enqueue_acquire/release_gl_objects", "objects", *av_fetch (av, i, 0), "OpenCL::Memory");
1138 1591
1139 if (ix) 1592 if (ix)
1140 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1593 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1141 else 1594 else
1142 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1595 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, num_objects, object_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1143 1596
1144 if (ev) 1597 if (ev)
1145 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1598 XPUSH_CLOBJ ("OpenCL::Event", ev);
1146 1599
1147#endif 1600#endif
1148 1601
1149void 1602void
1150enqueue_marker (OpenCL::Queue self)
1151 PPCODE:
1152 cl_event ev;
1153 NEED_SUCCESS (EnqueueMarker, (self, &ev));
1154 XPUSH_NEW_OBJ ("OpenCL::Event", ev);
1155
1156void
1157enqueue_wait_for_events (OpenCL::Queue self, ...) 1603enqueue_wait_for_events (OpenCL::Queue self, ...)
1158 CODE: 1604 CODE:
1159 EVENT_LIST (1, items - 1); 1605 EVENT_LIST (1, items - 1);
1606#if PREFER_1_1
1160 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr)); 1607 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
1608#else
1609 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0));
1610#endif
1161 1611
1162void 1612void
1613enqueue_marker (OpenCL::Queue self, ...)
1614 PPCODE:
1615 cl_event ev = 0;
1616 EVENT_LIST (1, items - 1);
1617#if PREFER_1_1
1618 if (!event_list_count)
1619 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
1620 else
1621#if CL_VERSION_1_2
1622 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1623#else
1624 {
1625 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr)); // also a barrier
1626 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
1627 }
1628#endif
1629#else
1630 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1631#endif
1632 if (ev)
1633 XPUSH_CLOBJ ("OpenCL::Event", ev);
1634
1635void
1163enqueue_barrier (OpenCL::Queue self) 1636enqueue_barrier (OpenCL::Queue self, ...)
1164 CODE: 1637 PPCODE:
1638 cl_event ev = 0;
1639 EVENT_LIST (1, items - 1);
1640#if PREFER_1_1
1641 if (!event_list_count && GIMME_V == G_VOID)
1165 NEED_SUCCESS (EnqueueBarrier, (self)); 1642 NEED_SUCCESS (EnqueueBarrier, (self));
1643 else
1644#if CL_VERSION_1_2
1645 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1646#else
1647 {
1648 if (event_list_count)
1649 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
1650
1651 if (GIMME_V != G_VOID)
1652 NEED_SUCCESS (EnqueueMarker, (self, &ev));
1653 }
1654#endif
1655#else
1656 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1657#endif
1658 if (ev)
1659 XPUSH_CLOBJ ("OpenCL::Event", ev);
1166 1660
1167void 1661void
1168flush (OpenCL::Queue self) 1662flush (OpenCL::Queue self)
1169 CODE: 1663 CODE:
1170 NEED_SUCCESS (Flush, (self)); 1664 NEED_SUCCESS (Flush, (self));
1188 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0)); 1682 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0));
1189 EXTEND (SP, 1); 1683 EXTEND (SP, 1);
1190 const int i = 0; 1684 const int i = 0;
1191 { 1685 {
1192 NEED_SUCCESS (RetainContext, (value [i])); 1686 NEED_SUCCESS (RetainContext, (value [i]));
1193 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 1687 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1194 } 1688 }
1195 1689
1196void 1690void
1197device (OpenCL::Queue self) 1691device (OpenCL::Queue self)
1198 PPCODE: 1692 PPCODE:
1199 cl_device_id value [1]; 1693 cl_device_id value [1];
1200 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0)); 1694 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0));
1201 EXTEND (SP, 1); 1695 EXTEND (SP, 1);
1202 const int i = 0; 1696 const int i = 0;
1203 { 1697 {
1204 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 1698 PUSH_CLOBJ ("OpenCL::Device", value [i]);
1205 } 1699 }
1206 1700
1207void 1701void
1208reference_count (OpenCL::Queue self) 1702reference_count (OpenCL::Queue self)
1209 PPCODE: 1703 PPCODE:
1296 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0)); 1790 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0));
1297 EXTEND (SP, 1); 1791 EXTEND (SP, 1);
1298 const int i = 0; 1792 const int i = 0;
1299 { 1793 {
1300 NEED_SUCCESS (RetainContext, (value [i])); 1794 NEED_SUCCESS (RetainContext, (value [i]));
1301 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 1795 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1302 } 1796 }
1303 1797
1304void 1798void
1305associated_memobject (OpenCL::Memory self) 1799associated_memobject (OpenCL::Memory self)
1306 PPCODE: 1800 PPCODE:
1308 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0)); 1802 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0));
1309 EXTEND (SP, 1); 1803 EXTEND (SP, 1);
1310 const int i = 0; 1804 const int i = 0;
1311 { 1805 {
1312 NEED_SUCCESS (RetainMemObject, (value [i])); 1806 NEED_SUCCESS (RetainMemObject, (value [i]));
1313 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Memory", value [i])); 1807 PUSH_CLOBJ ("OpenCL::Memory", value [i]);
1314 } 1808 }
1315 1809
1316#END:mem 1810#END:mem
1317 1811
1318#if cl_apple_gl_sharing || cl_khr_gl_sharing 1812#if cl_apple_gl_sharing || cl_khr_gl_sharing
1338 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!"); 1832 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!");
1339 1833
1340 cl_buffer_region crdata = { origin, size }; 1834 cl_buffer_region crdata = { origin, size };
1341 1835
1342 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res)); 1836 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res));
1343 XPUSH_NEW_OBJ ("OpenCL::Buffer", mem); 1837 XPUSH_CLOBJ ("OpenCL::Buffer", mem);
1344 1838
1345MODULE = OpenCL PACKAGE = OpenCL::Image 1839MODULE = OpenCL PACKAGE = OpenCL::Image
1346 1840
1347void 1841void
1348image_info (OpenCL::Image self, cl_image_info name) 1842image_info (OpenCL::Image self, cl_image_info name)
1349 PPCODE: 1843 PPCODE:
1350 INFO (Image) 1844 INFO (Image)
1845
1846void
1847format (OpenCL::Image self)
1848 PPCODE:
1849 cl_image_format format;
1850 NEED_SUCCESS (GetImageInfo, (self, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
1851 EXTEND (SP, 2);
1852 PUSHs (sv_2mortal (newSVuv (format.image_channel_order)));
1853 PUSHs (sv_2mortal (newSVuv (format.image_channel_data_type)));
1351 1854
1352#BEGIN:image 1855#BEGIN:image
1353 1856
1354void 1857void
1355element_size (OpenCL::Image self) 1858element_size (OpenCL::Image self)
1425 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0)); 1928 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0));
1426 EXTEND (SP, 1); 1929 EXTEND (SP, 1);
1427 const int i = 0; 1930 const int i = 0;
1428 { 1931 {
1429 NEED_SUCCESS (RetainContext, (value [i])); 1932 NEED_SUCCESS (RetainContext, (value [i]));
1430 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 1933 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1431 } 1934 }
1432 1935
1433void 1936void
1434normalized_coords (OpenCL::Sampler self) 1937normalized_coords (OpenCL::Sampler self)
1435 PPCODE: 1938 PPCODE:
1465DESTROY (OpenCL::Program self) 1968DESTROY (OpenCL::Program self)
1466 CODE: 1969 CODE:
1467 clReleaseProgram (self); 1970 clReleaseProgram (self);
1468 1971
1469void 1972void
1470build (OpenCL::Program self, OpenCL::Device device, SV *options = &PL_sv_undef) 1973build (OpenCL::Program self, SV *devices = &PL_sv_undef, SV *options = &PL_sv_undef, SV *notify = &PL_sv_undef)
1974 ALIAS:
1975 build_async = 1
1471 CODE: 1976 CODE:
1977 void (CL_CALLBACK *pfn_notify)(cl_program program, void *user_data) = 0;
1978 void *user_data = 0;
1979 cl_uint num_devices = 0;
1980 cl_device_id *device_list = 0;
1981
1982 if (SvOK (devices))
1983 {
1984 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV)
1985 croak ("clProgramBuild: devices must be undef or an array of OpenCL::Device objects.");
1986
1987 AV *av = (AV *)SvRV (devices);
1988 num_devices = av_len (av) + 1;
1989
1990 if (num_devices)
1991 {
1992 device_list = tmpbuf (sizeof (*device_list) * num_devices);
1993 int count;
1994 for (count = 0; count < num_devices; ++count)
1995 device_list [count] = SvCLOBJ ("clBuildProgram", "devices", *av_fetch (av, count, 1), "OpenCL::Device");
1996 }
1997 }
1998
1999 if (SvOK (notify))
2000 {
2001 NEED_SUCCESS (RetainProgram, (self));
2002 pfn_notify = eq_program_notify;
2003 user_data = SvREFCNT_inc (s_get_cv (notify));
2004 }
2005
2006 if (ix)
2007 build_program_async (self, num_devices, device_list, SvPVbyte_nolen (options), user_data);
2008 else
1472 NEED_SUCCESS (BuildProgram, (self, 1, &device, SvPVbyte_nolen (options), 0, 0)); 2009 NEED_SUCCESS (BuildProgram, (self, num_devices, device_list, SvPVbyte_nolen (options), pfn_notify, user_data));
1473 2010
1474void 2011void
1475build_info (OpenCL::Program self, OpenCL::Device device, cl_program_build_info name) 2012build_info (OpenCL::Program self, OpenCL::Device device, cl_program_build_info name)
1476 PPCODE: 2013 PPCODE:
1477 size_t size; 2014 size_t size;
1512 2049
1513void 2050void
1514kernel (OpenCL::Program program, SV *function) 2051kernel (OpenCL::Program program, SV *function)
1515 PPCODE: 2052 PPCODE:
1516 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res)); 2053 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res));
1517 XPUSH_NEW_OBJ ("OpenCL::Kernel", kernel); 2054 XPUSH_CLOBJ ("OpenCL::Kernel", kernel);
2055
2056void
2057kernels_in_program (OpenCL::Program program)
2058 PPCODE:
2059 cl_uint num_kernels;
2060 NEED_SUCCESS (CreateKernelsInProgram, (program, 0, 0, &num_kernels));
2061 cl_kernel *kernels = tmpbuf (sizeof (cl_kernel) * num_kernels);
2062 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0));
2063
2064 int i;
2065 EXTEND (SP, num_kernels);
2066 for (i = 0; i < num_kernels; ++i)
2067 PUSH_CLOBJ ("OpenCL::Kernel", kernels [i]);
1518 2068
1519void 2069void
1520info (OpenCL::Program self, cl_program_info name) 2070info (OpenCL::Program self, cl_program_info name)
1521 PPCODE: 2071 PPCODE:
1522 INFO (Program) 2072 INFO (Program)
1540 { 2090 {
1541 SV *sv = sv_2mortal (newSV (sizes [i])); 2091 SV *sv = sv_2mortal (newSV (sizes [i]));
1542 SvUPGRADE (sv, SVt_PV); 2092 SvUPGRADE (sv, SVt_PV);
1543 SvPOK_only (sv); 2093 SvPOK_only (sv);
1544 SvCUR_set (sv, sizes [i]); 2094 SvCUR_set (sv, sizes [i]);
1545 ptrs [i] = SvPVX (sv); 2095 ptrs [i] = (void *)SvPVX (sv);
1546 PUSHs (sv); 2096 PUSHs (sv);
1547 } 2097 }
1548 2098
1549 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARIES , sizeof (*ptrs ) * n, ptrs , &size)); 2099 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARIES , sizeof (*ptrs ) * n, ptrs , &size));
1550 if (size != sizeof (*ptrs) * n) XSRETURN_EMPTY; 2100 if (size != sizeof (*ptrs) * n) XSRETURN_EMPTY;
1570 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0)); 2120 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0));
1571 EXTEND (SP, 1); 2121 EXTEND (SP, 1);
1572 const int i = 0; 2122 const int i = 0;
1573 { 2123 {
1574 NEED_SUCCESS (RetainContext, (value [i])); 2124 NEED_SUCCESS (RetainContext, (value [i]));
1575 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2125 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1576 } 2126 }
1577 2127
1578void 2128void
1579devices (OpenCL::Program self) 2129devices (OpenCL::Program self)
1580 PPCODE: 2130 PPCODE:
1584 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0)); 2134 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0));
1585 int i, n = size / sizeof (*value); 2135 int i, n = size / sizeof (*value);
1586 EXTEND (SP, n); 2136 EXTEND (SP, n);
1587 for (i = 0; i < n; ++i) 2137 for (i = 0; i < n; ++i)
1588 { 2138 {
1589 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 2139 PUSH_CLOBJ ("OpenCL::Device", value [i]);
1590 } 2140 }
1591 2141
1592void 2142void
1593source (OpenCL::Program self) 2143source (OpenCL::Program self)
1594 PPCODE: 2144 PPCODE:
1748 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0)); 2298 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0));
1749 EXTEND (SP, 1); 2299 EXTEND (SP, 1);
1750 const int i = 0; 2300 const int i = 0;
1751 { 2301 {
1752 NEED_SUCCESS (RetainContext, (value [i])); 2302 NEED_SUCCESS (RetainContext, (value [i]));
1753 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2303 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1754 } 2304 }
1755 2305
1756void 2306void
1757program (OpenCL::Kernel self) 2307program (OpenCL::Kernel self)
1758 PPCODE: 2308 PPCODE:
1760 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0)); 2310 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0));
1761 EXTEND (SP, 1); 2311 EXTEND (SP, 1);
1762 const int i = 0; 2312 const int i = 0;
1763 { 2313 {
1764 NEED_SUCCESS (RetainProgram, (value [i])); 2314 NEED_SUCCESS (RetainProgram, (value [i]));
1765 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Program", value [i])); 2315 PUSH_CLOBJ ("OpenCL::Program", value [i]);
1766 } 2316 }
1767 2317
1768#END:kernel 2318#END:kernel
1769 2319
1770void 2320void
1830wait (OpenCL::Event self) 2380wait (OpenCL::Event self)
1831 CODE: 2381 CODE:
1832 clWaitForEvents (1, &self); 2382 clWaitForEvents (1, &self);
1833 2383
1834void 2384void
2385cb (OpenCL::Event self, cl_int command_exec_callback_type, SV *cb)
2386 CODE:
2387 clSetEventCallback (self, command_exec_callback_type, eq_event_notify, SvREFCNT_inc (s_get_cv (cb)));
2388
2389void
1835info (OpenCL::Event self, cl_event_info name) 2390info (OpenCL::Event self, cl_event_info name)
1836 PPCODE: 2391 PPCODE:
1837 INFO (Event) 2392 INFO (Event)
1838 2393
1839#BEGIN:event 2394#BEGIN:event
1845 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0)); 2400 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0));
1846 EXTEND (SP, 1); 2401 EXTEND (SP, 1);
1847 const int i = 0; 2402 const int i = 0;
1848 { 2403 {
1849 NEED_SUCCESS (RetainCommandQueue, (value [i])); 2404 NEED_SUCCESS (RetainCommandQueue, (value [i]));
1850 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Queue", value [i])); 2405 PUSH_CLOBJ ("OpenCL::Queue", value [i]);
1851 } 2406 }
1852 2407
1853void 2408void
1854command_type (OpenCL::Event self) 2409command_type (OpenCL::Event self)
1855 PPCODE: 2410 PPCODE:
1878 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0)); 2433 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0));
1879 EXTEND (SP, 1); 2434 EXTEND (SP, 1);
1880 const int i = 0; 2435 const int i = 0;
1881 { 2436 {
1882 NEED_SUCCESS (RetainContext, (value [i])); 2437 NEED_SUCCESS (RetainContext, (value [i]));
1883 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2438 PUSH_CLOBJ ("OpenCL::Context", value [i]);
1884 } 2439 }
1885 2440
1886#END:event 2441#END:event
1887 2442
1888void 2443void

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines