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

Comparing OpenCL/OpenCL.xs (file contents):
Revision 1.17 by root, Tue Nov 22 10:39:47 2011 UTC vs.
Revision 1.73 by root, Sat May 5 13:30:07 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#include "ecb.h"//D
6
7#define X_STACKSIZE sizeof (void *) * 512 * 1024 // 2-4mb should be enough, really
8#include "xthread.h"
9#include "schmorp.h"
10
11#ifdef I_DLFCN
12 #include <dlfcn.h>
13#endif
14
15// how stupid is that, the 1.2 header files define CL_VERSION_1_1,
16// but then fail to define the api functions unless you ALSO define
17// this. This breaks 100% of the opencl 1.1 apps, for what reason?
18// after all, the functions are deprecated, not removed.
19// in addition, you cannot test for this in any future-proof way.
20// each time a new opencl version comes out, you need to make a new
21// release.
22#define CL_USE_DEPRECATED_OPENCL_1_2_APIS /* just guessing, you stupid idiots */
23
24#ifndef PREFER_1_1
25 #define PREFER_1_1 1
26#endif
27
28#if PREFER_1_1
29 #define CL_USE_DEPRECATED_OPENCL_1_1_APIS
30#endif
31
32#ifdef __APPLE__
33 #include <OpenCL/opencl.h>
34#else
5#include <CL/opencl.h> 35 #include <CL/opencl.h>
36#endif
37
38#ifndef CL_VERSION_1_2
39 #undef PREFER_1_1
40 #define PREFER_1_1 1
41#endif
6 42
7typedef cl_platform_id OpenCL__Platform; 43typedef cl_platform_id OpenCL__Platform;
8typedef cl_device_id OpenCL__Device; 44typedef cl_device_id OpenCL__Device;
45typedef cl_device_id OpenCL__SubDevice;
9typedef cl_context OpenCL__Context; 46typedef cl_context OpenCL__Context;
10typedef cl_command_queue OpenCL__Queue; 47typedef cl_command_queue OpenCL__Queue;
11typedef cl_mem OpenCL__Memory; 48typedef cl_mem OpenCL__Memory;
12typedef cl_mem OpenCL__Buffer; 49typedef cl_mem OpenCL__Buffer;
50typedef cl_mem OpenCL__BufferObj;
13typedef cl_mem OpenCL__Image; 51typedef cl_mem OpenCL__Image;
14typedef cl_mem OpenCL__Image2D;
15typedef cl_mem OpenCL__Image3D;
16typedef cl_mem OpenCL__Memory_ornull; 52typedef cl_mem OpenCL__Memory_ornull;
17typedef cl_mem OpenCL__Buffer_ornull; 53typedef cl_mem OpenCL__Buffer_ornull;
18typedef cl_mem OpenCL__Image_ornull; 54typedef cl_mem OpenCL__Image_ornull;
19typedef cl_mem OpenCL__Image2D_ornull;
20typedef cl_mem OpenCL__Image3D_ornull;
21typedef cl_sampler OpenCL__Sampler; 55typedef cl_sampler OpenCL__Sampler;
22typedef cl_program OpenCL__Program; 56typedef cl_program OpenCL__Program;
23typedef cl_kernel OpenCL__Kernel; 57typedef cl_kernel OpenCL__Kernel;
24typedef cl_event OpenCL__Event; 58typedef cl_event OpenCL__Event;
25typedef cl_event OpenCL__UserEvent; 59typedef cl_event OpenCL__UserEvent;
26 60
27typedef SV *FUTURE; 61typedef struct mapped * OpenCL__Mapped;
62
63static HV
64 *stash_platform,
65 *stash_device,
66 *stash_subdevice,
67 *stash_context,
68 *stash_queue,
69 *stash_program,
70 *stash_kernel,
71 *stash_sampler,
72 *stash_event,
73 *stash_userevent,
74 *stash_memory,
75 *stash_buffer,
76 *stash_bufferobj,
77 *stash_image,
78 *stash_image1d,
79 *stash_image1darray,
80 *stash_image1dbuffer,
81 *stash_image2d,
82 *stash_image2darray,
83 *stash_image3d,
84 *stash_mapped,
85 *stash_mappedbuffer,
86 *stash_mappedimage;
28 87
29/*****************************************************************************/ 88/*****************************************************************************/
30 89
90// name must include a leading underscore
91// all of this horrors would be unneceesary if somebody wrote a proper OpenGL module
92// for perl. doh.
93static void *
94glsym (const char *name)
95{
96 void *fun = 0;
97
98 #if defined I_DLFCN && defined RTLD_DEFAULT
99 fun = dlsym (RTLD_DEFAULT, name + 1);
100 if (!fun) fun = dlsym (RTLD_DEFAULT, name);
101
102 if (!fun)
103 {
104 static void *libgl;
105 static const char *glso[] = {
106 "libGL.so.1",
107 "libGL.so.3",
108 "libGL.so.4.0",
109 "libGL.so",
110 "/usr/lib/libGL.so",
111 "/usr/X11R6/lib/libGL.1.dylib"
112 };
113 int i;
114
115 for (i = 0; !libgl && i < sizeof (glso) / sizeof (glso [0]); ++i)
116 {
117 libgl = dlopen (glso [i], RTLD_LAZY);
118 if (libgl)
119 break;
120 }
121
122 if (libgl)
123 {
124 fun = dlsym (libgl, name + 1);
125 if (!fun) fun = dlsym (libgl, name);
126 }
127 }
128 #endif
129
130 return fun;
131}
132
133/*****************************************************************************/
134
31/* up to two temporary buffers */ 135/* up to two temporary buffers */
32static void * 136static void * ecb_noinline
33tmpbuf (size_t size) 137tmpbuf (size_t size)
34{ 138{
139 enum { buffers = 4 };
35 static int idx; 140 static int idx;
36 static void *buf [2]; 141 static void *buf [buffers];
37 static size_t len [2]; 142 static size_t len [buffers];
38 143
39 idx ^= 1; 144 idx = (idx + 1) % buffers;
40 145
41 if (len [idx] < size) 146 if (len [idx] < size)
42 { 147 {
43 free (buf [idx]); 148 free (buf [idx]);
44 len [idx] = ((size + 31) & ~4095) + 4096 - 32; 149 len [idx] = ((size + 31) & ~4095) + 4096 - 32;
45 buf [idx] = malloc (len [idx]); 150 buf [idx] = malloc (len [idx]);
46 } 151 }
47 152
48 return buf [idx]; 153 return buf [idx];
154}
155
156static const char * ecb_noinline
157cv_get_name (CV *cv)
158{
159 static char fullname [256];
160
161 GV *gv = CvGV (cv); // gv better be nonzero
162
163 HV *stash = GvSTASH (gv);
164 const char *hvname = HvNAME_get (stash); // stash also better be nonzero
165 const char *gvname = GvNAME (gv);
166
167 snprintf (fullname, sizeof (fullname), "%s::%s", hvname, gvname);
168 return fullname;
49} 169}
50 170
51/*****************************************************************************/ 171/*****************************************************************************/
52 172
53typedef struct 173typedef struct
112 if (res) \ 232 if (res) \
113 FAIL (name); 233 FAIL (name);
114 234
115/*****************************************************************************/ 235/*****************************************************************************/
116 236
117#define NEW_MORTAL_OBJ(class,ptr) sv_setref_pv (sv_newmortal (), class, ptr)
118#define XPUSH_NEW_OBJ(class,ptr) XPUSHs (NEW_MORTAL_OBJ (class, ptr))
119
120static void * 237static SV *
238new_clobj (HV *stash, IV id)
239{
240 return sv_2mortal (sv_bless (newRV_noinc (newSViv (id)), stash));
241}
242
243#define PUSH_CLOBJ(stash,id) PUSHs (new_clobj ((stash), (IV)(id)))
244#define XPUSH_CLOBJ(stash,id) XPUSHs (new_clobj ((stash), (IV)(id)))
245
246/* cl objects are either \$iv, or [$iv, ...] */
247/* they can be upgraded at runtime to the array form */
248static void * ecb_noinline
121SvPTROBJ (const char *func, const char *svname, SV *sv, const char *pkg) 249SvCLOBJ (CV *cv, const char *svname, SV *sv, const char *pkg)
122{ 250{
251 // sv_derived_from is quite slow :(
123 if (SvROK (sv) && sv_derived_from (sv, pkg)) 252 if (SvROK (sv) && sv_derived_from (sv, pkg))
124 return (void *)SvIV (SvRV (sv)); 253 return (void *)SvIV (SvRV (sv));
125 254
126 croak ("%s: %s is not of type %s", func, svname, pkg); 255 croak ("%s: %s is not of type %s", cv_get_name (cv), svname, pkg);
256}
257
258// the "no-inherit" version of the above
259static void * ecb_noinline
260SvCLOBJ_ni (CV *cv, const char *svname, SV *sv, HV *stash)
261{
262 if (SvROK (sv) && SvSTASH (SvRV (sv)) == stash)
263 return (void *)SvIV (SvRV (sv));
264
265 croak ("%s: %s is not of type %s", cv_get_name (cv), svname, HvNAME (stash));
127} 266}
128 267
129/*****************************************************************************/ 268/*****************************************************************************/
269
270static cl_context_properties * ecb_noinline
271SvCONTEXTPROPERTIES (CV *cv, const char *svname, SV *sv, cl_context_properties *extra, int extracount)
272{
273 if (!sv || !SvOK (sv))
274 if (extra)
275 sv = sv_2mortal (newRV_noinc ((SV *)newAV ())); // slow, but rarely used hopefully
276 else
277 return 0;
278
279 if (SvROK (sv) && SvTYPE (SvRV (sv)) == SVt_PVAV)
280 {
281 AV *av = (AV *)SvRV (sv);
282 int i, len = av_len (av) + 1;
283 cl_context_properties *p = tmpbuf (sizeof (cl_context_properties) * (len + extracount + 1));
284 cl_context_properties *l = p;
285
286 if (len & 1)
287 croak ("%s: %s is not a property list (must contain an even number of elements)", cv_get_name (cv), svname);
288
289 while (extracount--)
290 *l++ = *extra++;
291
292 for (i = 0; i < len; i += 2)
293 {
294 cl_context_properties t = SvIV (*av_fetch (av, i , 0));
295 SV *p_sv = *av_fetch (av, i + 1, 0);
296 cl_context_properties v = SvIV (p_sv); // code below can override
297
298 switch (t)
299 {
300 case CL_CONTEXT_PLATFORM:
301 if (SvROK (p_sv))
302 v = (cl_context_properties)SvCLOBJ (cv, svname, p_sv, "OpenCL::Platform");
303 break;
304
305 case CL_GLX_DISPLAY_KHR:
306 if (!SvOK (p_sv))
307 {
308 void *func = glsym ("_glXGetCurrentDisplay");
309 if (func)
310 v = (cl_context_properties)((void *(*)(void))func)();
311 }
312 break;
313
314 case CL_GL_CONTEXT_KHR:
315 if (!SvOK (p_sv))
316 {
317 void *func = glsym ("_glXGetCurrentContext");
318 if (func)
319 v = (cl_context_properties)((void *(*)(void))func)();
320 }
321 break;
322
323 default:
324 /* unknown property, treat as int */
325 break;
326 }
327
328 *l++ = t;
329 *l++ = v;
330 }
331
332 *l = 0;
333
334 return p;
335 }
336
337 croak ("%s: %s is not a property list (either undef or [type => value, ...])", cv_get_name (cv), svname);
338}
339
340// parse an array of CLOBJ into a void ** array in C - works only for CLOBJs whose representation
341// is a pointer (and only on well-behaved systems).
342static void * ecb_noinline
343object_list (CV *cv, int or_undef, const char *argname, SV *arg, const char *klass, cl_uint *rcount)
344{
345 if (!SvROK (arg) || SvTYPE (SvRV (arg)) != SVt_PVAV)
346 croak ("%s: '%s' parameter must be %sa reference to an array of %s objects",
347 cv_get_name (cv), argname, or_undef ? "undef or " : "", klass);
348
349 AV *av = (AV *)SvRV (arg);
350 void **list = 0;
351 cl_uint count = av_len (av) + 1;
352
353 if (count)
354 {
355 list = tmpbuf (sizeof (*list) * count);
356 int i;
357 for (i = 0; i < count; ++i)
358 list [i] = SvCLOBJ (cv, argname, *av_fetch (av, i, 1), klass);
359 }
360
361 if (!count && !or_undef)
362 croak ("%s: '%s' must contain at least one %s object",
363 cv_get_name (cv), argname, klass);
364
365 *rcount = count;
366 return (void *)list;
367}
368
369/*****************************************************************************/
370/* callback stuff */
371
372/* default context callback, log to stderr */
373static void CL_CALLBACK
374context_default_notify (const char *msg, const void *info, size_t cb, void *data)
375{
376 fprintf (stderr, "OpenCL Context Notify: %s\n", msg);
377}
378
379typedef struct
380{
381 int free_cb;
382 void (*push)(void *data1, void *data2, void *data3);
383} eq_vtbl;
384
385typedef struct eq_item
386{
387 struct eq_item *next;
388 eq_vtbl *vtbl;
389 SV *cb;
390 void *data1, *data2, *data3;
391} eq_item;
392
393static void (*eq_signal_func)(void *signal_arg, int value);
394static void *eq_signal_arg;
395static xmutex_t eq_lock = X_MUTEX_INIT;
396static eq_item *eq_head, *eq_tail;
397
398static void ecb_noinline
399eq_enq (eq_vtbl *vtbl, SV *cb, void *data1, void *data2, void *data3)
400{
401 eq_item *item = malloc (sizeof (eq_item));
402
403 item->next = 0;
404 item->vtbl = vtbl;
405 item->cb = cb;
406 item->data1 = data1;
407 item->data2 = data2;
408 item->data3 = data3;
409
410 X_LOCK (eq_lock);
411
412 *(eq_head ? &eq_tail->next : &eq_head) = item;
413 eq_tail = item;
414
415 X_UNLOCK (eq_lock);
416
417 eq_signal_func (eq_signal_arg, 0);
418}
419
420static eq_item *
421eq_dec (void)
422{
423 eq_item *res;
424
425 X_LOCK (eq_lock);
426
427 res = eq_head;
428 if (res)
429 eq_head = res->next;
430
431 X_UNLOCK (eq_lock);
432
433 return res;
434}
435
436static void
437eq_poll (void)
438{
439 eq_item *item;
440
441 while ((item = eq_dec ()))
442 {
443 ENTER;
444 SAVETMPS;
445
446 dSP;
447 PUSHMARK (SP);
448 EXTEND (SP, 2);
449
450 if (item->vtbl->free_cb)
451 sv_2mortal (item->cb);
452
453 PUTBACK;
454 item->vtbl->push (item->data1, item->data2, item->data3);
455
456 SV *cb = item->cb;
457 free (item);
458
459 call_sv (cb, G_DISCARD | G_VOID);
460
461 FREETMPS;
462 LEAVE;
463 }
464}
465
466static void
467eq_poll_interrupt (pTHX_ void *c_arg, int value)
468{
469 eq_poll ();
470}
471
472/*****************************************************************************/
473/* context notify */
474
475static void ecb_noinline
476eq_context_push (void *data1, void *data2, void *data3)
477{
478 dSP;
479 PUSHs (sv_2mortal (newSVpv (data1, 0)));
480 PUSHs (sv_2mortal (newSVpvn (data2, (STRLEN)data3)));
481 PUTBACK;
482
483 free (data1);
484 free (data2);
485}
486
487static eq_vtbl eq_context_vtbl = { 0, eq_context_push };
488
489static void CL_CALLBACK
490eq_context_notify (const char *msg, const void *pvt, size_t cb, void *user_data)
491{
492 void *pvt_copy = malloc (cb);
493 memcpy (pvt_copy, pvt, cb);
494 eq_enq (&eq_context_vtbl, user_data, strdup (msg), pvt_copy, (void *)cb);
495}
496
497#define CONTEXT_NOTIFY_CALLBACK \
498 void (CL_CALLBACK *pfn_notify)(const char *, const void *, size_t, void *) = context_default_notify; \
499 void *user_data = 0; \
500 \
501 if (SvOK (notify)) \
502 { \
503 pfn_notify = eq_context_notify; \
504 user_data = s_get_cv (notify); \
505 }
506
507static SV * ecb_noinline
508new_clobj_context (cl_context ctx, void *user_data)
509{
510 SV *sv = new_clobj (stash_context, (IV)ctx);
511
512 if (user_data)
513 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0);
514
515 return sv;
516}
517
518#define XPUSH_CLOBJ_CONTEXT XPUSHs (new_clobj_context (ctx, user_data));
519
520/*****************************************************************************/
521/* build/compile/link notify */
522
523static void
524eq_program_push (void *data1, void *data2, void *data3)
525{
526 dSP;
527 PUSH_CLOBJ (stash_program, data1);
528 PUTBACK;
529}
530
531static eq_vtbl eq_program_vtbl = { 1, eq_program_push };
532
533static void CL_CALLBACK
534eq_program_notify (cl_program program, void *user_data)
535{
536 clRetainProgram (program);
537
538 eq_enq (&eq_program_vtbl, user_data, (void *)program, 0, 0);
539}
540
541typedef void (CL_CALLBACK *program_callback)(cl_program program, void *user_data);
542
543static program_callback ecb_noinline
544make_program_callback (SV *notify, void **ruser_data)
545{
546 if (SvOK (notify))
547 {
548 *ruser_data = SvREFCNT_inc (s_get_cv (notify));
549 return eq_program_notify;
550 }
551 else
552 {
553 *ruser_data = 0;
554 return 0;
555 }
556}
557
558struct build_args
559{
560 cl_program program;
561 char *options;
562 void *user_data;
563 cl_uint num_devices;
564};
565
566X_THREAD_PROC (build_program_thread)
567{
568 struct build_args *arg = thr_arg;
569
570 clBuildProgram (arg->program, arg->num_devices, arg->num_devices ? (void *)(arg + 1) : 0, arg->options, 0, 0);
571
572 if (arg->user_data)
573 eq_program_notify (arg->program, arg->user_data);
574 else
575 clReleaseProgram (arg->program);
576
577 free (arg->options);
578 free (arg);
579
580 return 0;
581}
582
583static void
584build_program_async (cl_program program, cl_uint num_devices, const cl_device_id *device_list, const char *options, void *user_data)
585{
586 struct build_args *arg = malloc (sizeof (struct build_args) + sizeof (*device_list) * num_devices);
587
588 arg->program = program;
589 arg->options = strdup (options);
590 arg->user_data = user_data;
591 arg->num_devices = num_devices;
592 memcpy (arg + 1, device_list, sizeof (*device_list) * num_devices);
593
594 xthread_t id;
595 thread_create (&id, build_program_thread, arg);
596}
597
598/*****************************************************************************/
599/* event objects */
600
601static void
602eq_event_push (void *data1, void *data2, void *data3)
603{
604 dSP;
605 PUSH_CLOBJ (stash_event, data1);
606 PUSHs (sv_2mortal (newSViv ((IV)data2)));
607 PUTBACK;
608}
609
610static eq_vtbl eq_event_vtbl = { 1, eq_event_push };
611
612static void CL_CALLBACK
613eq_event_notify (cl_event event, cl_int event_command_exec_status, void *user_data)
614{
615 clRetainEvent (event);
616 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0);
617}
618
619/*****************************************************************************/
620/* utilities for XS code */
130 621
131static size_t 622static size_t
132img_row_pitch (cl_mem img) 623img_row_pitch (cl_mem img)
133{ 624{
134 size_t res; 625 size_t res;
135 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0); 626 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0);
136 return res; 627 return res;
137} 628}
138 629
139static cl_event * 630static cl_event * ecb_noinline
140event_list (SV **items, int count) 631event_list (CV *cv, SV **items, cl_uint *rcount, cl_event extra)
141{ 632{
633 cl_uint count = *rcount;
634
635 if (count > 0x7fffffffU) // yeah, it's a hack - the caller might have underflowed
636 *rcount = count = 0;
637
638 if (!count && !extra)
639 return 0;
640
142 cl_event *list = tmpbuf (sizeof (cl_event) * count); 641 cl_event *list = tmpbuf (sizeof (cl_event) * (count + 1));
642 int i = 0;
143 643
144 while (count--) 644 while (count--)
645 if (SvOK (items [count]))
145 list [count] = SvPTROBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event"); 646 list [i++] = SvCLOBJ (cv, "wait_events", items [count], "OpenCL::Event");
146 647
648 if (extra)
649 list [i++] = extra;
650
651 *rcount = i;
652
147 return list; 653 return i ? list : 0;
148} 654}
149 655
150#define EVENT_LIST(items,count) \ 656#define EVENT_LIST(skip) \
151 cl_uint event_list_count = (count); \ 657 cl_uint event_list_count = items - (skip); \
152 cl_event *event_list_ptr = event_list (&ST (items), event_list_count) 658 cl_event *event_list_ptr = event_list (cv, &ST (skip), &event_list_count, 0)
153 659
154#define INFO(class) \ 660#define INFO(class) \
155{ \ 661{ \
156 size_t size; \ 662 size_t size; \
157 NEED_SUCCESS (Get ## class ## Info, (this, name, 0, 0, &size)); \ 663 NEED_SUCCESS (Get ## class ## Info, (self, name, 0, 0, &size)); \
158 SV *sv = sv_2mortal (newSV (size)); \ 664 SV *sv = sv_2mortal (newSV (size)); \
159 SvUPGRADE (sv, SVt_PV); \ 665 SvUPGRADE (sv, SVt_PV); \
160 SvPOK_only (sv); \ 666 SvPOK_only (sv); \
161 SvCUR_set (sv, size); \ 667 SvCUR_set (sv, size); \
162 NEED_SUCCESS (Get ## class ## Info, (this, name, size, SvPVX (sv), 0)); \ 668 NEED_SUCCESS (Get ## class ## Info, (self, name, size, SvPVX (sv), 0)); \
163 XPUSHs (sv); \ 669 XPUSHs (sv); \
164} 670}
165 671
672/*****************************************************************************/
673/* mapped_xxx */
674
675static OpenCL__Mapped
676SvMAPPED (SV *sv)
677{
678 // no typechecking atm., keep your fingers crossed
679 return (OpenCL__Mapped)SvMAGIC (SvRV (sv))->mg_ptr;
680}
681
682struct mapped
683{
684 cl_command_queue queue;
685 cl_mem memobj;
686 void *ptr;
687 size_t cb;
688 cl_event event;
689 size_t row_pitch;
690 size_t slice_pitch;
691};
692
693static SV *
694mapped_new (HV *stash, cl_command_queue queue, cl_mem memobj, cl_map_flags flags, void *ptr, size_t cb, cl_event ev, size_t row_pitch, size_t slice_pitch)
695{
696 SV *data = newSV (0);
697 SvUPGRADE (data, SVt_PVMG);
698
699 OpenCL__Mapped mapped;
700 New (0, mapped, 1, struct mapped);
701
702 clRetainCommandQueue (queue);
703
704 mapped->queue = queue;
705 mapped->memobj = memobj;
706 mapped->ptr = ptr;
707 mapped->cb = cb;
708 mapped->event = ev;
709 mapped->row_pitch = row_pitch;
710 mapped->slice_pitch = slice_pitch;
711
712 sv_magicext (data, 0, PERL_MAGIC_ext, 0, (char *)mapped, 0);
713
714 if (SvLEN (data))
715 Safefree (data);
716
717 SvPVX (data) = (char *)ptr;
718 SvCUR_set (data, cb);
719 SvLEN_set (data, 0);
720 SvPOK_only (data);
721
722 SV *obj = sv_2mortal (sv_bless (newRV_noinc (data), stash));
723
724 if (!(flags & CL_MAP_WRITE))
725 SvREADONLY_on (data);
726
727 return obj;
728}
729
730static void
731mapped_detach (SV *sv, OpenCL__Mapped mapped)
732{
733 SV *data = SvRV (sv);
734
735 // the next check checks both before AND after detach, where SvPVX should STILL be 0
736 if (SvPVX (data) != (char *)mapped->ptr)
737 warn ("FATAL: OpenCL memory mapped scalar changed location, detected");
738 else
739 {
740 SvREADONLY_off (data);
741 SvCUR_set (data, 0);
742 SvPVX (data) = 0;
743 SvOK_off (data);
744 }
745
746 mapped->ptr = 0;
747}
748
749static void
750mapped_unmap (CV *cv, SV *self, OpenCL__Mapped mapped, cl_command_queue queue, SV **wait_list, cl_uint event_list_count)
751{
752 cl_event *event_list_ptr = event_list (cv, wait_list, &event_list_count, mapped->event);
753 cl_event ev;
754
755 NEED_SUCCESS (EnqueueUnmapMemObject, (queue, mapped->memobj, mapped->ptr, event_list_count, event_list_ptr, &ev));
756
757 clReleaseEvent (mapped->event);
758 mapped->event = ev;
759
760 mapped_detach (self, mapped);
761}
762
763/*****************************************************************************/
764
166MODULE = OpenCL PACKAGE = OpenCL 765MODULE = OpenCL PACKAGE = OpenCL
167 766
168PROTOTYPES: ENABLE 767PROTOTYPES: ENABLE
169 768
769void
770poll ()
771 CODE:
772 eq_poll ();
773
774void
775_eq_initialise (IV func, IV arg)
776 CODE:
777 eq_signal_func = (void (*)(void *, int))func;
778 eq_signal_arg = (void*)arg;
779
170BOOT: 780BOOT:
171{ 781{
172 HV *stash = gv_stashpv ("OpenCL", 1); 782 HV *stash = gv_stashpv ("OpenCL", 1);
173 static const ivstr *civ, const_iv[] = { 783 static const ivstr *civ, const_iv[] = {
174 { sizeof (cl_char ), "SIZEOF_CHAR" }, 784 { sizeof (cl_char ), "SIZEOF_CHAR" },
175 { sizeof (cl_uchar ), "SIZEOF_UCHAR" }, 785 { sizeof (cl_uchar ), "SIZEOF_UCHAR" },
176 { sizeof (cl_short ), "SIZEOF_SHORT" }, 786 { sizeof (cl_short ), "SIZEOF_SHORT" },
177 { sizeof (cl_ushort), "SIZEOF_USHORT" }, 787 { sizeof (cl_ushort), "SIZEOF_USHORT" },
178 { sizeof (cl_int ), "SIZEOF_INT" }, 788 { sizeof (cl_int ), "SIZEOF_INT" },
179 { sizeof (cl_uint ), "SIZEOF_UINT" }, 789 { sizeof (cl_uint ), "SIZEOF_UINT" },
180 { sizeof (cl_long ), "SIZEOF_LONG" }, 790 { sizeof (cl_long ), "SIZEOF_LONG" },
181 { sizeof (cl_ulong ), "SIZEOF_ULONG" }, 791 { sizeof (cl_ulong ), "SIZEOF_ULONG" },
182 { sizeof (cl_half ), "SIZEOF_HALF" }, 792 { sizeof (cl_half ), "SIZEOF_HALF" },
183 { sizeof (cl_float ), "SIZEOF_FLOAT" }, 793 { sizeof (cl_float ), "SIZEOF_FLOAT" },
184 { sizeof (cl_double), "SIZEOF_DOUBLE" }, 794 { sizeof (cl_double), "SIZEOF_DOUBLE" },
795 { PREFER_1_1 , "PREFER_1_1" },
185#include "constiv.h" 796#include "constiv.h"
186 }; 797 };
798
187 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--) 799 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--)
188 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv)); 800 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv));
801
802 stash_platform = gv_stashpv ("OpenCL::Platform", GV_ADD);
803 stash_device = gv_stashpv ("OpenCL::Device", GV_ADD);
804 stash_subdevice = gv_stashpv ("OpenCL::SubDevice", GV_ADD);
805 stash_context = gv_stashpv ("OpenCL::Context", GV_ADD);
806 stash_queue = gv_stashpv ("OpenCL::Queue", GV_ADD);
807 stash_program = gv_stashpv ("OpenCL::Program", GV_ADD);
808 stash_kernel = gv_stashpv ("OpenCL::Kernel", GV_ADD);
809 stash_sampler = gv_stashpv ("OpenCL::Sampler", GV_ADD);
810 stash_event = gv_stashpv ("OpenCL::Event", GV_ADD);
811 stash_userevent = gv_stashpv ("OpenCL::UserEvent", GV_ADD);
812 stash_memory = gv_stashpv ("OpenCL::Memory", GV_ADD);
813 stash_buffer = gv_stashpv ("OpenCL::Buffer", GV_ADD);
814 stash_bufferobj = gv_stashpv ("OpenCL::BufferObj", GV_ADD);
815 stash_image = gv_stashpv ("OpenCL::Image", GV_ADD);
816 stash_image1d = gv_stashpv ("OpenCL::Image1D", GV_ADD);
817 stash_image1darray = gv_stashpv ("OpenCL::Image1DArray", GV_ADD);
818 stash_image1dbuffer = gv_stashpv ("OpenCL::Image1DBuffer", GV_ADD);
819 stash_image2d = gv_stashpv ("OpenCL::Image2D", GV_ADD);
820 stash_image2darray = gv_stashpv ("OpenCL::Image2DArray", GV_ADD);
821 stash_image3d = gv_stashpv ("OpenCL::Image3D", GV_ADD);
822 stash_mapped = gv_stashpv ("OpenCL::Mapped", GV_ADD);
823 stash_mappedbuffer = gv_stashpv ("OpenCL::MappedBuffer", GV_ADD);
824 stash_mappedimage = gv_stashpv ("OpenCL::MappedImage", GV_ADD);
825
826 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt);
189} 827}
190 828
191cl_int 829cl_int
192errno () 830errno ()
193 CODE: 831 CODE:
194 errno = res; 832 RETVAL = res;
833 OUTPUT:
834 RETVAL
195 835
196const char * 836const char *
197err2str (cl_int err) 837err2str (cl_int err = res)
198 838
199const char * 839const char *
200enum2str (cl_uint value) 840enum2str (cl_uint value)
201 841
202void 842void
210 list = tmpbuf (sizeof (*list) * count); 850 list = tmpbuf (sizeof (*list) * count);
211 NEED_SUCCESS (GetPlatformIDs, (count, list, 0)); 851 NEED_SUCCESS (GetPlatformIDs, (count, list, 0));
212 852
213 EXTEND (SP, count); 853 EXTEND (SP, count);
214 for (i = 0; i < count; ++i) 854 for (i = 0; i < count; ++i)
215 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Platform", list [i])); 855 PUSH_CLOBJ (stash_platform, list [i]);
216 856
217void 857void
218context_from_type (FUTURE properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, FUTURE notify = 0) 858context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
219 PPCODE: 859 PPCODE:
860 CONTEXT_NOTIFY_CALLBACK;
220 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (0, type, 0, 0, &res)); 861 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, pfn_notify, user_data, &res));
221 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 862 XPUSH_CLOBJ_CONTEXT;
222 863
223void 864void
224context (FUTURE properties, FUTURE devices, FUTURE notify = 0) 865context (cl_context_properties *properties, SV *devices, SV *notify = &PL_sv_undef)
225 PPCODE: 866 PPCODE:
226 /* der Gipfel der Kunst */ 867 cl_uint device_count;
868 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
869
870 CONTEXT_NOTIFY_CALLBACK;
871 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (properties, device_count, device_list, pfn_notify, user_data, &res));
872 XPUSH_CLOBJ_CONTEXT;
227 873
228void 874void
229wait_for_events (...) 875wait_for_events (...)
230 CODE: 876 CODE:
231 EVENT_LIST (0, items); 877 EVENT_LIST (0);
232 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr)); 878 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr));
233 879
234PROTOTYPES: DISABLE 880PROTOTYPES: DISABLE
235 881
236MODULE = OpenCL PACKAGE = OpenCL::Platform 882MODULE = OpenCL PACKAGE = OpenCL::Platform
237 883
238void 884void
239info (OpenCL::Platform this, cl_platform_info name) 885info (OpenCL::Platform self, cl_platform_info name)
240 PPCODE: 886 PPCODE:
241 INFO (Platform) 887 INFO (Platform)
242 888
889void
890unload_compiler (OpenCL::Platform self)
891 CODE:
892#if CL_VERSION_1_2
893 clUnloadPlatformCompiler (self);
894#endif
895
243#BEGIN:platform 896#BEGIN:platform
244 897
245void 898void
246profile (OpenCL::Platform this) 899profile (OpenCL::Platform self)
247 ALIAS: 900 ALIAS:
248 profile = CL_PLATFORM_PROFILE 901 profile = CL_PLATFORM_PROFILE
249 version = CL_PLATFORM_VERSION 902 version = CL_PLATFORM_VERSION
250 name = CL_PLATFORM_NAME 903 name = CL_PLATFORM_NAME
251 vendor = CL_PLATFORM_VENDOR 904 vendor = CL_PLATFORM_VENDOR
252 extensions = CL_PLATFORM_EXTENSIONS 905 extensions = CL_PLATFORM_EXTENSIONS
253 PPCODE: 906 PPCODE:
254 size_t size; 907 size_t size;
255 NEED_SUCCESS (GetPlatformInfo, (this, ix, 0, 0, &size)); 908 NEED_SUCCESS (GetPlatformInfo, (self, ix, 0, 0, &size));
256 char *value = tmpbuf (size); 909 char *value = tmpbuf (size);
257 NEED_SUCCESS (GetPlatformInfo, (this, ix, size, value, 0)); 910 NEED_SUCCESS (GetPlatformInfo, (self, ix, size, value, 0));
258 EXTEND (SP, 1); 911 EXTEND (SP, 1);
259 const int i = 0; 912 const int i = 0;
260 PUSHs (sv_2mortal (newSVpv (value, 0))); 913 PUSHs (sv_2mortal (newSVpv (value, 0)));
261 914
262#END:platform 915#END:platform
263 916
264void 917void
265devices (OpenCL::Platform this, cl_device_type type = CL_DEVICE_TYPE_ALL) 918devices (OpenCL::Platform self, cl_device_type type = CL_DEVICE_TYPE_ALL)
266 PPCODE: 919 PPCODE:
267 cl_device_id *list; 920 cl_device_id *list;
268 cl_uint count; 921 cl_uint count;
269 int i; 922 int i;
270 923
271 NEED_SUCCESS (GetDeviceIDs, (this, type, 0, 0, &count)); 924 NEED_SUCCESS (GetDeviceIDs, (self, type, 0, 0, &count));
272 list = tmpbuf (sizeof (*list) * count); 925 list = tmpbuf (sizeof (*list) * count);
273 NEED_SUCCESS (GetDeviceIDs, (this, type, count, list, 0)); 926 NEED_SUCCESS (GetDeviceIDs, (self, type, count, list, 0));
274 927
275 EXTEND (SP, count); 928 EXTEND (SP, count);
276 for (i = 0; i < count; ++i) 929 for (i = 0; i < count; ++i)
277 PUSHs (sv_setref_pv (sv_newmortal (), "OpenCL::Device", list [i])); 930 PUSH_CLOBJ (stash_device, list [i]);
278 931
279void 932void
280context (OpenCL::Platform this, FUTURE properties, SV *devices, FUTURE notify = 0) 933context (OpenCL::Platform self, SV *properties, SV *devices, SV *notify = &PL_sv_undef)
281 PPCODE: 934 PPCODE:
935 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
936 cl_context_properties *props = SvCONTEXTPROPERTIES (cv, "properties", properties, extra, 2);
937
938 cl_uint device_count;
939 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
940
941 CONTEXT_NOTIFY_CALLBACK;
942 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (props, device_count, device_list, pfn_notify, user_data, &res));
943 XPUSH_CLOBJ_CONTEXT;
944
945void
946context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
947 PPCODE:
948 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
949 cl_context_properties *props = SvCONTEXTPROPERTIES (cv, "properties", properties, extra, 2);
950
951 CONTEXT_NOTIFY_CALLBACK;
952 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, pfn_notify, user_data, &res));
953 XPUSH_CLOBJ_CONTEXT;
954
955MODULE = OpenCL PACKAGE = OpenCL::Device
956
957void
958info (OpenCL::Device self, cl_device_info name)
959 PPCODE:
960 INFO (Device)
961
962#if CL_VERSION_1_2
963
964void
965sub_devices (OpenCL::Device self, SV *properties)
966 PPCODE:
282 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV) 967 if (!SvROK (properties) || SvTYPE (SvRV (properties)) != SVt_PVAV)
283 croak ("OpenCL::Platform argument 'device' must be an arrayref with device objects, in call"); 968 croak ("OpenCL::Device::sub_devices: properties must be specified as reference to an array of property-value pairs");
284 969
285 AV *av = (AV *)SvRV (devices); 970 properties = SvRV (properties);
286 cl_uint num_devices = av_len (av) + 1; 971
287 cl_device_id *device_list = tmpbuf (sizeof (cl_device_id) * num_devices); 972 cl_uint count = av_len ((AV *)properties) + 1;
973 cl_device_partition_property *props = tmpbuf (sizeof (*props) * count + 1);
974
288 int i; 975 int i;
976 for (i = 0; i < count; ++i)
977 props [i] = (cl_device_partition_property)SvIV (*av_fetch ((AV *)properties, i, 0));
289 978
290 for (i = num_devices; i--; ) 979 props [count] = 0;
291 device_list [i] = SvPTROBJ ("clCreateContext", "devices", *av_fetch (av, i, 0), "OpenCL::Device");
292 980
293 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (0, num_devices, device_list, 0, 0, &res)); 981 cl_uint num_devices;
294 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 982 NEED_SUCCESS (CreateSubDevices, (self, props, 0, 0, &num_devices));
983 cl_device_id *list = tmpbuf (sizeof (*list) * num_devices);
984 NEED_SUCCESS (CreateSubDevices, (self, props, num_devices, list, 0));
295 985
296void 986 EXTEND (SP, num_devices);
297context_from_type (OpenCL::Platform this, FUTURE properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, FUTURE notify = 0) 987 for (i = 0; i < count; ++i)
298 PPCODE: 988 PUSH_CLOBJ (stash_subdevice, list [i]);
299 cl_context_properties props[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)this, 0 };
300 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, 0, 0, &res));
301 XPUSH_NEW_OBJ ("OpenCL::Context", ctx);
302 989
303MODULE = OpenCL PACKAGE = OpenCL::Device 990#endif
304
305void
306info (OpenCL::Device this, cl_device_info name)
307 PPCODE:
308 INFO (Device)
309 991
310#BEGIN:device 992#BEGIN:device
311 993
312void 994void
313type (OpenCL::Device this) 995type (OpenCL::Device self)
314 PPCODE: 996 PPCODE:
315 cl_device_type value [1]; 997 cl_device_type value [1];
316 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_TYPE, sizeof (value), value, 0)); 998 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_TYPE, sizeof (value), value, 0));
317 EXTEND (SP, 1); 999 EXTEND (SP, 1);
318 const int i = 0; 1000 const int i = 0;
319 PUSHs (sv_2mortal (newSViv (value [i]))); 1001 PUSHs (sv_2mortal (newSVuv (value [i])));
320 1002
321void 1003void
322vendor_id (OpenCL::Device this) 1004vendor_id (OpenCL::Device self)
323 ALIAS: 1005 ALIAS:
324 vendor_id = CL_DEVICE_VENDOR_ID 1006 vendor_id = CL_DEVICE_VENDOR_ID
325 max_compute_units = CL_DEVICE_MAX_COMPUTE_UNITS 1007 max_compute_units = CL_DEVICE_MAX_COMPUTE_UNITS
326 max_work_item_dimensions = CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS 1008 max_work_item_dimensions = CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS
327 preferred_vector_width_char = CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR 1009 preferred_vector_width_char = CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR
345 native_vector_width_int = CL_DEVICE_NATIVE_VECTOR_WIDTH_INT 1027 native_vector_width_int = CL_DEVICE_NATIVE_VECTOR_WIDTH_INT
346 native_vector_width_long = CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG 1028 native_vector_width_long = CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG
347 native_vector_width_float = CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT 1029 native_vector_width_float = CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT
348 native_vector_width_double = CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE 1030 native_vector_width_double = CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE
349 native_vector_width_half = CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF 1031 native_vector_width_half = CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF
350 reference_count_ext = CL_DEVICE_REFERENCE_COUNT_EXT 1032 reference_count_ext = CL_DEVICE_REFERENCE_COUNT_EXT
351 PPCODE: 1033 PPCODE:
352 cl_uint value [1]; 1034 cl_uint value [1];
353 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1035 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
354 EXTEND (SP, 1); 1036 EXTEND (SP, 1);
355 const int i = 0; 1037 const int i = 0;
356 PUSHs (sv_2mortal (newSVuv (value [i]))); 1038 PUSHs (sv_2mortal (newSVuv (value [i])));
357 1039
358void 1040void
359max_work_group_size (OpenCL::Device this) 1041max_work_group_size (OpenCL::Device self)
360 ALIAS: 1042 ALIAS:
361 max_work_group_size = CL_DEVICE_MAX_WORK_GROUP_SIZE 1043 max_work_group_size = CL_DEVICE_MAX_WORK_GROUP_SIZE
362 image2d_max_width = CL_DEVICE_IMAGE2D_MAX_WIDTH 1044 image2d_max_width = CL_DEVICE_IMAGE2D_MAX_WIDTH
363 image2d_max_height = CL_DEVICE_IMAGE2D_MAX_HEIGHT 1045 image2d_max_height = CL_DEVICE_IMAGE2D_MAX_HEIGHT
364 image3d_max_width = CL_DEVICE_IMAGE3D_MAX_WIDTH 1046 image3d_max_width = CL_DEVICE_IMAGE3D_MAX_WIDTH
366 image3d_max_depth = CL_DEVICE_IMAGE3D_MAX_DEPTH 1048 image3d_max_depth = CL_DEVICE_IMAGE3D_MAX_DEPTH
367 max_parameter_size = CL_DEVICE_MAX_PARAMETER_SIZE 1049 max_parameter_size = CL_DEVICE_MAX_PARAMETER_SIZE
368 profiling_timer_resolution = CL_DEVICE_PROFILING_TIMER_RESOLUTION 1050 profiling_timer_resolution = CL_DEVICE_PROFILING_TIMER_RESOLUTION
369 PPCODE: 1051 PPCODE:
370 size_t value [1]; 1052 size_t value [1];
371 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1053 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
372 EXTEND (SP, 1); 1054 EXTEND (SP, 1);
373 const int i = 0; 1055 const int i = 0;
374 PUSHs (sv_2mortal (newSVuv (value [i]))); 1056 PUSHs (sv_2mortal (newSVuv (value [i])));
375 1057
376void 1058void
377max_work_item_sizes (OpenCL::Device this) 1059max_work_item_sizes (OpenCL::Device self)
378 PPCODE: 1060 PPCODE:
379 size_t size; 1061 size_t size;
380 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_MAX_WORK_ITEM_SIZES, 0, 0, &size)); 1062 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_MAX_WORK_ITEM_SIZES, 0, 0, &size));
381 size_t *value = tmpbuf (size); 1063 size_t *value = tmpbuf (size);
382 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_MAX_WORK_ITEM_SIZES, size, value, 0)); 1064 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_MAX_WORK_ITEM_SIZES, size, value, 0));
383 int i, n = size / sizeof (*value); 1065 int i, n = size / sizeof (*value);
384 EXTEND (SP, n); 1066 EXTEND (SP, n);
385 for (i = 0; i < n; ++i) 1067 for (i = 0; i < n; ++i)
386 PUSHs (sv_2mortal (newSVuv (value [i]))); 1068 PUSHs (sv_2mortal (newSVuv (value [i])));
387 1069
388void 1070void
389address_bits (OpenCL::Device this) 1071address_bits (OpenCL::Device self)
390 PPCODE: 1072 PPCODE:
391 cl_bitfield value [1]; 1073 cl_bitfield value [1];
392 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_ADDRESS_BITS, sizeof (value), value, 0)); 1074 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_ADDRESS_BITS, sizeof (value), value, 0));
393 EXTEND (SP, 1); 1075 EXTEND (SP, 1);
394 const int i = 0; 1076 const int i = 0;
395 PUSHs (sv_2mortal (newSVuv (value [i]))); 1077 PUSHs (sv_2mortal (newSVuv (value [i])));
396 1078
397void 1079void
398max_mem_alloc_size (OpenCL::Device this) 1080max_mem_alloc_size (OpenCL::Device self)
399 ALIAS: 1081 ALIAS:
400 max_mem_alloc_size = CL_DEVICE_MAX_MEM_ALLOC_SIZE 1082 max_mem_alloc_size = CL_DEVICE_MAX_MEM_ALLOC_SIZE
401 global_mem_cache_size = CL_DEVICE_GLOBAL_MEM_CACHE_SIZE 1083 global_mem_cache_size = CL_DEVICE_GLOBAL_MEM_CACHE_SIZE
402 global_mem_size = CL_DEVICE_GLOBAL_MEM_SIZE 1084 global_mem_size = CL_DEVICE_GLOBAL_MEM_SIZE
403 max_constant_buffer_size = CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE 1085 max_constant_buffer_size = CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE
404 local_mem_size = CL_DEVICE_LOCAL_MEM_SIZE 1086 local_mem_size = CL_DEVICE_LOCAL_MEM_SIZE
405 PPCODE: 1087 PPCODE:
406 cl_ulong value [1]; 1088 cl_ulong value [1];
407 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1089 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
408 EXTEND (SP, 1); 1090 EXTEND (SP, 1);
409 const int i = 0; 1091 const int i = 0;
410 PUSHs (sv_2mortal (newSVuv (value [i]))); 1092 PUSHs (sv_2mortal (newSVuv (value [i])));
411 1093
412void 1094void
413single_fp_config (OpenCL::Device this) 1095single_fp_config (OpenCL::Device self)
414 ALIAS: 1096 ALIAS:
415 single_fp_config = CL_DEVICE_SINGLE_FP_CONFIG 1097 single_fp_config = CL_DEVICE_SINGLE_FP_CONFIG
416 double_fp_config = CL_DEVICE_DOUBLE_FP_CONFIG 1098 double_fp_config = CL_DEVICE_DOUBLE_FP_CONFIG
417 half_fp_config = CL_DEVICE_HALF_FP_CONFIG 1099 half_fp_config = CL_DEVICE_HALF_FP_CONFIG
418 PPCODE: 1100 PPCODE:
419 cl_device_fp_config value [1]; 1101 cl_device_fp_config value [1];
420 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1102 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
421 EXTEND (SP, 1); 1103 EXTEND (SP, 1);
422 const int i = 0; 1104 const int i = 0;
423 PUSHs (sv_2mortal (newSVuv (value [i]))); 1105 PUSHs (sv_2mortal (newSVuv (value [i])));
424 1106
425void 1107void
426global_mem_cache_type (OpenCL::Device this) 1108global_mem_cache_type (OpenCL::Device self)
427 PPCODE: 1109 PPCODE:
428 cl_device_mem_cache_type value [1]; 1110 cl_device_mem_cache_type value [1];
429 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, sizeof (value), value, 0)); 1111 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, sizeof (value), value, 0));
430 EXTEND (SP, 1); 1112 EXTEND (SP, 1);
431 const int i = 0; 1113 const int i = 0;
432 PUSHs (sv_2mortal (newSVuv (value [i]))); 1114 PUSHs (sv_2mortal (newSVuv (value [i])));
433 1115
434void 1116void
435local_mem_type (OpenCL::Device this) 1117local_mem_type (OpenCL::Device self)
436 PPCODE: 1118 PPCODE:
437 cl_device_local_mem_type value [1]; 1119 cl_device_local_mem_type value [1];
438 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_LOCAL_MEM_TYPE, sizeof (value), value, 0)); 1120 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_LOCAL_MEM_TYPE, sizeof (value), value, 0));
439 EXTEND (SP, 1); 1121 EXTEND (SP, 1);
440 const int i = 0; 1122 const int i = 0;
441 PUSHs (sv_2mortal (newSVuv (value [i]))); 1123 PUSHs (sv_2mortal (newSVuv (value [i])));
442 1124
443void 1125void
444error_correction_support (OpenCL::Device this) 1126error_correction_support (OpenCL::Device self)
445 ALIAS: 1127 ALIAS:
446 error_correction_support = CL_DEVICE_ERROR_CORRECTION_SUPPORT 1128 error_correction_support = CL_DEVICE_ERROR_CORRECTION_SUPPORT
447 endian_little = CL_DEVICE_ENDIAN_LITTLE 1129 endian_little = CL_DEVICE_ENDIAN_LITTLE
448 available = CL_DEVICE_AVAILABLE 1130 available = CL_DEVICE_AVAILABLE
449 compiler_available = CL_DEVICE_COMPILER_AVAILABLE 1131 compiler_available = CL_DEVICE_COMPILER_AVAILABLE
450 host_unified_memory = CL_DEVICE_HOST_UNIFIED_MEMORY 1132 host_unified_memory = CL_DEVICE_HOST_UNIFIED_MEMORY
451 PPCODE: 1133 PPCODE:
452 cl_bool value [1]; 1134 cl_bool value [1];
453 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1135 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
454 EXTEND (SP, 1); 1136 EXTEND (SP, 1);
455 const int i = 0; 1137 const int i = 0;
456 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no)); 1138 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no));
457 1139
458void 1140void
459execution_capabilities (OpenCL::Device this) 1141execution_capabilities (OpenCL::Device self)
460 PPCODE: 1142 PPCODE:
461 cl_device_exec_capabilities value [1]; 1143 cl_device_exec_capabilities value [1];
462 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_EXECUTION_CAPABILITIES, sizeof (value), value, 0)); 1144 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_EXECUTION_CAPABILITIES, sizeof (value), value, 0));
463 EXTEND (SP, 1); 1145 EXTEND (SP, 1);
464 const int i = 0; 1146 const int i = 0;
465 PUSHs (sv_2mortal (newSVuv (value [i]))); 1147 PUSHs (sv_2mortal (newSVuv (value [i])));
466 1148
467void 1149void
468properties (OpenCL::Device this) 1150properties (OpenCL::Device self)
469 PPCODE: 1151 PPCODE:
470 cl_command_queue_properties value [1]; 1152 cl_command_queue_properties value [1];
471 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0)); 1153 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0));
472 EXTEND (SP, 1); 1154 EXTEND (SP, 1);
473 const int i = 0; 1155 const int i = 0;
474 PUSHs (sv_2mortal (newSViv (value [i]))); 1156 PUSHs (sv_2mortal (newSVuv (value [i])));
475 1157
476void 1158void
477platform (OpenCL::Device this) 1159platform (OpenCL::Device self)
478 PPCODE: 1160 PPCODE:
479 cl_platform_id value [1]; 1161 cl_platform_id value [1];
480 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_PLATFORM, sizeof (value), value, 0)); 1162 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0));
481 EXTEND (SP, 1); 1163 EXTEND (SP, 1);
482 const int i = 0; 1164 const int i = 0;
483 { 1165 {
484 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Platform", value [i])); 1166 PUSH_CLOBJ (stash_platform, value [i]);
485 } 1167 }
486 1168
487void 1169void
488name (OpenCL::Device this) 1170name (OpenCL::Device self)
489 ALIAS: 1171 ALIAS:
490 name = CL_DEVICE_NAME 1172 name = CL_DEVICE_NAME
491 vendor = CL_DEVICE_VENDOR 1173 vendor = CL_DEVICE_VENDOR
492 driver_version = CL_DRIVER_VERSION 1174 driver_version = CL_DRIVER_VERSION
493 profile = CL_DEVICE_PROFILE 1175 profile = CL_DEVICE_PROFILE
494 version = CL_DEVICE_VERSION 1176 version = CL_DEVICE_VERSION
495 extensions = CL_DEVICE_EXTENSIONS 1177 extensions = CL_DEVICE_EXTENSIONS
496 PPCODE: 1178 PPCODE:
497 size_t size; 1179 size_t size;
498 NEED_SUCCESS (GetDeviceInfo, (this, ix, 0, 0, &size)); 1180 NEED_SUCCESS (GetDeviceInfo, (self, ix, 0, 0, &size));
499 char *value = tmpbuf (size); 1181 char *value = tmpbuf (size);
500 NEED_SUCCESS (GetDeviceInfo, (this, ix, size, value, 0)); 1182 NEED_SUCCESS (GetDeviceInfo, (self, ix, size, value, 0));
501 EXTEND (SP, 1); 1183 EXTEND (SP, 1);
502 const int i = 0; 1184 const int i = 0;
503 PUSHs (sv_2mortal (newSVpv (value, 0))); 1185 PUSHs (sv_2mortal (newSVpv (value, 0)));
504 1186
505void 1187void
506parent_device_ext (OpenCL::Device this) 1188parent_device_ext (OpenCL::Device self)
507 PPCODE: 1189 PPCODE:
508 cl_device_id value [1]; 1190 cl_device_id value [1];
509 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0)); 1191 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0));
510 EXTEND (SP, 1); 1192 EXTEND (SP, 1);
511 const int i = 0; 1193 const int i = 0;
512 { 1194 {
513 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 1195 PUSH_CLOBJ (stash_device, value [i]);
514 } 1196 }
515 1197
516void 1198void
517partition_types_ext (OpenCL::Device this) 1199partition_types_ext (OpenCL::Device self)
518 ALIAS: 1200 ALIAS:
519 partition_types_ext = CL_DEVICE_PARTITION_TYPES_EXT 1201 partition_types_ext = CL_DEVICE_PARTITION_TYPES_EXT
520 affinity_domains_ext = CL_DEVICE_AFFINITY_DOMAINS_EXT 1202 affinity_domains_ext = CL_DEVICE_AFFINITY_DOMAINS_EXT
521 partition_style_ext = CL_DEVICE_PARTITION_STYLE_EXT 1203 partition_style_ext = CL_DEVICE_PARTITION_STYLE_EXT
522 PPCODE: 1204 PPCODE:
523 size_t size; 1205 size_t size;
524 NEED_SUCCESS (GetDeviceInfo, (this, ix, 0, 0, &size)); 1206 NEED_SUCCESS (GetDeviceInfo, (self, ix, 0, 0, &size));
525 cl_device_partition_property_ext *value = tmpbuf (size); 1207 cl_device_partition_property_ext *value = tmpbuf (size);
526 NEED_SUCCESS (GetDeviceInfo, (this, ix, size, value, 0)); 1208 NEED_SUCCESS (GetDeviceInfo, (self, ix, size, value, 0));
527 int i, n = size / sizeof (*value); 1209 int i, n = size / sizeof (*value);
528 EXTEND (SP, n); 1210 EXTEND (SP, n);
529 for (i = 0; i < n; ++i) 1211 for (i = 0; i < n; ++i)
530 PUSHs (sv_2mortal (newSVuv (value [i]))); 1212 PUSHs (sv_2mortal (newSVuv (value [i])));
531 1213
532#END:device 1214#END:device
533 1215
1216MODULE = OpenCL PACKAGE = OpenCL::SubDevice
1217
1218#if CL_VERSION_1_2
1219
1220void
1221DESTROY (OpenCL::SubDevice self)
1222 CODE:
1223 clReleaseDevice (self);
1224
1225#endif
1226
534MODULE = OpenCL PACKAGE = OpenCL::Context 1227MODULE = OpenCL PACKAGE = OpenCL::Context
535 1228
536void 1229void
537DESTROY (OpenCL::Context context) 1230DESTROY (OpenCL::Context self)
538 CODE: 1231 CODE:
539 clReleaseContext (context); 1232 clReleaseContext (self);
540 1233
541void 1234void
542info (OpenCL::Context this, cl_context_info name) 1235info (OpenCL::Context self, cl_context_info name)
543 PPCODE: 1236 PPCODE:
544 INFO (Context) 1237 INFO (Context)
545 1238
546void 1239void
547queue (OpenCL::Context this, OpenCL::Device device, cl_command_queue_properties properties = 0) 1240queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0)
548 PPCODE: 1241 PPCODE:
549 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (this, device, properties, &res)); 1242 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res));
550 XPUSH_NEW_OBJ ("OpenCL::Queue", queue); 1243 XPUSH_CLOBJ (stash_queue, queue);
551 1244
552void 1245void
553user_event (OpenCL::Context this) 1246user_event (OpenCL::Context self)
554 PPCODE: 1247 PPCODE:
555 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (this, &res)); 1248 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res));
556 XPUSH_NEW_OBJ ("OpenCL::UserEvent", ev); 1249 XPUSH_CLOBJ (stash_userevent, ev);
557 1250
558void 1251void
559buffer (OpenCL::Context this, cl_mem_flags flags, size_t len) 1252buffer (OpenCL::Context self, cl_mem_flags flags, size_t len)
560 PPCODE: 1253 PPCODE:
561 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) 1254 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))
562 croak ("clCreateBuffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?"); 1255 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?");
563 1256
564 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (this, flags, len, 0, &res)); 1257 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res));
565 XPUSH_NEW_OBJ ("OpenCL::Buffer", mem); 1258 XPUSH_CLOBJ (stash_bufferobj, mem);
566 1259
567void 1260void
568buffer_sv (OpenCL::Context this, cl_mem_flags flags, SV *data) 1261buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data)
569 PPCODE: 1262 PPCODE:
570 STRLEN len; 1263 STRLEN len;
571 char *ptr = SvPVbyte (data, len); 1264 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
572
573 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) 1265 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)))
574 croak ("clCreateBuffer: have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?"); 1266 croak ("OpenCL::Context::buffer_sv: you have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?");
575
576 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (this, flags, len, ptr, &res)); 1267 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res));
577 XPUSH_NEW_OBJ ("OpenCL::Buffer", mem); 1268 XPUSH_CLOBJ (stash_bufferobj, mem);
578 1269
1270#if CL_VERSION_1_2
1271
579void 1272void
580image2d (OpenCL::Context this, 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) 1273image (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)
581 PPCODE: 1274 PPCODE:
582 STRLEN len; 1275 STRLEN len;
583 char *ptr = SvPVbyte (data, len); 1276 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
584 const cl_image_format format = { channel_order, channel_type }; 1277 const cl_image_format format = { channel_order, channel_type };
1278 const cl_image_desc desc = {
1279 type,
1280 width, height, depth,
1281 array_size, row_pitch, slice_pitch,
1282 num_mip_level, num_samples,
1283 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ (cv, "data", data, "OpenCL::Buffer") : 0
1284 };
585 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (this, flags, &format, width, height, row_pitch, ptr, &res)); 1285 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
586 XPUSH_NEW_OBJ ("OpenCL::Image2D", mem); 1286 HV *stash = stash_image;
1287 switch (type)
1288 {
1289 case CL_MEM_OBJECT_IMAGE1D_BUFFER: stash = stash_image1dbuffer; break;
1290 case CL_MEM_OBJECT_IMAGE1D: stash = stash_image1d; break;
1291 case CL_MEM_OBJECT_IMAGE1D_ARRAY: stash = stash_image2darray; break;
1292 case CL_MEM_OBJECT_IMAGE2D: stash = stash_image2d; break;
1293 case CL_MEM_OBJECT_IMAGE2D_ARRAY: stash = stash_image2darray; break;
1294 case CL_MEM_OBJECT_IMAGE3D: stash = stash_image3d; break;
1295 }
1296 XPUSH_CLOBJ (stash, mem);
587 1297
1298#endif
1299
588void 1300void
589image3d (OpenCL::Context this, 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) 1301image2d (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)
590 PPCODE: 1302 PPCODE:
591 STRLEN len; 1303 STRLEN len;
592 char *ptr = SvPVbyte (data, len); 1304 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
593 const cl_image_format format = { channel_order, channel_type }; 1305 const cl_image_format format = { channel_order, channel_type };
1306#if PREFER_1_1
1307 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res));
1308#else
1309 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, row_pitch, 0, 0, 0, 0 };
1310 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1311#endif
1312 XPUSH_CLOBJ (stash_image2d, mem);
1313
1314void
1315image3d (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)
1316 PPCODE:
1317 STRLEN len;
1318 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
1319 const cl_image_format format = { channel_order, channel_type };
1320#if PREFER_1_1
594 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (this, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res)); 1321 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res));
595 XPUSH_NEW_OBJ ("OpenCL::Image3D", mem); 1322#else
1323 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, row_pitch, slice_pitch, 0, 0, 0 };
1324 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1325#endif
1326 XPUSH_CLOBJ (stash_image3d, mem);
596 1327
1328#if cl_apple_gl_sharing || cl_khr_gl_sharing
1329
597void 1330void
1331gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj)
1332 PPCODE:
1333 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res));
1334 XPUSH_CLOBJ (stash_bufferobj, mem);
1335
1336void
1337gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer)
1338 PPCODE:
1339 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res));
1340 XPUSH_CLOBJ (stash_image2d, mem);
1341
1342#if CL_VERSION_1_2
1343
1344void
1345gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1346 ALIAS:
1347 PPCODE:
1348 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture, (self, flags, target, miplevel, texture, &res));
1349 cl_gl_object_type type;
1350 NEED_SUCCESS (GetGLObjectInfo, (mem, &type, 0)); // TODO: use target instead?
1351 HV *stash = stash_memory;
1352 switch (type)
1353 {
1354 case CL_GL_OBJECT_TEXTURE_BUFFER: stash = stash_image1dbuffer; break;
1355 case CL_GL_OBJECT_TEXTURE1D: stash = stash_image1d; break;
1356 case CL_GL_OBJECT_TEXTURE1D_ARRAY: stash = stash_image2darray; break;
1357 case CL_GL_OBJECT_TEXTURE2D: stash = stash_image2d; break;
1358 case CL_GL_OBJECT_TEXTURE2D_ARRAY: stash = stash_image2darray; break;
1359 case CL_GL_OBJECT_TEXTURE3D: stash = stash_image3d; break;
1360 }
1361 XPUSH_CLOBJ (stash, mem);
1362
1363#endif
1364
1365void
1366gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1367 PPCODE:
1368#if PREFER_1_1
1369 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res));
1370#else
1371 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1372#endif
1373 XPUSH_CLOBJ (stash_image2d, mem);
1374
1375void
1376gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1377 PPCODE:
1378#if PREFER_1_1
1379 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res));
1380#else
1381 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1382#endif
1383 XPUSH_CLOBJ (stash_image3d, mem);
1384
1385#endif
1386
1387void
598supported_image_formats (OpenCL::Context this, cl_mem_flags flags, cl_mem_object_type image_type) 1388supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type)
599 PPCODE: 1389 PPCODE:
600{ 1390{
601 cl_uint count; 1391 cl_uint count;
602 cl_image_format *list; 1392 cl_image_format *list;
603 int i; 1393 int i;
604 1394
605 NEED_SUCCESS (GetSupportedImageFormats, (this, flags, image_type, 0, 0, &count)); 1395 NEED_SUCCESS (GetSupportedImageFormats, (self, flags, image_type, 0, 0, &count));
606 Newx (list, count, cl_image_format); 1396 Newx (list, count, cl_image_format);
607 NEED_SUCCESS (GetSupportedImageFormats, (this, flags, image_type, count, list, 0)); 1397 NEED_SUCCESS (GetSupportedImageFormats, (self, flags, image_type, count, list, 0));
608 1398
609 EXTEND (SP, count); 1399 EXTEND (SP, count);
610 for (i = 0; i < count; ++i) 1400 for (i = 0; i < count; ++i)
611 { 1401 {
612 AV *av = newAV (); 1402 AV *av = newAV ();
615 PUSHs (sv_2mortal (newRV_noinc ((SV *)av))); 1405 PUSHs (sv_2mortal (newRV_noinc ((SV *)av)));
616 } 1406 }
617} 1407}
618 1408
619void 1409void
620sampler (OpenCL::Context this, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode) 1410sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode)
621 PPCODE: 1411 PPCODE:
622 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (this, normalized_coords, addressing_mode, filter_mode, &res)); 1412 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res));
623 XPUSH_NEW_OBJ ("OpenCL::Sampler", sampler); 1413 XPUSH_CLOBJ (stash_sampler, sampler);
624 1414
625void 1415void
626program_with_source (OpenCL::Context this, SV *program) 1416program_with_source (OpenCL::Context self, SV *program)
627 PPCODE: 1417 PPCODE:
628 STRLEN len; 1418 STRLEN len;
629 size_t len2; 1419 size_t len2;
630 const char *ptr = SvPVbyte (program, len); 1420 const char *ptr = SvPVbyte (program, len);
631 1421
632 len2 = len; 1422 len2 = len;
633 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (this, 1, &ptr, &len2, &res)); 1423 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res));
634 XPUSH_NEW_OBJ ("OpenCL::Program", prog); 1424 XPUSH_CLOBJ (stash_program, prog);
1425
1426void
1427program_with_binary (OpenCL::Context self, SV *devices, SV *binaries)
1428 PPCODE:
1429 cl_uint device_count;
1430 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
1431
1432 if (!SvROK (binaries) || SvTYPE (SvRV (binaries)) != SVt_PVAV)
1433 croak ("OpenCL::Context::program_with_binary: binaries must be specified as reference to an array of strings");
1434
1435 binaries = SvRV (binaries);
1436
1437 if (device_count != av_len ((AV *)binaries) + 1)
1438 croak ("OpenCL::Context::program_with_binary: differing numbers of devices and binaries are not allowed");
1439
1440 size_t *length_list = tmpbuf (sizeof (*length_list) * device_count);
1441 const unsigned char **binary_list = tmpbuf (sizeof (*binary_list) * device_count);
1442 cl_int *status_list = tmpbuf (sizeof (*status_list) * device_count);
1443
1444 int i;
1445 for (i = 0; i < device_count; ++i)
1446 {
1447 STRLEN len;
1448 binary_list [i] = (const unsigned char *)SvPVbyte (*av_fetch ((AV *)binaries, i, 0), len);
1449 length_list [i] = len;
1450 }
1451
1452 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithBinary, (self, device_count, device_list,
1453 length_list, binary_list,
1454 GIMME_V == G_ARRAY ? status_list : 0, &res));
1455
1456 EXTEND (SP, 2);
1457 PUSH_CLOBJ (stash_program, prog);
1458
1459 if (GIMME_V == G_ARRAY)
1460 {
1461 AV *av = newAV ();
1462 PUSHs (sv_2mortal (newRV_noinc ((SV *)av)));
1463
1464 for (i = device_count; i--; )
1465 av_store (av, i, newSViv (status_list [i]));
1466 }
1467
1468#if CL_VERSION_1_2
1469
1470void
1471program_with_built_in_kernels (OpenCL::Context self, SV *devices, SV *kernel_names)
1472 PPCODE:
1473 cl_uint device_count;
1474 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
1475
1476 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithBuiltInKernels, (self, device_count, device_list, SvPVbyte_nolen (kernel_names), &res));
1477
1478 XPUSH_CLOBJ (stash_program, prog);
1479
1480void
1481link_program (OpenCL::Context self, SV *devices, SV *options, SV *programs, SV *notify = &PL_sv_undef)
1482 CODE:
1483 cl_uint device_count = 0;
1484 cl_device_id *device_list = 0;
1485
1486 if (SvOK (devices))
1487 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
1488
1489 cl_uint program_count;
1490 cl_program *program_list = object_list (cv, 0, "programs", programs, "OpenCL::Program", &program_count);
1491
1492 void *user_data;
1493 program_callback pfn_notify = make_program_callback (notify, &user_data);
1494
1495 NEED_SUCCESS_ARG (cl_program prog, LinkProgram, (self, device_count, device_list, SvPVbyte_nolen (options),
1496 program_count, program_list, pfn_notify, user_data, &res));
1497
1498 XPUSH_CLOBJ (stash_program, prog);
1499
1500#endif
635 1501
636#BEGIN:context 1502#BEGIN:context
637 1503
638void 1504void
639reference_count (OpenCL::Context this) 1505reference_count (OpenCL::Context self)
640 ALIAS: 1506 ALIAS:
641 reference_count = CL_CONTEXT_REFERENCE_COUNT 1507 reference_count = CL_CONTEXT_REFERENCE_COUNT
642 num_devices = CL_CONTEXT_NUM_DEVICES 1508 num_devices = CL_CONTEXT_NUM_DEVICES
643 PPCODE: 1509 PPCODE:
644 cl_uint value [1]; 1510 cl_uint value [1];
645 NEED_SUCCESS (GetContextInfo, (this, ix, sizeof (value), value, 0)); 1511 NEED_SUCCESS (GetContextInfo, (self, ix, sizeof (value), value, 0));
646 EXTEND (SP, 1); 1512 EXTEND (SP, 1);
647 const int i = 0; 1513 const int i = 0;
648 PUSHs (sv_2mortal (newSVuv (value [i]))); 1514 PUSHs (sv_2mortal (newSVuv (value [i])));
649 1515
650void 1516void
651devices (OpenCL::Context this) 1517devices (OpenCL::Context self)
652 PPCODE: 1518 PPCODE:
653 size_t size; 1519 size_t size;
654 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_DEVICES, 0, 0, &size)); 1520 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, 0, 0, &size));
655 cl_device_id *value = tmpbuf (size); 1521 cl_device_id *value = tmpbuf (size);
656 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_DEVICES, size, value, 0)); 1522 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0));
657 int i, n = size / sizeof (*value); 1523 int i, n = size / sizeof (*value);
658 EXTEND (SP, n); 1524 EXTEND (SP, n);
659 for (i = 0; i < n; ++i) 1525 for (i = 0; i < n; ++i)
660 { 1526 {
661 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 1527 PUSH_CLOBJ (stash_device, value [i]);
662 } 1528 }
663 1529
664void 1530void
665properties (OpenCL::Context this) 1531properties (OpenCL::Context self)
666 PPCODE: 1532 PPCODE:
667 size_t size; 1533 size_t size;
668 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_PROPERTIES, 0, 0, &size)); 1534 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_PROPERTIES, 0, 0, &size));
669 cl_context_properties *value = tmpbuf (size); 1535 cl_context_properties *value = tmpbuf (size);
670 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_PROPERTIES, size, value, 0)); 1536 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_PROPERTIES, size, value, 0));
671 int i, n = size / sizeof (*value); 1537 int i, n = size / sizeof (*value);
672 EXTEND (SP, n); 1538 EXTEND (SP, n);
673 for (i = 0; i < n; ++i) 1539 for (i = 0; i < n; ++i)
674 PUSHs (sv_2mortal (newSVuv ((UV)value [i]))); 1540 PUSHs (sv_2mortal (newSVuv ((UV)value [i])));
675 1541
676#END:context 1542#END:context
677 1543
678MODULE = OpenCL PACKAGE = OpenCL::Queue 1544MODULE = OpenCL PACKAGE = OpenCL::Queue
679 1545
680void 1546void
681DESTROY (OpenCL::Queue this) 1547DESTROY (OpenCL::Queue self)
682 CODE: 1548 CODE:
683 clReleaseCommandQueue (this); 1549 clReleaseCommandQueue (self);
684 1550
685void 1551void
686enqueue_read_buffer (OpenCL::Queue this, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...) 1552read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...)
1553 ALIAS:
1554 enqueue_read_buffer = 0
687 PPCODE: 1555 PPCODE:
688 cl_event ev = 0;
689 EVENT_LIST (6, items - 6); 1556 EVENT_LIST (6);
690 1557
691 SvUPGRADE (data, SVt_PV); 1558 SvUPGRADE (data, SVt_PV);
692 SvGROW (data, len); 1559 SvGROW (data, len);
693 SvPOK_only (data); 1560 SvPOK_only (data);
694 SvCUR_set (data, len); 1561 SvCUR_set (data, len);
1562
1563 cl_event ev = 0;
695 NEED_SUCCESS (EnqueueReadBuffer, (this, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1564 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
696 1565
697 if (ev) 1566 if (ev)
698 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1567 XPUSH_CLOBJ (stash_event, ev);
699 1568
700void 1569void
701enqueue_write_buffer (OpenCL::Queue this, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...) 1570write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...)
1571 ALIAS:
1572 enqueue_write_buffer = 0
702 PPCODE: 1573 PPCODE:
703 cl_event ev = 0; 1574 EVENT_LIST (5);
1575
704 STRLEN len; 1576 STRLEN len;
705 char *ptr = SvPVbyte (data, len); 1577 char *ptr = SvPVbyte (data, len);
706 EVENT_LIST (5, items - 5);
707 1578
1579 cl_event ev = 0;
708 NEED_SUCCESS (EnqueueReadBuffer, (this, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1580 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
709 1581
710 if (ev) 1582 if (ev)
711 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1583 XPUSH_CLOBJ (stash_event, ev);
712 1584
1585#if CL_VERSION_1_2
1586
713void 1587void
714enqueue_copy_buffer (OpenCL::Queue this, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...) 1588fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...)
1589 ALIAS:
1590 enqueue_fill_buffer = 0
715 PPCODE: 1591 PPCODE:
1592 EVENT_LIST (5);
1593
1594 STRLEN len;
1595 char *ptr = SvPVbyte (data, len);
1596
716 cl_event ev = 0; 1597 cl_event ev = 0;
717 EVENT_LIST (6, items - 6);
718
719 NEED_SUCCESS (EnqueueCopyBuffer, (this, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1598 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
720 1599
721 if (ev) 1600 if (ev)
722 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1601 XPUSH_CLOBJ (stash_event, ev);
723 1602
724void 1603void
725enqueue_read_buffer_rect (OpenCL::Queue this, 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, ...) 1604fill_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, ...)
1605 ALIAS:
1606 enqueue_fill_image = 0
726 PPCODE: 1607 PPCODE:
1608 EVENT_LIST (12);
1609
1610 const size_t origin [3] = { x, y, z };
1611 const size_t region [3] = { width, height, depth };
1612
1613 const cl_float c_f [4] = { r, g, b, a };
1614 const cl_uint c_u [4] = { r, g, b, a };
1615 const cl_int c_s [4] = { r, g, b, a };
1616 const void *c_fus [3] = { &c_f, &c_u, &c_s };
1617 static const char fus [] = { 0, 0, 0, 0, 0, 0, 0, 2, 2, 2, 1, 1, 1, 0, 0 };
1618 cl_image_format format;
1619 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
1620 assert (sizeof (fus) == CL_FLOAT + 1 - CL_SNORM_INT8);
1621 if (format.image_channel_data_type < CL_SNORM_INT8 || CL_FLOAT < format.image_channel_data_type)
1622 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported.");
1623
727 cl_event ev = 0; 1624 cl_event ev = 0;
1625 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type - CL_SNORM_INT8]],
1626 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1627
1628 if (ev)
1629 XPUSH_CLOBJ (stash_event, ev);
1630
1631#endif
1632
1633void
1634copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...)
1635 ALIAS:
1636 enqueue_copy_buffer = 0
1637 PPCODE:
1638 EVENT_LIST (6);
1639
1640 cl_event ev = 0;
1641 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1642
1643 if (ev)
1644 XPUSH_CLOBJ (stash_event, ev);
1645
1646void
1647read_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, ...)
1648 ALIAS:
1649 enqueue_read_buffer_rect = 0
1650 PPCODE:
1651 EVENT_LIST (17);
1652
728 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1653 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
729 const size_t host_origin[3] = { host_x, host_y, host_z }; 1654 const size_t host_origin[3] = { host_x, host_y, host_z };
730 const size_t region[3] = { width, height, depth }; 1655 const size_t region[3] = { width, height, depth };
731 EVENT_LIST (17, items - 17);
732 1656
733 if (!buf_row_pitch) 1657 if (!buf_row_pitch)
734 buf_row_pitch = region [0]; 1658 buf_row_pitch = region [0];
735 1659
736 if (!buf_slice_pitch) 1660 if (!buf_slice_pitch)
746 1670
747 SvUPGRADE (data, SVt_PV); 1671 SvUPGRADE (data, SVt_PV);
748 SvGROW (data, len); 1672 SvGROW (data, len);
749 SvPOK_only (data); 1673 SvPOK_only (data);
750 SvCUR_set (data, len); 1674 SvCUR_set (data, len);
1675
1676 cl_event ev = 0;
751 NEED_SUCCESS (EnqueueReadBufferRect, (this, 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)); 1677 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));
752 1678
753 if (ev) 1679 if (ev)
754 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1680 XPUSH_CLOBJ (stash_event, ev);
755 1681
756void 1682void
757enqueue_write_buffer_rect (OpenCL::Queue this, 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, ...) 1683write_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, ...)
1684 ALIAS:
1685 enqueue_write_buffer_rect = 0
758 PPCODE: 1686 PPCODE:
759 cl_event ev = 0; 1687 EVENT_LIST (17);
1688
760 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1689 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
761 const size_t host_origin[3] = { host_x, host_y, host_z }; 1690 const size_t host_origin[3] = { host_x, host_y, host_z };
762 const size_t region[3] = { width, height, depth }; 1691 const size_t region[3] = { width, height, depth };
763 STRLEN len; 1692 STRLEN len;
764 char *ptr = SvPVbyte (data, len); 1693 char *ptr = SvPVbyte (data, len);
765 EVENT_LIST (17, items - 17);
766 1694
767 if (!buf_row_pitch) 1695 if (!buf_row_pitch)
768 buf_row_pitch = region [0]; 1696 buf_row_pitch = region [0];
769 1697
770 if (!buf_slice_pitch) 1698 if (!buf_slice_pitch)
779 size_t min_len = host_row_pitch * host_slice_pitch * region [2]; 1707 size_t min_len = host_row_pitch * host_slice_pitch * region [2];
780 1708
781 if (len < min_len) 1709 if (len < min_len)
782 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1710 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
783 1711
1712 cl_event ev = 0;
784 NEED_SUCCESS (EnqueueWriteBufferRect, (this, 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)); 1713 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));
785 1714
786 if (ev) 1715 if (ev)
787 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1716 XPUSH_CLOBJ (stash_event, ev);
788 1717
789void 1718void
790enqueue_read_image (OpenCL::Queue this, 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, ...) 1719copy_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, ...)
1720 ALIAS:
1721 enqueue_copy_buffer_rect = 0
791 PPCODE: 1722 PPCODE:
1723 EVENT_LIST (16);
1724
1725 const size_t src_origin[3] = { src_x, src_y, src_z };
1726 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1727 const size_t region[3] = { width, height, depth };
1728
792 cl_event ev = 0; 1729 cl_event ev = 0;
1730 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));
1731
1732 if (ev)
1733 XPUSH_CLOBJ (stash_event, ev);
1734
1735void
1736read_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, ...)
1737 ALIAS:
1738 enqueue_read_image = 0
1739 PPCODE:
1740 EVENT_LIST (12);
1741
793 const size_t src_origin[3] = { src_x, src_y, src_z }; 1742 const size_t src_origin[3] = { src_x, src_y, src_z };
794 const size_t region[3] = { width, height, depth }; 1743 const size_t region[3] = { width, height, depth };
795 EVENT_LIST (12, items - 12);
796 1744
797 if (!row_pitch) 1745 if (!row_pitch)
798 row_pitch = img_row_pitch (src); 1746 row_pitch = img_row_pitch (src);
799 1747
800 if (depth > 1 && !slice_pitch) 1748 if (depth > 1 && !slice_pitch)
804 1752
805 SvUPGRADE (data, SVt_PV); 1753 SvUPGRADE (data, SVt_PV);
806 SvGROW (data, len); 1754 SvGROW (data, len);
807 SvPOK_only (data); 1755 SvPOK_only (data);
808 SvCUR_set (data, len); 1756 SvCUR_set (data, len);
1757
1758 cl_event ev = 0;
809 NEED_SUCCESS (EnqueueReadImage, (this, src, blocking, src_origin, region, row_pitch, slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1759 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));
810 1760
811 if (ev) 1761 if (ev)
812 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1762 XPUSH_CLOBJ (stash_event, ev);
813 1763
814void 1764void
815enqueue_write_image (OpenCL::Queue this, 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, ...) 1765write_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, ...)
1766 ALIAS:
1767 enqueue_write_image = 0
816 PPCODE: 1768 PPCODE:
817 cl_event ev = 0; 1769 EVENT_LIST (12);
1770
818 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1771 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
819 const size_t region[3] = { width, height, depth }; 1772 const size_t region[3] = { width, height, depth };
820 STRLEN len; 1773 STRLEN len;
821 char *ptr = SvPVbyte (data, len); 1774 char *ptr = SvPVbyte (data, len);
822 EVENT_LIST (12, items - 12);
823 1775
824 if (!row_pitch) 1776 if (!row_pitch)
825 row_pitch = img_row_pitch (dst); 1777 row_pitch = img_row_pitch (dst);
826 1778
827 if (depth > 1 && !slice_pitch) 1779 if (depth > 1 && !slice_pitch)
830 size_t min_len = slice_pitch ? slice_pitch * depth : row_pitch * height; 1782 size_t min_len = slice_pitch ? slice_pitch * depth : row_pitch * height;
831 1783
832 if (len < min_len) 1784 if (len < min_len)
833 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1785 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
834 1786
1787 cl_event ev = 0;
835 NEED_SUCCESS (EnqueueWriteImage, (this, dst, blocking, dst_origin, region, row_pitch, slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1788 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));
836 1789
837 if (ev) 1790 if (ev)
838 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1791 XPUSH_CLOBJ (stash_event, ev);
839 1792
840void 1793void
841enqueue_copy_buffer_rect (OpenCL::Queue this, 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, ...) 1794copy_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, ...)
1795 ALIAS:
1796 enqueue_copy_image = 0
842 PPCODE: 1797 PPCODE:
843 cl_event ev = 0; 1798 EVENT_LIST (12);
1799
844 const size_t src_origin[3] = { src_x, src_y, src_z }; 1800 const size_t src_origin[3] = { src_x, src_y, src_z };
845 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1801 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
846 const size_t region[3] = { width, height, depth }; 1802 const size_t region[3] = { width, height, depth };
847 EVENT_LIST (16, items - 16);
848 1803
1804 cl_event ev = 0;
849 NEED_SUCCESS (EnqueueCopyBufferRect, (this, 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)); 1805 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
850 1806
851 if (ev) 1807 if (ev)
852 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1808 XPUSH_CLOBJ (stash_event, ev);
853 1809
854void 1810void
855enqueue_copy_buffer_to_image (OpenCL::Queue this, 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, ...) 1811copy_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, ...)
1812 ALIAS:
1813 enqueue_copy_image_to_buffer = 0
856 PPCODE: 1814 PPCODE:
1815 EVENT_LIST (10);
1816
1817 const size_t src_origin[3] = { src_x, src_y, src_z };
1818 const size_t region [3] = { width, height, depth };
1819
857 cl_event ev = 0; 1820 cl_event ev = 0;
1821 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1822
1823 if (ev)
1824 XPUSH_CLOBJ (stash_event, ev);
1825
1826void
1827copy_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, ...)
1828 ALIAS:
1829 enqueue_copy_buffer_to_image = 0
1830 PPCODE:
1831 EVENT_LIST (10);
1832
858 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1833 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1834 const size_t region [3] = { width, height, depth };
1835
1836 cl_event ev = 0;
1837 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1838
1839 if (ev)
1840 XPUSH_CLOBJ (stash_event, ev);
1841
1842void
1843map_buffer (OpenCL::Queue self, OpenCL::Buffer buf, cl_bool blocking = 1, cl_map_flags map_flags = CL_MAP_READ | CL_MAP_WRITE, size_t offset = 0, SV *cb_ = &PL_sv_undef, ...)
1844 ALIAS:
1845 enqueue_map_buffer = 0
1846 PPCODE:
1847 EVENT_LIST (6);
1848
1849 size_t cb = SvIV (cb_);
1850
1851 if (!SvOK (cb_))
1852 {
1853 NEED_SUCCESS (GetMemObjectInfo, (buf, CL_MEM_SIZE, sizeof (cb), &cb, 0));
1854 cb -= offset;
1855 }
1856
1857 cl_event ev;
1858 NEED_SUCCESS_ARG (void *ptr, EnqueueMapBuffer, (self, buf, blocking, map_flags, offset, cb, event_list_count, event_list_ptr, &ev, &res));
1859 XPUSHs (mapped_new (stash_mappedbuffer, self, buf, map_flags, ptr, cb, ev, 0, 0));
1860
1861void
1862map_image (OpenCL::Queue self, OpenCL::Image img, cl_bool blocking = 1, cl_map_flags map_flags = CL_MAP_READ | CL_MAP_WRITE, size_t x = 0, size_t y = 0, size_t z = 0, SV *width_ = &PL_sv_undef, SV *height_ = &PL_sv_undef, SV *depth_ = &PL_sv_undef, ...)
1863 ALIAS:
1864 enqueue_map_image = 0
1865 PPCODE:
1866 size_t width = SvIV (width_);
1867 if (!SvOK (width_))
1868 {
1869 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_WIDTH, sizeof (width), &width, 0));
1870 width -= x;
1871 }
1872
1873 size_t height = SvIV (width_);
1874 if (!SvOK (height_))
1875 {
1876 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_HEIGHT, sizeof (height), &height, 0));
1877 height -= y;
1878 }
1879
1880 size_t depth = SvIV (width_);
1881 if (!SvOK (depth_))
1882 {
1883 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_DEPTH, sizeof (depth), &depth, 0));
1884 depth -= z;
1885
1886 // stupid opencl returns 0 for depth, but requires 1 for 2d images
1887 if (!depth)
1888 depth = 1;
1889 }
1890
1891 const size_t origin[3] = { x, y, z };
859 const size_t region[3] = { width, height, depth }; 1892 const size_t region[3] = { width, height, depth };
1893 size_t row_pitch, slice_pitch;
860 EVENT_LIST (10, items - 10); 1894 EVENT_LIST (10);
861 1895
862 NEED_SUCCESS (EnqueueCopyBufferToImage, (this, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1896 cl_event ev;
1897 NEED_SUCCESS_ARG (void *ptr, EnqueueMapImage, (self, img, blocking, map_flags, origin, region, &row_pitch, &slice_pitch, event_list_count, event_list_ptr, &ev, &res));
1898
1899 size_t cb = slice_pitch ? slice_pitch * region [2]
1900 : row_pitch ? row_pitch * region [1]
1901 : region [0];
1902
1903 XPUSHs (mapped_new (stash_mappedimage, self, img, map_flags, ptr, cb, ev, row_pitch, slice_pitch));
1904
1905void
1906unmap (OpenCL::Queue self, OpenCL::Mapped mapped, ...)
1907 PPCODE:
1908 mapped_unmap (cv, ST (1), mapped, self, &ST (2), items - 2);
1909 if (GIMME_V != G_VOID)
1910 {
1911 clRetainEvent (mapped->event);
1912 XPUSH_CLOBJ (stash_event, mapped->event);
1913 }
1914
1915void
1916task (OpenCL::Queue self, OpenCL::Kernel kernel, ...)
1917 ALIAS:
1918 enqueue_task = 0
1919 PPCODE:
1920 EVENT_LIST (2);
1921
1922 cl_event ev = 0;
1923 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
863 1924
864 if (ev) 1925 if (ev)
865 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1926 XPUSH_CLOBJ (stash_event, ev);
866 1927
867void 1928void
868enqueue_copy_image (OpenCL::Queue this, 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, ...)
869 PPCODE:
870 cl_event ev = 0;
871 const size_t src_origin[3] = { src_x, src_y, src_z };
872 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
873 const size_t region[3] = { width, height, depth };
874 EVENT_LIST (12, items - 12);
875
876 NEED_SUCCESS (EnqueueCopyImage, (this, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
877
878 if (ev)
879 XPUSH_NEW_OBJ ("OpenCL::Event", ev);
880
881void
882enqueue_copy_image_to_buffer (OpenCL::Queue this, 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, ...)
883 PPCODE:
884 cl_event ev = 0;
885 const size_t src_origin[3] = { src_x, src_y, src_z };
886 const size_t region[3] = { width, height, depth };
887 EVENT_LIST (10, items - 10);
888
889 NEED_SUCCESS (EnqueueCopyImageToBuffer, (this, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
890
891 if (ev)
892 XPUSH_NEW_OBJ ("OpenCL::Event", ev);
893
894void
895enqueue_task (OpenCL::Queue this, OpenCL::Kernel kernel, ...)
896 PPCODE:
897 cl_event ev = 0;
898 EVENT_LIST (2, items - 2);
899
900 NEED_SUCCESS (EnqueueTask, (this, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
901
902 if (ev)
903 XPUSH_NEW_OBJ ("OpenCL::Event", ev);
904
905void
906enqueue_nd_range_kernel (OpenCL::Queue this, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...) 1929nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...)
1930 ALIAS:
1931 enqueue_nd_range_kernel = 0
907 PPCODE: 1932 PPCODE:
908 cl_event ev = 0; 1933 EVENT_LIST (5);
1934
909 size_t *gwo = 0, *gws, *lws = 0; 1935 size_t *gwo = 0, *gws, *lws = 0;
910 int gws_len; 1936 int gws_len;
911 size_t *lists; 1937 size_t *lists;
912 int i; 1938 int i;
913 EVENT_LIST (5, items - 5);
914 1939
915 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV) 1940 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV)
916 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference"); 1941 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference");
917 1942
918 gws_len = AvFILLp (SvRV (global_work_size)) + 1; 1943 gws_len = AvFILLp (SvRV (global_work_size)) + 1;
919 1944
920 lists = tmpbuf (sizeof (size_t) * 3 * gws_len); 1945 lists = tmpbuf (sizeof (size_t) * 3 * gws_len);
921 1946
922 gws = lists + gws_len * 0; 1947 gws = lists + gws_len * 0;
923 for (i = 0; i < gws_len; ++i) 1948 for (i = 0; i < gws_len; ++i)
1949 {
924 gws [i] = SvIV (AvARRAY (SvRV (global_work_size))[i]); 1950 gws [i] = SvIV (AvARRAY (SvRV (global_work_size))[i]);
1951 // at least nvidia crashes for 0-sized work group sizes, work around
1952 if (!gws [i])
1953 croak ("clEnqueueNDRangeKernel: global_work_size[%d] is zero, must be non-zero", i);
1954 }
925 1955
926 if (SvOK (global_work_offset)) 1956 if (SvOK (global_work_offset))
927 { 1957 {
928 if (!SvROK (global_work_offset) || SvTYPE (SvRV (global_work_offset)) != SVt_PVAV) 1958 if (!SvROK (global_work_offset) || SvTYPE (SvRV (global_work_offset)) != SVt_PVAV)
929 croak ("clEnqueueNDRangeKernel: global_work_offset must be undef or an array reference"); 1959 croak ("clEnqueueNDRangeKernel: global_work_offset must be undef or an array reference");
936 gwo [i] = SvIV (AvARRAY (SvRV (global_work_offset))[i]); 1966 gwo [i] = SvIV (AvARRAY (SvRV (global_work_offset))[i]);
937 } 1967 }
938 1968
939 if (SvOK (local_work_size)) 1969 if (SvOK (local_work_size))
940 { 1970 {
941 if (SvOK (local_work_size) && !SvROK (local_work_size) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV) 1971 if ((SvOK (local_work_size) && !SvROK (local_work_size)) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV)
942 croak ("clEnqueueNDRangeKernel: global_work_size must be undef or an array reference"); 1972 croak ("clEnqueueNDRangeKernel: local_work_size must be undef or an array reference");
943 1973
944 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len) 1974 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len)
945 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size"); 1975 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size");
946 1976
947 lws = lists + gws_len * 2; 1977 lws = lists + gws_len * 2;
948 for (i = 0; i < gws_len; ++i) 1978 for (i = 0; i < gws_len; ++i)
1979 {
949 lws [i] = SvIV (AvARRAY (SvRV (local_work_size))[i]); 1980 lws [i] = SvIV (AvARRAY (SvRV (local_work_size))[i]);
1981 // at least nvidia crashes for 0-sized work group sizes, work around
1982 if (!lws [i])
1983 croak ("clEnqueueNDRangeKernel: local_work_size[%d] is zero, must be non-zero", i);
1984 }
950 } 1985 }
951 1986
1987 cl_event ev = 0;
952 NEED_SUCCESS (EnqueueNDRangeKernel, (this, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1988 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
953 1989
954 if (ev) 1990 if (ev)
955 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1991 XPUSH_CLOBJ (stash_event, ev);
956 1992
1993#if CL_VERSION_1_2
1994
957void 1995void
958enqueue_marker (OpenCL::Queue this) 1996migrate_mem_objects (OpenCL::Queue self, SV *objects, cl_mem_migration_flags flags, ...)
1997 ALIAS:
1998 enqueue_migrate_mem_objects = 0
959 PPCODE: 1999 PPCODE:
2000 EVENT_LIST (3);
2001
2002 cl_uint obj_count;
2003 cl_mem *obj_list = object_list (cv, 0, "objects", objects, "OpenCL::Memory", &obj_count);
2004
2005 cl_event ev = 0;
2006 NEED_SUCCESS (EnqueueMigrateMemObjects, (self, obj_count, obj_list, flags, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2007
2008 if (ev)
2009 XPUSH_CLOBJ (stash_event, ev);
2010
2011#endif
2012
2013#if cl_apple_gl_sharing || cl_khr_gl_sharing
2014
2015void
2016acquire_gl_objects (OpenCL::Queue self, SV *objects, ...)
2017 ALIAS:
2018 release_gl_objects = 1
2019 enqueue_acquire_gl_objects = 0
2020 enqueue_release_gl_objects = 1
2021 PPCODE:
2022 EVENT_LIST (2);
2023
2024 cl_uint obj_count;
2025 cl_mem *obj_list = object_list (cv, 0, "objects", objects, "OpenCL::Memory", &obj_count);
2026
960 cl_event ev; 2027 cl_event ev = 0;
961 NEED_SUCCESS (EnqueueMarker, (this, &ev));
962 XPUSH_NEW_OBJ ("OpenCL::Event", ev);
963 2028
2029 if (ix)
2030 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, obj_count, obj_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2031 else
2032 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, obj_count, obj_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2033
2034 if (ev)
2035 XPUSH_CLOBJ (stash_event, ev);
2036
2037#endif
2038
964void 2039void
965enqueue_wait_for_events (OpenCL::Queue this, ...) 2040wait_for_events (OpenCL::Queue self, ...)
2041 ALIAS:
2042 enqueue_wait_for_events = 0
966 CODE: 2043 CODE:
967 EVENT_LIST (1, items - 1); 2044 EVENT_LIST (1);
2045#if PREFER_1_1
968 NEED_SUCCESS (EnqueueWaitForEvents, (this, event_list_count, event_list_ptr)); 2046 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
2047#else
2048 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0));
2049#endif
969 2050
970void 2051void
971enqueue_barrier (OpenCL::Queue this) 2052marker (OpenCL::Queue self, ...)
2053 ALIAS:
2054 enqueue_marker = 0
972 CODE: 2055 PPCODE:
2056 EVENT_LIST (1);
2057 cl_event ev = 0;
2058#if PREFER_1_1
2059 if (!event_list_count)
2060 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
2061 else
2062#if CL_VERSION_1_2
2063 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2064#else
2065 {
2066 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr)); // also a barrier
2067 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
2068 }
2069#endif
2070#else
2071 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2072#endif
2073 if (ev)
2074 XPUSH_CLOBJ (stash_event, ev);
2075
2076void
2077barrier (OpenCL::Queue self, ...)
2078 ALIAS:
2079 enqueue_barrier = 0
2080 PPCODE:
2081 EVENT_LIST (1);
2082 cl_event ev = 0;
2083#if PREFER_1_1
2084 if (!event_list_count && GIMME_V == G_VOID)
973 NEED_SUCCESS (EnqueueBarrier, (this)); 2085 NEED_SUCCESS (EnqueueBarrier, (self));
2086 else
2087#if CL_VERSION_1_2
2088 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2089#else
2090 {
2091 if (event_list_count)
2092 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
974 2093
2094 if (GIMME_V != G_VOID)
2095 NEED_SUCCESS (EnqueueMarker, (self, &ev));
2096 }
2097#endif
2098#else
2099 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2100#endif
2101 if (ev)
2102 XPUSH_CLOBJ (stash_event, ev);
2103
975void 2104void
976flush (OpenCL::Queue this) 2105flush (OpenCL::Queue self)
977 CODE: 2106 CODE:
978 NEED_SUCCESS (Flush, (this)); 2107 NEED_SUCCESS (Flush, (self));
979 2108
980void 2109void
981finish (OpenCL::Queue this) 2110finish (OpenCL::Queue self)
982 CODE: 2111 CODE:
983 NEED_SUCCESS (Finish, (this)); 2112 NEED_SUCCESS (Finish, (self));
984 2113
985void 2114void
986info (OpenCL::Queue this, cl_command_queue_info name) 2115info (OpenCL::Queue self, cl_command_queue_info name)
987 PPCODE: 2116 PPCODE:
988 INFO (CommandQueue) 2117 INFO (CommandQueue)
989 2118
990#BEGIN:command_queue 2119#BEGIN:command_queue
991 2120
992void 2121void
993context (OpenCL::Queue this) 2122context (OpenCL::Queue self)
994 PPCODE: 2123 PPCODE:
995 cl_context value [1]; 2124 cl_context value [1];
996 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_CONTEXT, sizeof (value), value, 0)); 2125 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0));
997 EXTEND (SP, 1); 2126 EXTEND (SP, 1);
998 const int i = 0; 2127 const int i = 0;
999 { 2128 {
1000 NEED_SUCCESS (RetainContext, (value [i])); 2129 NEED_SUCCESS (RetainContext, (value [i]));
1001 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2130 PUSH_CLOBJ (stash_context, value [i]);
1002 } 2131 }
1003 2132
1004void 2133void
1005device (OpenCL::Queue this) 2134device (OpenCL::Queue self)
1006 PPCODE: 2135 PPCODE:
1007 cl_device_id value [1]; 2136 cl_device_id value [1];
1008 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_DEVICE, sizeof (value), value, 0)); 2137 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0));
1009 EXTEND (SP, 1); 2138 EXTEND (SP, 1);
1010 const int i = 0; 2139 const int i = 0;
1011 { 2140 {
1012 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 2141 PUSH_CLOBJ (stash_device, value [i]);
1013 } 2142 }
1014 2143
1015void 2144void
1016reference_count (OpenCL::Queue this) 2145reference_count (OpenCL::Queue self)
1017 PPCODE: 2146 PPCODE:
1018 cl_uint value [1]; 2147 cl_uint value [1];
1019 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_REFERENCE_COUNT, sizeof (value), value, 0)); 2148 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_REFERENCE_COUNT, sizeof (value), value, 0));
1020 EXTEND (SP, 1); 2149 EXTEND (SP, 1);
1021 const int i = 0; 2150 const int i = 0;
1022 PUSHs (sv_2mortal (newSVuv (value [i]))); 2151 PUSHs (sv_2mortal (newSVuv (value [i])));
1023 2152
1024void 2153void
1025properties (OpenCL::Queue this) 2154properties (OpenCL::Queue self)
1026 PPCODE: 2155 PPCODE:
1027 cl_command_queue_properties value [1]; 2156 cl_command_queue_properties value [1];
1028 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_PROPERTIES, sizeof (value), value, 0)); 2157 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_PROPERTIES, sizeof (value), value, 0));
1029 EXTEND (SP, 1); 2158 EXTEND (SP, 1);
1030 const int i = 0; 2159 const int i = 0;
1031 PUSHs (sv_2mortal (newSViv (value [i]))); 2160 PUSHs (sv_2mortal (newSVuv (value [i])));
1032 2161
1033#END:command_queue 2162#END:command_queue
1034 2163
1035MODULE = OpenCL PACKAGE = OpenCL::Memory 2164MODULE = OpenCL PACKAGE = OpenCL::Memory
1036 2165
1037void 2166void
1038DESTROY (OpenCL::Memory this) 2167DESTROY (OpenCL::Memory self)
1039 CODE: 2168 CODE:
1040 clReleaseMemObject (this); 2169 clReleaseMemObject (self);
1041 2170
1042void 2171void
1043info (OpenCL::Memory this, cl_mem_info name) 2172info (OpenCL::Memory self, cl_mem_info name)
1044 PPCODE: 2173 PPCODE:
1045 INFO (MemObject) 2174 INFO (MemObject)
1046 2175
1047#BEGIN:mem 2176#BEGIN:mem
1048 2177
1049void 2178void
1050type (OpenCL::Memory this) 2179type (OpenCL::Memory self)
1051 PPCODE: 2180 PPCODE:
1052 cl_mem_object_type value [1]; 2181 cl_mem_object_type value [1];
1053 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_TYPE, sizeof (value), value, 0)); 2182 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_TYPE, sizeof (value), value, 0));
1054 EXTEND (SP, 1); 2183 EXTEND (SP, 1);
1055 const int i = 0; 2184 const int i = 0;
1056 PUSHs (sv_2mortal (newSViv (value [i]))); 2185 PUSHs (sv_2mortal (newSVuv (value [i])));
1057 2186
1058void 2187void
1059flags (OpenCL::Memory this) 2188flags (OpenCL::Memory self)
1060 PPCODE: 2189 PPCODE:
1061 cl_mem_flags value [1]; 2190 cl_mem_flags value [1];
1062 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_FLAGS, sizeof (value), value, 0)); 2191 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_FLAGS, sizeof (value), value, 0));
1063 EXTEND (SP, 1); 2192 EXTEND (SP, 1);
1064 const int i = 0; 2193 const int i = 0;
1065 PUSHs (sv_2mortal (newSViv (value [i]))); 2194 PUSHs (sv_2mortal (newSVuv (value [i])));
1066 2195
1067void 2196void
1068size (OpenCL::Memory this) 2197size (OpenCL::Memory self)
1069 ALIAS: 2198 ALIAS:
1070 size = CL_MEM_SIZE 2199 size = CL_MEM_SIZE
1071 offset = CL_MEM_OFFSET 2200 offset = CL_MEM_OFFSET
1072 PPCODE: 2201 PPCODE:
1073 size_t value [1]; 2202 size_t value [1];
1074 NEED_SUCCESS (GetMemObjectInfo, (this, ix, sizeof (value), value, 0)); 2203 NEED_SUCCESS (GetMemObjectInfo, (self, ix, sizeof (value), value, 0));
1075 EXTEND (SP, 1); 2204 EXTEND (SP, 1);
1076 const int i = 0; 2205 const int i = 0;
1077 PUSHs (sv_2mortal (newSVuv (value [i]))); 2206 PUSHs (sv_2mortal (newSVuv (value [i])));
1078 2207
1079void 2208void
1080host_ptr (OpenCL::Memory this) 2209host_ptr (OpenCL::Memory self)
1081 PPCODE: 2210 PPCODE:
1082 void * value [1]; 2211 void * value [1];
1083 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_HOST_PTR, sizeof (value), value, 0)); 2212 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_HOST_PTR, sizeof (value), value, 0));
1084 EXTEND (SP, 1); 2213 EXTEND (SP, 1);
1085 const int i = 0; 2214 const int i = 0;
1086 PUSHs (sv_2mortal (newSVuv ((IV)(intptr_t)value [i]))); 2215 PUSHs (sv_2mortal (newSVuv ((IV)(intptr_t)value [i])));
1087 2216
1088void 2217void
1089map_count (OpenCL::Memory this) 2218map_count (OpenCL::Memory self)
1090 ALIAS: 2219 ALIAS:
1091 map_count = CL_MEM_MAP_COUNT 2220 map_count = CL_MEM_MAP_COUNT
1092 reference_count = CL_MEM_REFERENCE_COUNT 2221 reference_count = CL_MEM_REFERENCE_COUNT
1093 PPCODE: 2222 PPCODE:
1094 cl_uint value [1]; 2223 cl_uint value [1];
1095 NEED_SUCCESS (GetMemObjectInfo, (this, ix, sizeof (value), value, 0)); 2224 NEED_SUCCESS (GetMemObjectInfo, (self, ix, sizeof (value), value, 0));
1096 EXTEND (SP, 1); 2225 EXTEND (SP, 1);
1097 const int i = 0; 2226 const int i = 0;
1098 PUSHs (sv_2mortal (newSVuv (value [i]))); 2227 PUSHs (sv_2mortal (newSVuv (value [i])));
1099 2228
1100void 2229void
1101context (OpenCL::Memory this) 2230context (OpenCL::Memory self)
1102 PPCODE: 2231 PPCODE:
1103 cl_context value [1]; 2232 cl_context value [1];
1104 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_CONTEXT, sizeof (value), value, 0)); 2233 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0));
1105 EXTEND (SP, 1); 2234 EXTEND (SP, 1);
1106 const int i = 0; 2235 const int i = 0;
1107 { 2236 {
1108 NEED_SUCCESS (RetainContext, (value [i])); 2237 NEED_SUCCESS (RetainContext, (value [i]));
1109 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2238 PUSH_CLOBJ (stash_context, value [i]);
1110 } 2239 }
1111 2240
1112void 2241void
1113associated_memobject (OpenCL::Memory this) 2242associated_memobject (OpenCL::Memory self)
1114 PPCODE: 2243 PPCODE:
1115 cl_mem value [1]; 2244 cl_mem value [1];
1116 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0)); 2245 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0));
1117 EXTEND (SP, 1); 2246 EXTEND (SP, 1);
1118 const int i = 0; 2247 const int i = 0;
1119 { 2248 {
1120 NEED_SUCCESS (RetainMemObject, (value [i])); 2249 NEED_SUCCESS (RetainMemObject, (value [i]));
1121 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Memory", value [i])); 2250 PUSH_CLOBJ (stash_memory, value [i]);
1122 } 2251 }
1123 2252
1124#END:mem 2253#END:mem
1125 2254
2255#if cl_apple_gl_sharing || cl_khr_gl_sharing
2256
2257void
2258gl_object_info (OpenCL::Memory self)
2259 PPCODE:
2260 cl_gl_object_type type;
2261 cl_GLuint name;
2262 NEED_SUCCESS (GetGLObjectInfo, (self, &type, &name));
2263 EXTEND (SP, 2);
2264 PUSHs (sv_2mortal (newSVuv (type)));
2265 PUSHs (sv_2mortal (newSVuv (name)));
2266
2267#endif
2268
2269MODULE = OpenCL PACKAGE = OpenCL::BufferObj
2270
2271void
2272sub_buffer_region (OpenCL::BufferObj self, cl_mem_flags flags, size_t origin, size_t size)
2273 PPCODE:
2274 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR | CL_MEM_ALLOC_HOST_PTR))
2275 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!");
2276
2277 cl_buffer_region crdata = { origin, size };
2278
2279 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res));
2280 XPUSH_CLOBJ (stash_buffer, mem);
2281
1126MODULE = OpenCL PACKAGE = OpenCL::Image 2282MODULE = OpenCL PACKAGE = OpenCL::Image
1127 2283
1128void 2284void
1129image_info (OpenCL::Image this, cl_image_info name) 2285image_info (OpenCL::Image self, cl_image_info name)
1130 PPCODE: 2286 PPCODE:
1131 INFO (Image) 2287 INFO (Image)
1132 2288
2289void
2290format (OpenCL::Image self)
2291 PPCODE:
2292 cl_image_format format;
2293 NEED_SUCCESS (GetImageInfo, (self, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
2294 EXTEND (SP, 2);
2295 PUSHs (sv_2mortal (newSVuv (format.image_channel_order)));
2296 PUSHs (sv_2mortal (newSVuv (format.image_channel_data_type)));
2297
1133#BEGIN:image 2298#BEGIN:image
1134 2299
1135void 2300void
1136element_size (OpenCL::Image this) 2301element_size (OpenCL::Image self)
1137 ALIAS: 2302 ALIAS:
1138 element_size = CL_IMAGE_ELEMENT_SIZE 2303 element_size = CL_IMAGE_ELEMENT_SIZE
1139 row_pitch = CL_IMAGE_ROW_PITCH 2304 row_pitch = CL_IMAGE_ROW_PITCH
1140 slice_pitch = CL_IMAGE_SLICE_PITCH 2305 slice_pitch = CL_IMAGE_SLICE_PITCH
1141 width = CL_IMAGE_WIDTH 2306 width = CL_IMAGE_WIDTH
1142 height = CL_IMAGE_HEIGHT 2307 height = CL_IMAGE_HEIGHT
1143 depth = CL_IMAGE_DEPTH 2308 depth = CL_IMAGE_DEPTH
1144 PPCODE: 2309 PPCODE:
1145 size_t value [1]; 2310 size_t value [1];
1146 NEED_SUCCESS (GetImageInfo, (this, ix, sizeof (value), value, 0)); 2311 NEED_SUCCESS (GetImageInfo, (self, ix, sizeof (value), value, 0));
1147 EXTEND (SP, 1); 2312 EXTEND (SP, 1);
1148 const int i = 0; 2313 const int i = 0;
1149 PUSHs (sv_2mortal (newSVuv (value [i]))); 2314 PUSHs (sv_2mortal (newSVuv (value [i])));
1150 2315
1151#END:image 2316#END:image
1152 2317
2318#if cl_apple_gl_sharing || cl_khr_gl_sharing
2319
2320#BEGIN:gl_texture
2321
2322void
2323target (OpenCL::Image self)
2324 PPCODE:
2325 cl_GLenum value [1];
2326 NEED_SUCCESS (GetGLTextureInfo, (self, CL_GL_TEXTURE_TARGET, sizeof (value), value, 0));
2327 EXTEND (SP, 1);
2328 const int i = 0;
2329 PUSHs (sv_2mortal (newSVuv (value [i])));
2330
2331void
2332gl_mipmap_level (OpenCL::Image self)
2333 PPCODE:
2334 cl_GLint value [1];
2335 NEED_SUCCESS (GetGLTextureInfo, (self, CL_GL_MIPMAP_LEVEL, sizeof (value), value, 0));
2336 EXTEND (SP, 1);
2337 const int i = 0;
2338 PUSHs (sv_2mortal (newSViv (value [i])));
2339
2340#END:gl_texture
2341
2342#endif
2343
1153MODULE = OpenCL PACKAGE = OpenCL::Sampler 2344MODULE = OpenCL PACKAGE = OpenCL::Sampler
1154 2345
1155void 2346void
1156DESTROY (OpenCL::Sampler this) 2347DESTROY (OpenCL::Sampler self)
1157 CODE: 2348 CODE:
1158 clReleaseSampler (this); 2349 clReleaseSampler (self);
1159 2350
1160void 2351void
1161info (OpenCL::Sampler this, cl_sampler_info name) 2352info (OpenCL::Sampler self, cl_sampler_info name)
1162 PPCODE: 2353 PPCODE:
1163 INFO (Sampler) 2354 INFO (Sampler)
1164 2355
1165#BEGIN:sampler 2356#BEGIN:sampler
1166 2357
1167void 2358void
1168reference_count (OpenCL::Sampler this) 2359reference_count (OpenCL::Sampler self)
1169 PPCODE: 2360 PPCODE:
1170 cl_uint value [1]; 2361 cl_uint value [1];
1171 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_REFERENCE_COUNT, sizeof (value), value, 0)); 2362 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_REFERENCE_COUNT, sizeof (value), value, 0));
1172 EXTEND (SP, 1); 2363 EXTEND (SP, 1);
1173 const int i = 0; 2364 const int i = 0;
1174 PUSHs (sv_2mortal (newSVuv (value [i]))); 2365 PUSHs (sv_2mortal (newSVuv (value [i])));
1175 2366
1176void 2367void
1177context (OpenCL::Sampler this) 2368context (OpenCL::Sampler self)
1178 PPCODE: 2369 PPCODE:
1179 cl_context value [1]; 2370 cl_context value [1];
1180 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_CONTEXT, sizeof (value), value, 0)); 2371 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0));
1181 EXTEND (SP, 1); 2372 EXTEND (SP, 1);
1182 const int i = 0; 2373 const int i = 0;
1183 { 2374 {
1184 NEED_SUCCESS (RetainContext, (value [i])); 2375 NEED_SUCCESS (RetainContext, (value [i]));
1185 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2376 PUSH_CLOBJ (stash_context, value [i]);
1186 } 2377 }
1187 2378
1188void 2379void
1189normalized_coords (OpenCL::Sampler this) 2380normalized_coords (OpenCL::Sampler self)
1190 PPCODE: 2381 PPCODE:
1191 cl_addressing_mode value [1]; 2382 cl_addressing_mode value [1];
1192 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0)); 2383 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0));
1193 EXTEND (SP, 1); 2384 EXTEND (SP, 1);
1194 const int i = 0; 2385 const int i = 0;
1195 PUSHs (sv_2mortal (newSViv (value [i]))); 2386 PUSHs (sv_2mortal (newSVuv (value [i])));
1196 2387
1197void 2388void
1198addressing_mode (OpenCL::Sampler this) 2389addressing_mode (OpenCL::Sampler self)
1199 PPCODE: 2390 PPCODE:
1200 cl_filter_mode value [1]; 2391 cl_filter_mode value [1];
1201 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0)); 2392 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0));
1202 EXTEND (SP, 1); 2393 EXTEND (SP, 1);
1203 const int i = 0; 2394 const int i = 0;
1204 PUSHs (sv_2mortal (newSViv (value [i]))); 2395 PUSHs (sv_2mortal (newSVuv (value [i])));
1205 2396
1206void 2397void
1207filter_mode (OpenCL::Sampler this) 2398filter_mode (OpenCL::Sampler self)
1208 PPCODE: 2399 PPCODE:
1209 cl_bool value [1]; 2400 cl_bool value [1];
1210 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_FILTER_MODE, sizeof (value), value, 0)); 2401 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_FILTER_MODE, sizeof (value), value, 0));
1211 EXTEND (SP, 1); 2402 EXTEND (SP, 1);
1212 const int i = 0; 2403 const int i = 0;
1213 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no)); 2404 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no));
1214 2405
1215#END:sampler 2406#END:sampler
1216 2407
1217MODULE = OpenCL PACKAGE = OpenCL::Program 2408MODULE = OpenCL PACKAGE = OpenCL::Program
1218 2409
1219void 2410void
1220DESTROY (OpenCL::Program this) 2411DESTROY (OpenCL::Program self)
1221 CODE: 2412 CODE:
1222 clReleaseProgram (this); 2413 clReleaseProgram (self);
1223 2414
1224void 2415void
1225build (OpenCL::Program this, OpenCL::Device device, SV *options = &PL_sv_undef) 2416build (OpenCL::Program self, SV *devices = &PL_sv_undef, SV *options = &PL_sv_undef, SV *notify = &PL_sv_undef)
2417 ALIAS:
2418 build_async = 1
1226 CODE: 2419 CODE:
2420 cl_uint device_count = 0;
2421 cl_device_id *device_list = 0;
2422
2423 if (SvOK (devices))
2424 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
2425
2426 void *user_data;
2427 program_callback pfn_notify = make_program_callback (notify, &user_data);
2428
2429 if (ix)
2430 build_program_async (self, device_count, device_list, SvPVbyte_nolen (options), user_data);
2431 else
2432 NEED_SUCCESS (BuildProgram, (self, device_count, device_list, SvPVbyte_nolen (options), pfn_notify, user_data));
2433
2434#if CL_VERSION_1_2
2435
2436void
2437compile (OpenCL::Program self, SV *devices, SV *options = &PL_sv_undef, SV *headers = &PL_sv_undef, SV *notify = &PL_sv_undef)
2438 CODE:
2439 cl_uint device_count = 0;
2440 cl_device_id *device_list = 0;
2441
2442 if (SvOK (devices))
2443 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
2444
2445 cl_uint header_count = 0;
2446 cl_program *header_list = 0;
2447 const char **header_name = 0;
2448
2449 if (SvOK (headers))
2450 {
2451 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVHV)
2452 croak ("clCompileProgram: headers must be undef or a hashref of name => OpenCL::Program pairs");
2453
2454 HV *hv = (HV *)SvRV (devices);
2455
2456 header_count = hv_iterinit (hv);
2457 header_list = tmpbuf (sizeof (*header_list) * header_count);
2458 header_name = tmpbuf (sizeof (*header_name) * header_count);
2459
2460 HE *he;
2461 int i = 0;
2462 while (he = hv_iternext (hv))
2463 {
2464 header_name [i] = SvPVbyte_nolen (HeSVKEY_force (he));
2465 header_list [i] = SvCLOBJ (cv, "headers", HeVAL (he), "OpenCL::Program");
2466 ++i;
2467 }
2468 }
2469
2470 void *user_data;
2471 program_callback pfn_notify = make_program_callback (notify, &user_data);
2472
1227 NEED_SUCCESS (BuildProgram, (this, 1, &device, SvPVbyte_nolen (options), 0, 0)); 2473 NEED_SUCCESS (CompileProgram, (self, device_count, device_list, SvPVbyte_nolen (options),
2474 header_count, header_list, header_name, pfn_notify, user_data));
1228 2475
2476#endif
2477
1229void 2478void
1230build_info (OpenCL::Program this, OpenCL::Device device, cl_program_build_info name) 2479build_info (OpenCL::Program self, OpenCL::Device device, cl_program_build_info name)
1231 PPCODE: 2480 PPCODE:
1232 size_t size; 2481 size_t size;
1233 NEED_SUCCESS (GetProgramBuildInfo, (this, device, name, 0, 0, &size)); 2482 NEED_SUCCESS (GetProgramBuildInfo, (self, device, name, 0, 0, &size));
1234 SV *sv = sv_2mortal (newSV (size)); 2483 SV *sv = sv_2mortal (newSV (size));
1235 SvUPGRADE (sv, SVt_PV); 2484 SvUPGRADE (sv, SVt_PV);
1236 SvPOK_only (sv); 2485 SvPOK_only (sv);
1237 SvCUR_set (sv, size); 2486 SvCUR_set (sv, size);
1238 NEED_SUCCESS (GetProgramBuildInfo, (this, device, name, size, SvPVX (sv), 0)); 2487 NEED_SUCCESS (GetProgramBuildInfo, (self, device, name, size, SvPVX (sv), 0));
1239 XPUSHs (sv); 2488 XPUSHs (sv);
1240 2489
1241#BEGIN:program_build 2490#BEGIN:program_build
1242 2491
1243void 2492void
1244build_status (OpenCL::Program this, OpenCL::Device device) 2493build_status (OpenCL::Program self, OpenCL::Device device)
1245 PPCODE: 2494 PPCODE:
1246 cl_build_status value [1]; 2495 cl_build_status value [1];
1247 NEED_SUCCESS (GetProgramBuildInfo, (this, device, CL_PROGRAM_BUILD_STATUS, sizeof (value), value, 0)); 2496 NEED_SUCCESS (GetProgramBuildInfo, (self, device, CL_PROGRAM_BUILD_STATUS, sizeof (value), value, 0));
1248 EXTEND (SP, 1); 2497 EXTEND (SP, 1);
1249 const int i = 0; 2498 const int i = 0;
1250 PUSHs (sv_2mortal (newSViv (value [i]))); 2499 PUSHs (sv_2mortal (newSViv (value [i])));
1251 2500
1252void 2501void
1253build_options (OpenCL::Program this, OpenCL::Device device) 2502build_options (OpenCL::Program self, OpenCL::Device device)
1254 ALIAS: 2503 ALIAS:
1255 build_options = CL_PROGRAM_BUILD_OPTIONS 2504 build_options = CL_PROGRAM_BUILD_OPTIONS
1256 build_log = CL_PROGRAM_BUILD_LOG 2505 build_log = CL_PROGRAM_BUILD_LOG
1257 PPCODE: 2506 PPCODE:
1258 size_t size; 2507 size_t size;
1259 NEED_SUCCESS (GetProgramBuildInfo, (this, device, ix, 0, 0, &size)); 2508 NEED_SUCCESS (GetProgramBuildInfo, (self, device, ix, 0, 0, &size));
1260 char *value = tmpbuf (size); 2509 char *value = tmpbuf (size);
1261 NEED_SUCCESS (GetProgramBuildInfo, (this, device, ix, size, value, 0)); 2510 NEED_SUCCESS (GetProgramBuildInfo, (self, device, ix, size, value, 0));
1262 EXTEND (SP, 1); 2511 EXTEND (SP, 1);
1263 const int i = 0; 2512 const int i = 0;
1264 PUSHs (sv_2mortal (newSVpv (value, 0))); 2513 PUSHs (sv_2mortal (newSVpv (value, 0)));
1265 2514
2515void
2516binary_type (OpenCL::Program self, OpenCL::Device device)
2517 PPCODE:
2518 cl_program_binary_type value [1];
2519 NEED_SUCCESS (GetProgramBuildInfo, (self, device, CL_PROGRAM_BINARY_TYPE, sizeof (value), value, 0));
2520 EXTEND (SP, 1);
2521 const int i = 0;
2522 PUSHs (sv_2mortal (newSVuv (UV)value [i])));
2523
1266#END:program_build 2524#END:program_build
1267 2525
1268void 2526void
1269kernel (OpenCL::Program program, SV *function) 2527kernel (OpenCL::Program program, SV *function)
1270 PPCODE: 2528 PPCODE:
1271 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res)); 2529 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res));
1272 XPUSH_NEW_OBJ ("OpenCL::Kernel", kernel); 2530 XPUSH_CLOBJ (stash_kernel, kernel);
1273 2531
1274void 2532void
2533kernels_in_program (OpenCL::Program program)
2534 PPCODE:
2535 cl_uint num_kernels;
2536 NEED_SUCCESS (CreateKernelsInProgram, (program, 0, 0, &num_kernels));
2537 cl_kernel *kernels = tmpbuf (sizeof (cl_kernel) * num_kernels);
2538 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0));
2539
2540 int i;
2541 EXTEND (SP, num_kernels);
2542 for (i = 0; i < num_kernels; ++i)
2543 PUSH_CLOBJ (stash_kernel, kernels [i]);
2544
2545void
1275info (OpenCL::Program this, cl_program_info name) 2546info (OpenCL::Program self, cl_program_info name)
1276 PPCODE: 2547 PPCODE:
1277 INFO (Program) 2548 INFO (Program)
1278 2549
1279void 2550void
1280binaries (OpenCL::Program this) 2551binaries (OpenCL::Program self)
1281 PPCODE: 2552 PPCODE:
1282 cl_uint n, i; 2553 cl_uint n, i;
1283 size_t size; 2554 size_t size;
1284 2555
1285 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_NUM_DEVICES , sizeof (n) , &n , 0)); 2556 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_NUM_DEVICES , sizeof (n) , &n , 0));
1286 if (!n) XSRETURN_EMPTY; 2557 if (!n) XSRETURN_EMPTY;
1287 2558
1288 size_t *sizes = tmpbuf (sizeof (*sizes) * n); 2559 size_t *sizes = tmpbuf (sizeof (*sizes) * n);
1289 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARY_SIZES, sizeof (*sizes) * n, sizes, &size)); 2560 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARY_SIZES, sizeof (*sizes) * n, sizes, &size));
1290 if (size != sizeof (*sizes) * n) XSRETURN_EMPTY; 2561 if (size != sizeof (*sizes) * n) XSRETURN_EMPTY;
1291 unsigned char **ptrs = tmpbuf (sizeof (*ptrs) * n); 2562 unsigned char **ptrs = tmpbuf (sizeof (*ptrs) * n);
1292 2563
1293 EXTEND (SP, n); 2564 EXTEND (SP, n);
1294 for (i = 0; i < n; ++i) 2565 for (i = 0; i < n; ++i)
1295 { 2566 {
1296 SV *sv = sv_2mortal (newSV (sizes [i])); 2567 SV *sv = sv_2mortal (newSV (sizes [i]));
1297 SvUPGRADE (sv, SVt_PV); 2568 SvUPGRADE (sv, SVt_PV);
1298 SvPOK_only (sv); 2569 SvPOK_only (sv);
1299 SvCUR_set (sv, sizes [i]); 2570 SvCUR_set (sv, sizes [i]);
1300 ptrs [i] = SvPVX (sv); 2571 ptrs [i] = (void *)SvPVX (sv);
1301 PUSHs (sv); 2572 PUSHs (sv);
1302 } 2573 }
1303 2574
1304 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARIES , sizeof (*ptrs ) * n, ptrs , &size)); 2575 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARIES , sizeof (*ptrs ) * n, ptrs , &size));
1305 if (size != sizeof (*ptrs) * n) XSRETURN_EMPTY; 2576 if (size != sizeof (*ptrs) * n) XSRETURN_EMPTY;
1306 2577
1307#BEGIN:program 2578#BEGIN:program
1308 2579
1309void 2580void
1310reference_count (OpenCL::Program this) 2581reference_count (OpenCL::Program self)
1311 ALIAS: 2582 ALIAS:
1312 reference_count = CL_PROGRAM_REFERENCE_COUNT 2583 reference_count = CL_PROGRAM_REFERENCE_COUNT
1313 num_devices = CL_PROGRAM_NUM_DEVICES 2584 num_devices = CL_PROGRAM_NUM_DEVICES
1314 PPCODE: 2585 PPCODE:
1315 cl_uint value [1]; 2586 cl_uint value [1];
1316 NEED_SUCCESS (GetProgramInfo, (this, ix, sizeof (value), value, 0)); 2587 NEED_SUCCESS (GetProgramInfo, (self, ix, sizeof (value), value, 0));
1317 EXTEND (SP, 1); 2588 EXTEND (SP, 1);
1318 const int i = 0; 2589 const int i = 0;
1319 PUSHs (sv_2mortal (newSVuv (value [i]))); 2590 PUSHs (sv_2mortal (newSVuv (value [i])));
1320 2591
1321void 2592void
1322context (OpenCL::Program this) 2593context (OpenCL::Program self)
1323 PPCODE: 2594 PPCODE:
1324 cl_context value [1]; 2595 cl_context value [1];
1325 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_CONTEXT, sizeof (value), value, 0)); 2596 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0));
1326 EXTEND (SP, 1); 2597 EXTEND (SP, 1);
1327 const int i = 0; 2598 const int i = 0;
1328 { 2599 {
1329 NEED_SUCCESS (RetainContext, (value [i])); 2600 NEED_SUCCESS (RetainContext, (value [i]));
1330 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2601 PUSH_CLOBJ (stash_context, value [i]);
1331 } 2602 }
1332 2603
1333void 2604void
1334devices (OpenCL::Program this) 2605devices (OpenCL::Program self)
1335 PPCODE: 2606 PPCODE:
1336 size_t size; 2607 size_t size;
1337 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_DEVICES, 0, 0, &size)); 2608 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, 0, 0, &size));
1338 cl_device_id *value = tmpbuf (size); 2609 cl_device_id *value = tmpbuf (size);
1339 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_DEVICES, size, value, 0)); 2610 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0));
1340 int i, n = size / sizeof (*value); 2611 int i, n = size / sizeof (*value);
1341 EXTEND (SP, n); 2612 EXTEND (SP, n);
1342 for (i = 0; i < n; ++i) 2613 for (i = 0; i < n; ++i)
1343 { 2614 {
1344 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 2615 PUSH_CLOBJ (stash_device, value [i]);
1345 } 2616 }
1346 2617
1347void 2618void
1348source (OpenCL::Program this) 2619source (OpenCL::Program self)
1349 PPCODE: 2620 PPCODE:
1350 size_t size; 2621 size_t size;
1351 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_SOURCE, 0, 0, &size)); 2622 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_SOURCE, 0, 0, &size));
1352 char *value = tmpbuf (size); 2623 char *value = tmpbuf (size);
1353 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_SOURCE, size, value, 0)); 2624 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_SOURCE, size, value, 0));
1354 EXTEND (SP, 1); 2625 EXTEND (SP, 1);
1355 const int i = 0; 2626 const int i = 0;
1356 PUSHs (sv_2mortal (newSVpv (value, 0))); 2627 PUSHs (sv_2mortal (newSVpv (value, 0)));
1357 2628
1358void 2629void
1359binary_sizes (OpenCL::Program this) 2630binary_sizes (OpenCL::Program self)
1360 PPCODE: 2631 PPCODE:
1361 size_t size; 2632 size_t size;
1362 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARY_SIZES, 0, 0, &size)); 2633 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARY_SIZES, 0, 0, &size));
1363 size_t *value = tmpbuf (size); 2634 size_t *value = tmpbuf (size);
1364 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARY_SIZES, size, value, 0)); 2635 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARY_SIZES, size, value, 0));
1365 int i, n = size / sizeof (*value); 2636 int i, n = size / sizeof (*value);
1366 EXTEND (SP, n); 2637 EXTEND (SP, n);
1367 for (i = 0; i < n; ++i) 2638 for (i = 0; i < n; ++i)
1368 PUSHs (sv_2mortal (newSVuv (value [i]))); 2639 PUSHs (sv_2mortal (newSVuv (value [i])));
1369 2640
1370#END:program 2641#END:program
1371 2642
1372MODULE = OpenCL PACKAGE = OpenCL::Kernel 2643MODULE = OpenCL PACKAGE = OpenCL::Kernel
1373 2644
1374void 2645void
1375DESTROY (OpenCL::Kernel this) 2646DESTROY (OpenCL::Kernel self)
1376 CODE: 2647 CODE:
1377 clReleaseKernel (this); 2648 clReleaseKernel (self);
1378 2649
1379void 2650void
2651setf (OpenCL::Kernel self, const char *format, ...)
2652 CODE:
2653 int i;
2654 for (i = 2; ; ++i)
2655 {
2656 while (*format == ' ')
2657 ++format;
2658
2659 char type = *format++;
2660
2661 if (!type)
2662 break;
2663
2664 if (i >= items)
2665 croak ("OpenCL::Kernel::setf format string too long (not enough arguments)");
2666
2667 SV *sv = ST (i);
2668
2669 union
2670 {
2671 cl_char cc; cl_uchar cC; cl_short cs; cl_ushort cS;
2672 cl_int ci; cl_uint cI; cl_long cl; cl_ulong cL;
2673 cl_half ch; cl_float cf; cl_double cd;
2674 cl_mem cm;
2675 cl_sampler ca;
2676 size_t cz;
2677 cl_event ce;
2678 } arg;
2679 size_t size;
2680 int nullarg = 0;
2681
2682 switch (type)
2683 {
2684 case 'c': arg.cc = SvIV (sv); size = sizeof (arg.cc); break;
2685 case 'C': arg.cC = SvUV (sv); size = sizeof (arg.cC); break;
2686 case 's': arg.cs = SvIV (sv); size = sizeof (arg.cs); break;
2687 case 'S': arg.cS = SvUV (sv); size = sizeof (arg.cS); break;
2688 case 'i': arg.ci = SvIV (sv); size = sizeof (arg.ci); break;
2689 case 'I': arg.cI = SvUV (sv); size = sizeof (arg.cI); break;
2690 case 'l': arg.cl = SvIV (sv); size = sizeof (arg.cl); break;
2691 case 'L': arg.cL = SvUV (sv); size = sizeof (arg.cL); break;
2692
2693 case 'h': arg.ch = SvUV (sv); size = sizeof (arg.ch); break;
2694 case 'f': arg.cf = SvNV (sv); size = sizeof (arg.cf); break;
2695 case 'd': arg.cd = SvNV (sv); size = sizeof (arg.cd); break;
2696
2697 case 'z': nullarg = 1; size = SvIV (sv); break;
2698
2699 case 'm': nullarg = !SvOK (sv); arg.cm = SvCLOBJ (cv, "m", sv, "OpenCL::Memory" ); size = sizeof (arg.cm); break;
2700 case 'a': nullarg = !SvOK (sv); arg.ca = SvCLOBJ (cv, "a", sv, "OpenCL::Sampler"); size = sizeof (arg.ca); break;
2701 case 'e': nullarg = !SvOK (sv); arg.ca = SvCLOBJ (cv, "e", sv, "OpenCL::Event" ); size = sizeof (arg.ce); break;
2702
2703 default:
2704 croak ("OpenCL::Kernel::setf format character '%c' not supported", type);
2705 }
2706
2707 res = clSetKernelArg (self, i - 2, size, nullarg ? 0 : &arg);
2708 if (res)
2709 croak ("OpenCL::Kernel::setf kernel parameter '%c' (#%d): %s", type, i - 2, err2str (res));
2710 }
2711
2712 if (i != items)
2713 croak ("OpenCL::Kernel::setf format string too short (too many arguments)");
2714
2715void
1380set_char (OpenCL::Kernel this, cl_uint idx, cl_char value) 2716set_char (OpenCL::Kernel self, cl_uint idx, cl_char value)
1381 CODE: 2717 CODE:
1382 clSetKernelArg (this, idx, sizeof (value), &value); 2718 clSetKernelArg (self, idx, sizeof (value), &value);
1383 2719
1384void 2720void
1385set_uchar (OpenCL::Kernel this, cl_uint idx, cl_uchar value) 2721set_uchar (OpenCL::Kernel self, cl_uint idx, cl_uchar value)
1386 CODE: 2722 CODE:
1387 clSetKernelArg (this, idx, sizeof (value), &value); 2723 clSetKernelArg (self, idx, sizeof (value), &value);
1388 2724
1389void 2725void
1390set_short (OpenCL::Kernel this, cl_uint idx, cl_short value) 2726set_short (OpenCL::Kernel self, cl_uint idx, cl_short value)
1391 CODE: 2727 CODE:
1392 clSetKernelArg (this, idx, sizeof (value), &value); 2728 clSetKernelArg (self, idx, sizeof (value), &value);
1393 2729
1394void 2730void
1395set_ushort (OpenCL::Kernel this, cl_uint idx, cl_ushort value) 2731set_ushort (OpenCL::Kernel self, cl_uint idx, cl_ushort value)
1396 CODE: 2732 CODE:
1397 clSetKernelArg (this, idx, sizeof (value), &value); 2733 clSetKernelArg (self, idx, sizeof (value), &value);
1398 2734
1399void 2735void
1400set_int (OpenCL::Kernel this, cl_uint idx, cl_int value) 2736set_int (OpenCL::Kernel self, cl_uint idx, cl_int value)
1401 CODE: 2737 CODE:
1402 clSetKernelArg (this, idx, sizeof (value), &value); 2738 clSetKernelArg (self, idx, sizeof (value), &value);
1403 2739
1404void 2740void
1405set_uint (OpenCL::Kernel this, cl_uint idx, cl_uint value) 2741set_uint (OpenCL::Kernel self, cl_uint idx, cl_uint value)
1406 CODE: 2742 CODE:
1407 clSetKernelArg (this, idx, sizeof (value), &value); 2743 clSetKernelArg (self, idx, sizeof (value), &value);
1408 2744
1409void 2745void
1410set_long (OpenCL::Kernel this, cl_uint idx, cl_long value) 2746set_long (OpenCL::Kernel self, cl_uint idx, cl_long value)
1411 CODE: 2747 CODE:
1412 clSetKernelArg (this, idx, sizeof (value), &value); 2748 clSetKernelArg (self, idx, sizeof (value), &value);
1413 2749
1414void 2750void
1415set_ulong (OpenCL::Kernel this, cl_uint idx, cl_ulong value) 2751set_ulong (OpenCL::Kernel self, cl_uint idx, cl_ulong value)
1416 CODE: 2752 CODE:
1417 clSetKernelArg (this, idx, sizeof (value), &value); 2753 clSetKernelArg (self, idx, sizeof (value), &value);
1418 2754
1419void 2755void
1420set_half (OpenCL::Kernel this, cl_uint idx, cl_half value) 2756set_half (OpenCL::Kernel self, cl_uint idx, cl_half value)
1421 CODE: 2757 CODE:
1422 clSetKernelArg (this, idx, sizeof (value), &value); 2758 clSetKernelArg (self, idx, sizeof (value), &value);
1423 2759
1424void 2760void
1425set_float (OpenCL::Kernel this, cl_uint idx, cl_float value) 2761set_float (OpenCL::Kernel self, cl_uint idx, cl_float value)
1426 CODE: 2762 CODE:
1427 clSetKernelArg (this, idx, sizeof (value), &value); 2763 clSetKernelArg (self, idx, sizeof (value), &value);
1428 2764
1429void 2765void
1430set_double (OpenCL::Kernel this, cl_uint idx, cl_double value) 2766set_double (OpenCL::Kernel self, cl_uint idx, cl_double value)
1431 CODE: 2767 CODE:
1432 clSetKernelArg (this, idx, sizeof (value), &value); 2768 clSetKernelArg (self, idx, sizeof (value), &value);
1433 2769
1434void 2770void
1435set_memory (OpenCL::Kernel this, cl_uint idx, OpenCL::Memory_ornull value) 2771set_memory (OpenCL::Kernel self, cl_uint idx, OpenCL::Memory_ornull value)
1436 CODE: 2772 CODE:
1437 clSetKernelArg (this, idx, sizeof (value), &value); 2773 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
1438 2774
1439void 2775void
1440set_buffer (OpenCL::Kernel this, cl_uint idx, OpenCL::Buffer_ornull value) 2776set_buffer (OpenCL::Kernel self, cl_uint idx, OpenCL::Buffer_ornull value)
1441 CODE: 2777 CODE:
1442 clSetKernelArg (this, idx, sizeof (value), &value); 2778 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
1443 2779
1444void 2780void
1445set_image2d (OpenCL::Kernel this, cl_uint idx, OpenCL::Image2D_ornull value) 2781set_image (OpenCL::Kernel self, cl_uint idx, OpenCL::Image_ornull value)
2782 ALIAS:
2783 set_image2d = 0
2784 set_image3d = 0
1446 CODE: 2785 CODE:
1447 clSetKernelArg (this, idx, sizeof (value), &value); 2786 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
1448 2787
1449void 2788void
1450set_image3d (OpenCL::Kernel this, cl_uint idx, OpenCL::Image3D_ornull value)
1451 CODE:
1452 clSetKernelArg (this, idx, sizeof (value), &value);
1453
1454void
1455set_sampler (OpenCL::Kernel this, cl_uint idx, OpenCL::Sampler value) 2789set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value)
1456 CODE: 2790 CODE:
1457 clSetKernelArg (this, idx, sizeof (value), &value); 2791 clSetKernelArg (self, idx, sizeof (value), &value);
1458 2792
1459void 2793void
2794set_local (OpenCL::Kernel self, cl_uint idx, size_t size)
2795 CODE:
2796 clSetKernelArg (self, idx, size, 0);
2797
2798void
1460set_event (OpenCL::Kernel this, cl_uint idx, OpenCL::Event value) 2799set_event (OpenCL::Kernel self, cl_uint idx, OpenCL::Event value)
1461 CODE: 2800 CODE:
1462 clSetKernelArg (this, idx, sizeof (value), &value); 2801 clSetKernelArg (self, idx, sizeof (value), &value);
1463 2802
1464void 2803void
1465info (OpenCL::Kernel this, cl_kernel_info name) 2804info (OpenCL::Kernel self, cl_kernel_info name)
1466 PPCODE: 2805 PPCODE:
1467 INFO (Kernel) 2806 INFO (Kernel)
1468 2807
1469#BEGIN:kernel 2808#BEGIN:kernel
1470 2809
1471void 2810void
1472function_name (OpenCL::Kernel this) 2811function_name (OpenCL::Kernel self)
1473 PPCODE: 2812 PPCODE:
1474 size_t size; 2813 size_t size;
1475 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_FUNCTION_NAME, 0, 0, &size)); 2814 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_FUNCTION_NAME, 0, 0, &size));
1476 char *value = tmpbuf (size); 2815 char *value = tmpbuf (size);
1477 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_FUNCTION_NAME, size, value, 0)); 2816 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_FUNCTION_NAME, size, value, 0));
1478 EXTEND (SP, 1); 2817 EXTEND (SP, 1);
1479 const int i = 0; 2818 const int i = 0;
1480 PUSHs (sv_2mortal (newSVpv (value, 0))); 2819 PUSHs (sv_2mortal (newSVpv (value, 0)));
1481 2820
1482void 2821void
1483num_args (OpenCL::Kernel this) 2822num_args (OpenCL::Kernel self)
1484 ALIAS: 2823 ALIAS:
1485 num_args = CL_KERNEL_NUM_ARGS 2824 num_args = CL_KERNEL_NUM_ARGS
1486 reference_count = CL_KERNEL_REFERENCE_COUNT 2825 reference_count = CL_KERNEL_REFERENCE_COUNT
1487 PPCODE: 2826 PPCODE:
1488 cl_uint value [1]; 2827 cl_uint value [1];
1489 NEED_SUCCESS (GetKernelInfo, (this, ix, sizeof (value), value, 0)); 2828 NEED_SUCCESS (GetKernelInfo, (self, ix, sizeof (value), value, 0));
1490 EXTEND (SP, 1); 2829 EXTEND (SP, 1);
1491 const int i = 0; 2830 const int i = 0;
1492 PUSHs (sv_2mortal (newSVuv (value [i]))); 2831 PUSHs (sv_2mortal (newSVuv (value [i])));
1493 2832
1494void 2833void
1495context (OpenCL::Kernel this) 2834context (OpenCL::Kernel self)
1496 PPCODE: 2835 PPCODE:
1497 cl_context value [1]; 2836 cl_context value [1];
1498 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_CONTEXT, sizeof (value), value, 0)); 2837 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0));
1499 EXTEND (SP, 1); 2838 EXTEND (SP, 1);
1500 const int i = 0; 2839 const int i = 0;
1501 { 2840 {
1502 NEED_SUCCESS (RetainContext, (value [i])); 2841 NEED_SUCCESS (RetainContext, (value [i]));
1503 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2842 PUSH_CLOBJ (stash_context, value [i]);
1504 } 2843 }
1505 2844
1506void 2845void
1507program (OpenCL::Kernel this) 2846program (OpenCL::Kernel self)
1508 PPCODE: 2847 PPCODE:
1509 cl_program value [1]; 2848 cl_program value [1];
1510 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_PROGRAM, sizeof (value), value, 0)); 2849 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0));
1511 EXTEND (SP, 1); 2850 EXTEND (SP, 1);
1512 const int i = 0; 2851 const int i = 0;
1513 { 2852 {
1514 NEED_SUCCESS (RetainProgram, (value [i])); 2853 NEED_SUCCESS (RetainProgram, (value [i]));
1515 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Program", value [i])); 2854 PUSH_CLOBJ (stash_program, value [i]);
1516 } 2855 }
1517 2856
1518#END:kernel 2857#END:kernel
1519 2858
1520void 2859void
1521work_group_info (OpenCL::Kernel this, OpenCL::Device device, cl_kernel_work_group_info name) 2860work_group_info (OpenCL::Kernel self, OpenCL::Device device, cl_kernel_work_group_info name)
1522 PPCODE: 2861 PPCODE:
1523 size_t size; 2862 size_t size;
1524 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, name, 0, 0, &size)); 2863 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, name, 0, 0, &size));
1525 SV *sv = sv_2mortal (newSV (size)); 2864 SV *sv = sv_2mortal (newSV (size));
1526 SvUPGRADE (sv, SVt_PV); 2865 SvUPGRADE (sv, SVt_PV);
1527 SvPOK_only (sv); 2866 SvPOK_only (sv);
1528 SvCUR_set (sv, size); 2867 SvCUR_set (sv, size);
1529 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, name, size, SvPVX (sv), 0)); 2868 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, name, size, SvPVX (sv), 0));
1530 XPUSHs (sv); 2869 XPUSHs (sv);
1531 2870
1532#BEGIN:kernel_work_group 2871#BEGIN:kernel_work_group
1533 2872
1534void 2873void
1535work_group_size (OpenCL::Kernel this, OpenCL::Device device) 2874work_group_size (OpenCL::Kernel self, OpenCL::Device device)
1536 ALIAS: 2875 ALIAS:
1537 work_group_size = CL_KERNEL_WORK_GROUP_SIZE 2876 work_group_size = CL_KERNEL_WORK_GROUP_SIZE
1538 preferred_work_group_size_multiple = CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE 2877 preferred_work_group_size_multiple = CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE
1539 PPCODE: 2878 PPCODE:
1540 size_t value [1]; 2879 size_t value [1];
1541 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, ix, sizeof (value), value, 0)); 2880 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, ix, sizeof (value), value, 0));
1542 EXTEND (SP, 1); 2881 EXTEND (SP, 1);
1543 const int i = 0; 2882 const int i = 0;
1544 PUSHs (sv_2mortal (newSVuv (value [i]))); 2883 PUSHs (sv_2mortal (newSVuv (value [i])));
1545 2884
1546void 2885void
1547compile_work_group_size (OpenCL::Kernel this, OpenCL::Device device) 2886compile_work_group_size (OpenCL::Kernel self, OpenCL::Device device)
1548 PPCODE: 2887 PPCODE:
1549 size_t size; 2888 size_t size;
1550 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, 0, 0, &size)); 2889 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, 0, 0, &size));
1551 size_t *value = tmpbuf (size); 2890 size_t *value = tmpbuf (size);
1552 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, size, value, 0)); 2891 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, size, value, 0));
1553 int i, n = size / sizeof (*value); 2892 int i, n = size / sizeof (*value);
1554 EXTEND (SP, n); 2893 EXTEND (SP, n);
1555 for (i = 0; i < n; ++i) 2894 for (i = 0; i < n; ++i)
1556 PUSHs (sv_2mortal (newSVuv (value [i]))); 2895 PUSHs (sv_2mortal (newSVuv (value [i])));
1557 2896
1558void 2897void
1559local_mem_size (OpenCL::Kernel this, OpenCL::Device device) 2898local_mem_size (OpenCL::Kernel self, OpenCL::Device device)
1560 ALIAS: 2899 ALIAS:
1561 local_mem_size = CL_KERNEL_LOCAL_MEM_SIZE 2900 local_mem_size = CL_KERNEL_LOCAL_MEM_SIZE
1562 private_mem_size = CL_KERNEL_PRIVATE_MEM_SIZE 2901 private_mem_size = CL_KERNEL_PRIVATE_MEM_SIZE
1563 PPCODE: 2902 PPCODE:
1564 cl_ulong value [1]; 2903 cl_ulong value [1];
1565 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, ix, sizeof (value), value, 0)); 2904 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, ix, sizeof (value), value, 0));
1566 EXTEND (SP, 1); 2905 EXTEND (SP, 1);
1567 const int i = 0; 2906 const int i = 0;
1568 PUSHs (sv_2mortal (newSVuv (value [i]))); 2907 PUSHs (sv_2mortal (newSVuv (value [i])));
1569 2908
1570#END:kernel_work_group 2909#END:kernel_work_group
1571 2910
2911#if CL_VERSION_1_2
2912
2913void
2914arg_info (OpenCL::Kernel self, cl_uint idx, cl_kernel_arg_info name)
2915 PPCODE:
2916 size_t size;
2917 NEED_SUCCESS (GetKernelArgInfo, (self, idx, name, 0, 0, &size));
2918 SV *sv = sv_2mortal (newSV (size));
2919 SvUPGRADE (sv, SVt_PV);
2920 SvPOK_only (sv);
2921 SvCUR_set (sv, size);
2922 NEED_SUCCESS (GetKernelArgInfo, (self, idx, name, size, SvPVX (sv), 0));
2923 XPUSHs (sv);
2924
2925#BEGIN:kernel_arg
2926
2927void
2928arg_address_qualifier (OpenCL::Kernel self, cl_uint idx)
2929 PPCODE:
2930 cl_kernel_arg_address_qualifier value [1];
2931 NEED_SUCCESS (GetKernelArgInfo, (self, idx, CL_KERNEL_ARG_ADDRESS_QUALIFIER, sizeof (value), value, 0));
2932 EXTEND (SP, 1);
2933 const int i = 0;
2934 PUSHs (sv_2mortal (newSVuv (value [i])));
2935
2936void
2937arg_access_qualifier (OpenCL::Kernel self, cl_uint idx)
2938 PPCODE:
2939 cl_kernel_arg_access_qualifier value [1];
2940 NEED_SUCCESS (GetKernelArgInfo, (self, idx, CL_KERNEL_ARG_ACCESS_QUALIFIER, sizeof (value), value, 0));
2941 EXTEND (SP, 1);
2942 const int i = 0;
2943 PUSHs (sv_2mortal (newSVuv (value [i])));
2944
2945void
2946arg_type_name (OpenCL::Kernel self, cl_uint idx)
2947 ALIAS:
2948 arg_type_name = CL_KERNEL_ARG_TYPE_NAME
2949 arg_name = CL_KERNEL_ARG_NAME
2950 PPCODE:
2951 size_t size;
2952 NEED_SUCCESS (GetKernelArgInfo, (self, idx, ix, 0, 0, &size));
2953 char *value = tmpbuf (size);
2954 NEED_SUCCESS (GetKernelArgInfo, (self, idx, ix, size, value, 0));
2955 EXTEND (SP, 1);
2956 const int i = 0;
2957 PUSHs (sv_2mortal (newSVpv (value, 0)));
2958
2959void
2960arg_type_qualifier (OpenCL::Kernel self, cl_uint idx)
2961 PPCODE:
2962 cl_kernel_arg_type_qualifier value [1];
2963 NEED_SUCCESS (GetKernelArgInfo, (self, idx, CL_KERNEL_ARG_TYPE_QUALIFIER, sizeof (value), value, 0));
2964 EXTEND (SP, 1);
2965 const int i = 0;
2966 PUSHs (sv_2mortal (newSVuv (value [i])));
2967
2968#END:kernel_arg
2969
2970#endif
2971
1572MODULE = OpenCL PACKAGE = OpenCL::Event 2972MODULE = OpenCL PACKAGE = OpenCL::Event
1573 2973
1574void 2974void
1575DESTROY (OpenCL::Event this) 2975DESTROY (OpenCL::Event self)
1576 CODE: 2976 CODE:
1577 clReleaseEvent (this); 2977 clReleaseEvent (self);
1578 2978
1579void 2979void
1580wait (OpenCL::Event this) 2980wait (OpenCL::Event self)
1581 CODE: 2981 CODE:
1582 clWaitForEvents (1, &this); 2982 clWaitForEvents (1, &self);
1583 2983
1584void 2984void
2985cb (OpenCL::Event self, cl_int command_exec_callback_type, SV *cb)
2986 CODE:
2987 clSetEventCallback (self, command_exec_callback_type, eq_event_notify, SvREFCNT_inc (s_get_cv (cb)));
2988
2989void
1585info (OpenCL::Event this, cl_event_info name) 2990info (OpenCL::Event self, cl_event_info name)
1586 PPCODE: 2991 PPCODE:
1587 INFO (Event) 2992 INFO (Event)
1588 2993
1589#BEGIN:event 2994#BEGIN:event
1590 2995
1591void 2996void
1592command_queue (OpenCL::Event this) 2997command_queue (OpenCL::Event self)
1593 PPCODE: 2998 PPCODE:
1594 cl_command_queue value [1]; 2999 cl_command_queue value [1];
1595 NEED_SUCCESS (GetEventInfo, (this, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0)); 3000 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0));
1596 EXTEND (SP, 1); 3001 EXTEND (SP, 1);
1597 const int i = 0; 3002 const int i = 0;
1598 { 3003 {
1599 NEED_SUCCESS (RetainCommandQueue, (value [i])); 3004 NEED_SUCCESS (RetainCommandQueue, (value [i]));
1600 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Queue", value [i])); 3005 PUSH_CLOBJ (stash_queue, value [i]);
1601 } 3006 }
1602 3007
1603void 3008void
1604command_type (OpenCL::Event this) 3009command_type (OpenCL::Event self)
1605 PPCODE: 3010 PPCODE:
1606 cl_command_type value [1]; 3011 cl_command_type value [1];
1607 NEED_SUCCESS (GetEventInfo, (this, CL_EVENT_COMMAND_TYPE, sizeof (value), value, 0)); 3012 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_TYPE, sizeof (value), value, 0));
1608 EXTEND (SP, 1); 3013 EXTEND (SP, 1);
1609 const int i = 0; 3014 const int i = 0;
1610 PUSHs (sv_2mortal (newSVuv (value [i]))); 3015 PUSHs (sv_2mortal (newSVuv (value [i])));
1611 3016
1612void 3017void
1613reference_count (OpenCL::Event this) 3018reference_count (OpenCL::Event self)
1614 ALIAS: 3019 ALIAS:
1615 reference_count = CL_EVENT_REFERENCE_COUNT 3020 reference_count = CL_EVENT_REFERENCE_COUNT
1616 command_execution_status = CL_EVENT_COMMAND_EXECUTION_STATUS 3021 command_execution_status = CL_EVENT_COMMAND_EXECUTION_STATUS
1617 PPCODE: 3022 PPCODE:
1618 cl_uint value [1]; 3023 cl_uint value [1];
1619 NEED_SUCCESS (GetEventInfo, (this, ix, sizeof (value), value, 0)); 3024 NEED_SUCCESS (GetEventInfo, (self, ix, sizeof (value), value, 0));
1620 EXTEND (SP, 1); 3025 EXTEND (SP, 1);
1621 const int i = 0; 3026 const int i = 0;
1622 PUSHs (sv_2mortal (newSVuv (value [i]))); 3027 PUSHs (sv_2mortal (newSVuv (value [i])));
1623 3028
1624void 3029void
1625context (OpenCL::Event this) 3030context (OpenCL::Event self)
1626 PPCODE: 3031 PPCODE:
1627 cl_context value [1]; 3032 cl_context value [1];
1628 NEED_SUCCESS (GetEventInfo, (this, CL_EVENT_CONTEXT, sizeof (value), value, 0)); 3033 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0));
1629 EXTEND (SP, 1); 3034 EXTEND (SP, 1);
1630 const int i = 0; 3035 const int i = 0;
1631 { 3036 {
1632 NEED_SUCCESS (RetainContext, (value [i])); 3037 NEED_SUCCESS (RetainContext, (value [i]));
1633 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 3038 PUSH_CLOBJ (stash_context, value [i]);
1634 } 3039 }
1635 3040
1636#END:event 3041#END:event
1637 3042
1638void 3043void
1639profiling_info (OpenCL::Event this, cl_profiling_info name) 3044profiling_info (OpenCL::Event self, cl_profiling_info name)
1640 PPCODE: 3045 PPCODE:
1641 INFO (EventProfiling) 3046 INFO (EventProfiling)
1642 3047
1643#BEGIN:profiling 3048#BEGIN:profiling
1644 3049
1645void 3050void
1646profiling_command_queued (OpenCL::Event this) 3051profiling_command_queued (OpenCL::Event self)
1647 ALIAS: 3052 ALIAS:
1648 profiling_command_queued = CL_PROFILING_COMMAND_QUEUED 3053 profiling_command_queued = CL_PROFILING_COMMAND_QUEUED
1649 profiling_command_submit = CL_PROFILING_COMMAND_SUBMIT 3054 profiling_command_submit = CL_PROFILING_COMMAND_SUBMIT
1650 profiling_command_start = CL_PROFILING_COMMAND_START 3055 profiling_command_start = CL_PROFILING_COMMAND_START
1651 profiling_command_end = CL_PROFILING_COMMAND_END 3056 profiling_command_end = CL_PROFILING_COMMAND_END
1652 PPCODE: 3057 PPCODE:
1653 cl_ulong value [1]; 3058 cl_ulong value [1];
1654 NEED_SUCCESS (GetEventProfilingInfo, (this, ix, sizeof (value), value, 0)); 3059 NEED_SUCCESS (GetEventProfilingInfo, (self, ix, sizeof (value), value, 0));
1655 EXTEND (SP, 1); 3060 EXTEND (SP, 1);
1656 const int i = 0; 3061 const int i = 0;
1657 PUSHs (sv_2mortal (newSVuv (value [i]))); 3062 PUSHs (sv_2mortal (newSVuv (value [i])));
1658 3063
1659#END:profiling 3064#END:profiling
1660 3065
1661MODULE = OpenCL PACKAGE = OpenCL::UserEvent 3066MODULE = OpenCL PACKAGE = OpenCL::UserEvent
1662 3067
1663void 3068void
1664set_status (OpenCL::UserEvent this, cl_int execution_status) 3069set_status (OpenCL::UserEvent self, cl_int execution_status)
1665 CODE: 3070 CODE:
1666 clSetUserEventStatus (this, execution_status); 3071 clSetUserEventStatus (self, execution_status);
1667 3072
3073MODULE = OpenCL PACKAGE = OpenCL::Mapped
3074
3075void
3076DESTROY (SV *self)
3077 CODE:
3078 OpenCL__Mapped mapped = SvMAPPED (self);
3079
3080 clEnqueueUnmapMemObject (mapped->queue, mapped->memobj, mapped->ptr, 1, &mapped->event, 0);
3081 mapped_detach (self, mapped);
3082
3083 clReleaseCommandQueue (mapped->queue);
3084 clReleaseEvent (mapped->event);
3085 Safefree (mapped);
3086
3087void
3088unmap (OpenCL::Mapped self, ...)
3089 CODE:
3090 mapped_unmap (cv, ST (0), self, self->queue, &ST (1), items - 1);
3091
3092bool
3093mapped (OpenCL::Mapped self)
3094 CODE:
3095 RETVAL = !!self->ptr;
3096 OUTPUT:
3097 RETVAL
3098
3099void
3100wait (OpenCL::Mapped self)
3101 PPCODE:
3102 if (self->event)
3103 NEED_SUCCESS (WaitForEvents, (1, &self->event));
3104
3105void
3106event (OpenCL::Mapped self)
3107 PPCODE:
3108 if (!self->event)
3109 XSRETURN_UNDEF;
3110
3111 clRetainEvent (self->event);
3112 XPUSH_CLOBJ (stash_event, self->event);
3113
3114size_t
3115size (OpenCL::Mapped self)
3116 CODE:
3117 RETVAL = self->cb;
3118 OUTPUT:
3119 RETVAL
3120
3121IV
3122ptr (OpenCL::Mapped self)
3123 CODE:
3124 RETVAL = PTR2IV (self->ptr);
3125 OUTPUT:
3126 RETVAL
3127
3128void
3129set (OpenCL::Mapped self, size_t offset, SV *data)
3130 CODE:
3131 STRLEN len;
3132 const char *ptr = SvPVbyte (data, len);
3133
3134 if (offset + len > self->cb)
3135 croak ("OpenCL::Mapped::set out of bound condition detected");
3136
3137 memcpy (offset + (char *)self->ptr, ptr, len);
3138
3139MODULE = OpenCL PACKAGE = OpenCL::MappedBuffer
3140
3141MODULE = OpenCL PACKAGE = OpenCL::MappedImage
3142
3143IV
3144row_pitch (OpenCL::Mapped self)
3145 ALIAS:
3146 slice_pitch = 1
3147 CODE:
3148 RETVAL = ix ? self->slice_pitch : self->row_pitch;
3149 OUTPUT:
3150 RETVAL
3151
3152
3153

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines