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 1.74 by root, Sat May 5 13:56:00 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 #define CLHDR(name) <OpenCL/name>
34#else
35 #define CLHDR(name) <CL/name>
36#endif
37
5#include <CL/opencl.h> 38#include CLHDR(opencl.h)
39
40#ifndef CL_VERSION_1_2
41 #include CLHDR(cl_d3d9.h)
42#endif
43
44#if _WIN32
45 #include CLHDR(cl_d3d10.h)
46 #if CL_VERSION_1_2
47 #include CLHDR<cl_d3d11.h>
48 #endif
49 #include CLHDR<cl_dx9_media_sharing.h.h>
50#endif
51
52#ifndef CL_VERSION_1_2
53 #undef PREFER_1_1
54 #define PREFER_1_1 1
55#endif
56
57// make sure all constants we might use are actually defined
58#include "default.h"
6 59
7typedef cl_platform_id OpenCL__Platform; 60typedef cl_platform_id OpenCL__Platform;
8typedef cl_device_id OpenCL__Device; 61typedef cl_device_id OpenCL__Device;
62typedef cl_device_id OpenCL__SubDevice;
9typedef cl_context OpenCL__Context; 63typedef cl_context OpenCL__Context;
10typedef cl_command_queue OpenCL__Queue; 64typedef cl_command_queue OpenCL__Queue;
11typedef cl_mem OpenCL__Memory; 65typedef cl_mem OpenCL__Memory;
12typedef cl_mem OpenCL__Buffer; 66typedef cl_mem OpenCL__Buffer;
67typedef cl_mem OpenCL__BufferObj;
13typedef cl_mem OpenCL__Image; 68typedef cl_mem OpenCL__Image;
14typedef cl_mem OpenCL__Image2D;
15typedef cl_mem OpenCL__Image3D;
16typedef cl_mem OpenCL__Memory_ornull; 69typedef cl_mem OpenCL__Memory_ornull;
17typedef cl_mem OpenCL__Buffer_ornull; 70typedef cl_mem OpenCL__Buffer_ornull;
18typedef cl_mem OpenCL__Image_ornull; 71typedef cl_mem OpenCL__Image_ornull;
19typedef cl_mem OpenCL__Image2D_ornull;
20typedef cl_mem OpenCL__Image3D_ornull;
21typedef cl_sampler OpenCL__Sampler; 72typedef cl_sampler OpenCL__Sampler;
22typedef cl_program OpenCL__Program; 73typedef cl_program OpenCL__Program;
23typedef cl_kernel OpenCL__Kernel; 74typedef cl_kernel OpenCL__Kernel;
24typedef cl_event OpenCL__Event; 75typedef cl_event OpenCL__Event;
25typedef cl_event OpenCL__UserEvent; 76typedef cl_event OpenCL__UserEvent;
26 77
27typedef SV *FUTURE; 78typedef struct mapped * OpenCL__Mapped;
79
80static HV
81 *stash_platform,
82 *stash_device,
83 *stash_subdevice,
84 *stash_context,
85 *stash_queue,
86 *stash_program,
87 *stash_kernel,
88 *stash_sampler,
89 *stash_event,
90 *stash_userevent,
91 *stash_memory,
92 *stash_buffer,
93 *stash_bufferobj,
94 *stash_image,
95 *stash_image1d,
96 *stash_image1darray,
97 *stash_image1dbuffer,
98 *stash_image2d,
99 *stash_image2darray,
100 *stash_image3d,
101 *stash_mapped,
102 *stash_mappedbuffer,
103 *stash_mappedimage;
28 104
29/*****************************************************************************/ 105/*****************************************************************************/
30 106
107// name must include a leading underscore
108// all of this horrors would be unneceesary if somebody wrote a proper OpenGL module
109// for perl. doh.
110static void *
111glsym (const char *name)
112{
113 void *fun = 0;
114
115 #if defined I_DLFCN && defined RTLD_DEFAULT
116 fun = dlsym (RTLD_DEFAULT, name + 1);
117 if (!fun) fun = dlsym (RTLD_DEFAULT, name);
118
119 if (!fun)
120 {
121 static void *libgl;
122 static const char *glso[] = {
123 "libGL.so.1",
124 "libGL.so.3",
125 "libGL.so.4.0",
126 "libGL.so",
127 "/usr/lib/libGL.so",
128 "/usr/X11R6/lib/libGL.1.dylib"
129 };
130 int i;
131
132 for (i = 0; !libgl && i < sizeof (glso) / sizeof (glso [0]); ++i)
133 {
134 libgl = dlopen (glso [i], RTLD_LAZY);
135 if (libgl)
136 break;
137 }
138
139 if (libgl)
140 {
141 fun = dlsym (libgl, name + 1);
142 if (!fun) fun = dlsym (libgl, name);
143 }
144 }
145 #endif
146
147 return fun;
148}
149
150/*****************************************************************************/
151
31/* up to two temporary buffers */ 152/* up to two temporary buffers */
32static void * 153static void * ecb_noinline
33tmpbuf (size_t size) 154tmpbuf (size_t size)
34{ 155{
156 enum { buffers = 4 };
35 static int idx; 157 static int idx;
36 static void *buf [2]; 158 static void *buf [buffers];
37 static size_t len [2]; 159 static size_t len [buffers];
38 160
39 idx ^= 1; 161 idx = (idx + 1) % buffers;
40 162
41 if (len [idx] < size) 163 if (len [idx] < size)
42 { 164 {
43 free (buf [idx]); 165 free (buf [idx]);
44 len [idx] = ((size + 31) & ~4095) + 4096 - 32; 166 len [idx] = ((size + 31) & ~4095) + 4096 - 32;
45 buf [idx] = malloc (len [idx]); 167 buf [idx] = malloc (len [idx]);
46 } 168 }
47 169
48 return buf [idx]; 170 return buf [idx];
171}
172
173static const char * ecb_noinline
174cv_get_name (CV *cv)
175{
176 static char fullname [256];
177
178 GV *gv = CvGV (cv); // gv better be nonzero
179
180 HV *stash = GvSTASH (gv);
181 const char *hvname = HvNAME_get (stash); // stash also better be nonzero
182 const char *gvname = GvNAME (gv);
183
184 snprintf (fullname, sizeof (fullname), "%s::%s", hvname, gvname);
185 return fullname;
49} 186}
50 187
51/*****************************************************************************/ 188/*****************************************************************************/
52 189
53typedef struct 190typedef struct
112 if (res) \ 249 if (res) \
113 FAIL (name); 250 FAIL (name);
114 251
115/*****************************************************************************/ 252/*****************************************************************************/
116 253
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 * 254static SV *
255new_clobj (HV *stash, IV id)
256{
257 return sv_2mortal (sv_bless (newRV_noinc (newSViv (id)), stash));
258}
259
260#define PUSH_CLOBJ(stash,id) PUSHs (new_clobj ((stash), (IV)(id)))
261#define XPUSH_CLOBJ(stash,id) XPUSHs (new_clobj ((stash), (IV)(id)))
262
263/* cl objects are either \$iv, or [$iv, ...] */
264/* they can be upgraded at runtime to the array form */
265static void * ecb_noinline
121SvPTROBJ (const char *func, const char *svname, SV *sv, const char *pkg) 266SvCLOBJ (CV *cv, const char *svname, SV *sv, const char *pkg)
122{ 267{
268 // sv_derived_from is quite slow :(
123 if (SvROK (sv) && sv_derived_from (sv, pkg)) 269 if (SvROK (sv) && sv_derived_from (sv, pkg))
124 return (void *)SvIV (SvRV (sv)); 270 return (void *)SvIV (SvRV (sv));
125 271
126 croak ("%s: %s is not of type %s", func, svname, pkg); 272 croak ("%s: %s is not of type %s", cv_get_name (cv), svname, pkg);
273}
274
275// the "no-inherit" version of the above
276static void * ecb_noinline
277SvCLOBJ_ni (CV *cv, const char *svname, SV *sv, HV *stash)
278{
279 if (SvROK (sv) && SvSTASH (SvRV (sv)) == stash)
280 return (void *)SvIV (SvRV (sv));
281
282 croak ("%s: %s is not of type %s", cv_get_name (cv), svname, HvNAME (stash));
127} 283}
128 284
129/*****************************************************************************/ 285/*****************************************************************************/
286
287static cl_context_properties * ecb_noinline
288SvCONTEXTPROPERTIES (CV *cv, const char *svname, SV *sv, cl_context_properties *extra, int extracount)
289{
290 if (!sv || !SvOK (sv))
291 if (extra)
292 sv = sv_2mortal (newRV_noinc ((SV *)newAV ())); // slow, but rarely used hopefully
293 else
294 return 0;
295
296 if (SvROK (sv) && SvTYPE (SvRV (sv)) == SVt_PVAV)
297 {
298 AV *av = (AV *)SvRV (sv);
299 int i, len = av_len (av) + 1;
300 cl_context_properties *p = tmpbuf (sizeof (cl_context_properties) * (len + extracount + 1));
301 cl_context_properties *l = p;
302
303 if (len & 1)
304 croak ("%s: %s is not a property list (must contain an even number of elements)", cv_get_name (cv), svname);
305
306 while (extracount--)
307 *l++ = *extra++;
308
309 for (i = 0; i < len; i += 2)
310 {
311 cl_context_properties t = SvIV (*av_fetch (av, i , 0));
312 SV *p_sv = *av_fetch (av, i + 1, 0);
313 cl_context_properties v = SvIV (p_sv); // code below can override
314
315 switch (t)
316 {
317 case CL_CONTEXT_PLATFORM:
318 if (SvROK (p_sv))
319 v = (cl_context_properties)SvCLOBJ (cv, svname, p_sv, "OpenCL::Platform");
320 break;
321
322 case CL_GLX_DISPLAY_KHR:
323 if (!SvOK (p_sv))
324 {
325 void *func = glsym ("_glXGetCurrentDisplay");
326 if (func)
327 v = (cl_context_properties)((void *(*)(void))func)();
328 }
329 break;
330
331 case CL_GL_CONTEXT_KHR:
332 if (!SvOK (p_sv))
333 {
334 void *func = glsym ("_glXGetCurrentContext");
335 if (func)
336 v = (cl_context_properties)((void *(*)(void))func)();
337 }
338 break;
339
340 default:
341 /* unknown property, treat as int */
342 break;
343 }
344
345 *l++ = t;
346 *l++ = v;
347 }
348
349 *l = 0;
350
351 return p;
352 }
353
354 croak ("%s: %s is not a property list (either undef or [type => value, ...])", cv_get_name (cv), svname);
355}
356
357// parse an array of CLOBJ into a void ** array in C - works only for CLOBJs whose representation
358// is a pointer (and only on well-behaved systems).
359static void * ecb_noinline
360object_list (CV *cv, int or_undef, const char *argname, SV *arg, const char *klass, cl_uint *rcount)
361{
362 if (!SvROK (arg) || SvTYPE (SvRV (arg)) != SVt_PVAV)
363 croak ("%s: '%s' parameter must be %sa reference to an array of %s objects",
364 cv_get_name (cv), argname, or_undef ? "undef or " : "", klass);
365
366 AV *av = (AV *)SvRV (arg);
367 void **list = 0;
368 cl_uint count = av_len (av) + 1;
369
370 if (count)
371 {
372 list = tmpbuf (sizeof (*list) * count);
373 int i;
374 for (i = 0; i < count; ++i)
375 list [i] = SvCLOBJ (cv, argname, *av_fetch (av, i, 1), klass);
376 }
377
378 if (!count && !or_undef)
379 croak ("%s: '%s' must contain at least one %s object",
380 cv_get_name (cv), argname, klass);
381
382 *rcount = count;
383 return (void *)list;
384}
385
386/*****************************************************************************/
387/* callback stuff */
388
389/* default context callback, log to stderr */
390static void CL_CALLBACK
391context_default_notify (const char *msg, const void *info, size_t cb, void *data)
392{
393 fprintf (stderr, "OpenCL Context Notify: %s\n", msg);
394}
395
396typedef struct
397{
398 int free_cb;
399 void (*push)(void *data1, void *data2, void *data3);
400} eq_vtbl;
401
402typedef struct eq_item
403{
404 struct eq_item *next;
405 eq_vtbl *vtbl;
406 SV *cb;
407 void *data1, *data2, *data3;
408} eq_item;
409
410static void (*eq_signal_func)(void *signal_arg, int value);
411static void *eq_signal_arg;
412static xmutex_t eq_lock = X_MUTEX_INIT;
413static eq_item *eq_head, *eq_tail;
414
415static void ecb_noinline
416eq_enq (eq_vtbl *vtbl, SV *cb, void *data1, void *data2, void *data3)
417{
418 eq_item *item = malloc (sizeof (eq_item));
419
420 item->next = 0;
421 item->vtbl = vtbl;
422 item->cb = cb;
423 item->data1 = data1;
424 item->data2 = data2;
425 item->data3 = data3;
426
427 X_LOCK (eq_lock);
428
429 *(eq_head ? &eq_tail->next : &eq_head) = item;
430 eq_tail = item;
431
432 X_UNLOCK (eq_lock);
433
434 eq_signal_func (eq_signal_arg, 0);
435}
436
437static eq_item *
438eq_dec (void)
439{
440 eq_item *res;
441
442 X_LOCK (eq_lock);
443
444 res = eq_head;
445 if (res)
446 eq_head = res->next;
447
448 X_UNLOCK (eq_lock);
449
450 return res;
451}
452
453static void
454eq_poll (void)
455{
456 eq_item *item;
457
458 while ((item = eq_dec ()))
459 {
460 ENTER;
461 SAVETMPS;
462
463 dSP;
464 PUSHMARK (SP);
465 EXTEND (SP, 2);
466
467 if (item->vtbl->free_cb)
468 sv_2mortal (item->cb);
469
470 PUTBACK;
471 item->vtbl->push (item->data1, item->data2, item->data3);
472
473 SV *cb = item->cb;
474 free (item);
475
476 call_sv (cb, G_DISCARD | G_VOID);
477
478 FREETMPS;
479 LEAVE;
480 }
481}
482
483static void
484eq_poll_interrupt (pTHX_ void *c_arg, int value)
485{
486 eq_poll ();
487}
488
489/*****************************************************************************/
490/* context notify */
491
492static void ecb_noinline
493eq_context_push (void *data1, void *data2, void *data3)
494{
495 dSP;
496 PUSHs (sv_2mortal (newSVpv (data1, 0)));
497 PUSHs (sv_2mortal (newSVpvn (data2, (STRLEN)data3)));
498 PUTBACK;
499
500 free (data1);
501 free (data2);
502}
503
504static eq_vtbl eq_context_vtbl = { 0, eq_context_push };
505
506static void CL_CALLBACK
507eq_context_notify (const char *msg, const void *pvt, size_t cb, void *user_data)
508{
509 void *pvt_copy = malloc (cb);
510 memcpy (pvt_copy, pvt, cb);
511 eq_enq (&eq_context_vtbl, user_data, strdup (msg), pvt_copy, (void *)cb);
512}
513
514#define CONTEXT_NOTIFY_CALLBACK \
515 void (CL_CALLBACK *pfn_notify)(const char *, const void *, size_t, void *) = context_default_notify; \
516 void *user_data = 0; \
517 \
518 if (SvOK (notify)) \
519 { \
520 pfn_notify = eq_context_notify; \
521 user_data = s_get_cv (notify); \
522 }
523
524static SV * ecb_noinline
525new_clobj_context (cl_context ctx, void *user_data)
526{
527 SV *sv = new_clobj (stash_context, (IV)ctx);
528
529 if (user_data)
530 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0);
531
532 return sv;
533}
534
535#define XPUSH_CLOBJ_CONTEXT XPUSHs (new_clobj_context (ctx, user_data));
536
537/*****************************************************************************/
538/* build/compile/link notify */
539
540static void
541eq_program_push (void *data1, void *data2, void *data3)
542{
543 dSP;
544 PUSH_CLOBJ (stash_program, data1);
545 PUTBACK;
546}
547
548static eq_vtbl eq_program_vtbl = { 1, eq_program_push };
549
550static void CL_CALLBACK
551eq_program_notify (cl_program program, void *user_data)
552{
553 clRetainProgram (program);
554
555 eq_enq (&eq_program_vtbl, user_data, (void *)program, 0, 0);
556}
557
558typedef void (CL_CALLBACK *program_callback)(cl_program program, void *user_data);
559
560static program_callback ecb_noinline
561make_program_callback (SV *notify, void **ruser_data)
562{
563 if (SvOK (notify))
564 {
565 *ruser_data = SvREFCNT_inc (s_get_cv (notify));
566 return eq_program_notify;
567 }
568 else
569 {
570 *ruser_data = 0;
571 return 0;
572 }
573}
574
575struct build_args
576{
577 cl_program program;
578 char *options;
579 void *user_data;
580 cl_uint num_devices;
581};
582
583X_THREAD_PROC (build_program_thread)
584{
585 struct build_args *arg = thr_arg;
586
587 clBuildProgram (arg->program, arg->num_devices, arg->num_devices ? (void *)(arg + 1) : 0, arg->options, 0, 0);
588
589 if (arg->user_data)
590 eq_program_notify (arg->program, arg->user_data);
591 else
592 clReleaseProgram (arg->program);
593
594 free (arg->options);
595 free (arg);
596
597 return 0;
598}
599
600static void
601build_program_async (cl_program program, cl_uint num_devices, const cl_device_id *device_list, const char *options, void *user_data)
602{
603 struct build_args *arg = malloc (sizeof (struct build_args) + sizeof (*device_list) * num_devices);
604
605 arg->program = program;
606 arg->options = strdup (options);
607 arg->user_data = user_data;
608 arg->num_devices = num_devices;
609 memcpy (arg + 1, device_list, sizeof (*device_list) * num_devices);
610
611 xthread_t id;
612 thread_create (&id, build_program_thread, arg);
613}
614
615/*****************************************************************************/
616/* event objects */
617
618static void
619eq_event_push (void *data1, void *data2, void *data3)
620{
621 dSP;
622 PUSH_CLOBJ (stash_event, data1);
623 PUSHs (sv_2mortal (newSViv ((IV)data2)));
624 PUTBACK;
625}
626
627static eq_vtbl eq_event_vtbl = { 1, eq_event_push };
628
629static void CL_CALLBACK
630eq_event_notify (cl_event event, cl_int event_command_exec_status, void *user_data)
631{
632 clRetainEvent (event);
633 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0);
634}
635
636/*****************************************************************************/
637/* utilities for XS code */
130 638
131static size_t 639static size_t
132img_row_pitch (cl_mem img) 640img_row_pitch (cl_mem img)
133{ 641{
134 size_t res; 642 size_t res;
135 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0); 643 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0);
136 return res; 644 return res;
137} 645}
138 646
139static cl_event * 647static cl_event * ecb_noinline
140event_list (SV **items, int count) 648event_list (CV *cv, SV **items, cl_uint *rcount, cl_event extra)
141{ 649{
650 cl_uint count = *rcount;
651
652 if (count > 0x7fffffffU) // yeah, it's a hack - the caller might have underflowed
653 *rcount = count = 0;
654
655 if (!count && !extra)
656 return 0;
657
142 cl_event *list = tmpbuf (sizeof (cl_event) * count); 658 cl_event *list = tmpbuf (sizeof (cl_event) * (count + 1));
659 int i = 0;
143 660
144 while (count--) 661 while (count--)
662 if (SvOK (items [count]))
145 list [count] = SvPTROBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event"); 663 list [i++] = SvCLOBJ (cv, "wait_events", items [count], "OpenCL::Event");
146 664
665 if (extra)
666 list [i++] = extra;
667
668 *rcount = i;
669
147 return list; 670 return i ? list : 0;
148} 671}
149 672
150#define EVENT_LIST(items,count) \ 673#define EVENT_LIST(skip) \
151 cl_uint event_list_count = (count); \ 674 cl_uint event_list_count = items - (skip); \
152 cl_event *event_list_ptr = event_list (&ST (items), event_list_count) 675 cl_event *event_list_ptr = event_list (cv, &ST (skip), &event_list_count, 0)
153 676
154#define INFO(class) \ 677#define INFO(class) \
155{ \ 678{ \
156 size_t size; \ 679 size_t size; \
157 NEED_SUCCESS (Get ## class ## Info, (this, name, 0, 0, &size)); \ 680 NEED_SUCCESS (Get ## class ## Info, (self, name, 0, 0, &size)); \
158 SV *sv = sv_2mortal (newSV (size)); \ 681 SV *sv = sv_2mortal (newSV (size)); \
159 SvUPGRADE (sv, SVt_PV); \ 682 SvUPGRADE (sv, SVt_PV); \
160 SvPOK_only (sv); \ 683 SvPOK_only (sv); \
161 SvCUR_set (sv, size); \ 684 SvCUR_set (sv, size); \
162 NEED_SUCCESS (Get ## class ## Info, (this, name, size, SvPVX (sv), 0)); \ 685 NEED_SUCCESS (Get ## class ## Info, (self, name, size, SvPVX (sv), 0)); \
163 XPUSHs (sv); \ 686 XPUSHs (sv); \
164} 687}
165 688
689/*****************************************************************************/
690/* mapped_xxx */
691
692static OpenCL__Mapped
693SvMAPPED (SV *sv)
694{
695 // no typechecking atm., keep your fingers crossed
696 return (OpenCL__Mapped)SvMAGIC (SvRV (sv))->mg_ptr;
697}
698
699struct mapped
700{
701 cl_command_queue queue;
702 cl_mem memobj;
703 void *ptr;
704 size_t cb;
705 cl_event event;
706 size_t row_pitch;
707 size_t slice_pitch;
708};
709
710static SV *
711mapped_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)
712{
713 SV *data = newSV (0);
714 SvUPGRADE (data, SVt_PVMG);
715
716 OpenCL__Mapped mapped;
717 New (0, mapped, 1, struct mapped);
718
719 clRetainCommandQueue (queue);
720
721 mapped->queue = queue;
722 mapped->memobj = memobj;
723 mapped->ptr = ptr;
724 mapped->cb = cb;
725 mapped->event = ev;
726 mapped->row_pitch = row_pitch;
727 mapped->slice_pitch = slice_pitch;
728
729 sv_magicext (data, 0, PERL_MAGIC_ext, 0, (char *)mapped, 0);
730
731 if (SvLEN (data))
732 Safefree (data);
733
734 SvPVX (data) = (char *)ptr;
735 SvCUR_set (data, cb);
736 SvLEN_set (data, 0);
737 SvPOK_only (data);
738
739 SV *obj = sv_2mortal (sv_bless (newRV_noinc (data), stash));
740
741 if (!(flags & CL_MAP_WRITE))
742 SvREADONLY_on (data);
743
744 return obj;
745}
746
747static void
748mapped_detach (SV *sv, OpenCL__Mapped mapped)
749{
750 SV *data = SvRV (sv);
751
752 // the next check checks both before AND after detach, where SvPVX should STILL be 0
753 if (SvPVX (data) != (char *)mapped->ptr)
754 warn ("FATAL: OpenCL memory mapped scalar changed location, detected");
755 else
756 {
757 SvREADONLY_off (data);
758 SvCUR_set (data, 0);
759 SvPVX (data) = 0;
760 SvOK_off (data);
761 }
762
763 mapped->ptr = 0;
764}
765
766static void
767mapped_unmap (CV *cv, SV *self, OpenCL__Mapped mapped, cl_command_queue queue, SV **wait_list, cl_uint event_list_count)
768{
769 cl_event *event_list_ptr = event_list (cv, wait_list, &event_list_count, mapped->event);
770 cl_event ev;
771
772 NEED_SUCCESS (EnqueueUnmapMemObject, (queue, mapped->memobj, mapped->ptr, event_list_count, event_list_ptr, &ev));
773
774 clReleaseEvent (mapped->event);
775 mapped->event = ev;
776
777 mapped_detach (self, mapped);
778}
779
780/*****************************************************************************/
781
166MODULE = OpenCL PACKAGE = OpenCL 782MODULE = OpenCL PACKAGE = OpenCL
167 783
168PROTOTYPES: ENABLE 784PROTOTYPES: ENABLE
169 785
786void
787poll ()
788 CODE:
789 eq_poll ();
790
791void
792_eq_initialise (IV func, IV arg)
793 CODE:
794 eq_signal_func = (void (*)(void *, int))func;
795 eq_signal_arg = (void*)arg;
796
170BOOT: 797BOOT:
171{ 798{
172 HV *stash = gv_stashpv ("OpenCL", 1); 799 HV *stash = gv_stashpv ("OpenCL", 1);
173 static const ivstr *civ, const_iv[] = { 800 static const ivstr *civ, const_iv[] = {
174 { sizeof (cl_char ), "SIZEOF_CHAR" }, 801 { sizeof (cl_char ), "SIZEOF_CHAR" },
175 { sizeof (cl_uchar ), "SIZEOF_UCHAR" }, 802 { sizeof (cl_uchar ), "SIZEOF_UCHAR" },
176 { sizeof (cl_short ), "SIZEOF_SHORT" }, 803 { sizeof (cl_short ), "SIZEOF_SHORT" },
177 { sizeof (cl_ushort), "SIZEOF_USHORT" }, 804 { sizeof (cl_ushort), "SIZEOF_USHORT" },
178 { sizeof (cl_int ), "SIZEOF_INT" }, 805 { sizeof (cl_int ), "SIZEOF_INT" },
179 { sizeof (cl_uint ), "SIZEOF_UINT" }, 806 { sizeof (cl_uint ), "SIZEOF_UINT" },
180 { sizeof (cl_long ), "SIZEOF_LONG" }, 807 { sizeof (cl_long ), "SIZEOF_LONG" },
181 { sizeof (cl_ulong ), "SIZEOF_ULONG" }, 808 { sizeof (cl_ulong ), "SIZEOF_ULONG" },
182 { sizeof (cl_half ), "SIZEOF_HALF" }, 809 { sizeof (cl_half ), "SIZEOF_HALF" },
183 { sizeof (cl_float ), "SIZEOF_FLOAT" }, 810 { sizeof (cl_float ), "SIZEOF_FLOAT" },
184 { sizeof (cl_double), "SIZEOF_DOUBLE" }, 811 { sizeof (cl_double), "SIZEOF_DOUBLE" },
812 { PREFER_1_1 , "PREFER_1_1" },
185#include "constiv.h" 813#include "constiv.h"
186 }; 814 };
815
187 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--) 816 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)); 817 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv));
818
819 stash_platform = gv_stashpv ("OpenCL::Platform", GV_ADD);
820 stash_device = gv_stashpv ("OpenCL::Device", GV_ADD);
821 stash_subdevice = gv_stashpv ("OpenCL::SubDevice", GV_ADD);
822 stash_context = gv_stashpv ("OpenCL::Context", GV_ADD);
823 stash_queue = gv_stashpv ("OpenCL::Queue", GV_ADD);
824 stash_program = gv_stashpv ("OpenCL::Program", GV_ADD);
825 stash_kernel = gv_stashpv ("OpenCL::Kernel", GV_ADD);
826 stash_sampler = gv_stashpv ("OpenCL::Sampler", GV_ADD);
827 stash_event = gv_stashpv ("OpenCL::Event", GV_ADD);
828 stash_userevent = gv_stashpv ("OpenCL::UserEvent", GV_ADD);
829 stash_memory = gv_stashpv ("OpenCL::Memory", GV_ADD);
830 stash_buffer = gv_stashpv ("OpenCL::Buffer", GV_ADD);
831 stash_bufferobj = gv_stashpv ("OpenCL::BufferObj", GV_ADD);
832 stash_image = gv_stashpv ("OpenCL::Image", GV_ADD);
833 stash_image1d = gv_stashpv ("OpenCL::Image1D", GV_ADD);
834 stash_image1darray = gv_stashpv ("OpenCL::Image1DArray", GV_ADD);
835 stash_image1dbuffer = gv_stashpv ("OpenCL::Image1DBuffer", GV_ADD);
836 stash_image2d = gv_stashpv ("OpenCL::Image2D", GV_ADD);
837 stash_image2darray = gv_stashpv ("OpenCL::Image2DArray", GV_ADD);
838 stash_image3d = gv_stashpv ("OpenCL::Image3D", GV_ADD);
839 stash_mapped = gv_stashpv ("OpenCL::Mapped", GV_ADD);
840 stash_mappedbuffer = gv_stashpv ("OpenCL::MappedBuffer", GV_ADD);
841 stash_mappedimage = gv_stashpv ("OpenCL::MappedImage", GV_ADD);
842
843 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt);
189} 844}
190 845
191cl_int 846cl_int
192errno () 847errno ()
193 CODE: 848 CODE:
194 errno = res; 849 RETVAL = res;
850 OUTPUT:
851 RETVAL
195 852
196const char * 853const char *
197err2str (cl_int err) 854err2str (cl_int err = res)
198 855
199const char * 856const char *
200enum2str (cl_uint value) 857enum2str (cl_uint value)
201 858
202void 859void
210 list = tmpbuf (sizeof (*list) * count); 867 list = tmpbuf (sizeof (*list) * count);
211 NEED_SUCCESS (GetPlatformIDs, (count, list, 0)); 868 NEED_SUCCESS (GetPlatformIDs, (count, list, 0));
212 869
213 EXTEND (SP, count); 870 EXTEND (SP, count);
214 for (i = 0; i < count; ++i) 871 for (i = 0; i < count; ++i)
215 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Platform", list [i])); 872 PUSH_CLOBJ (stash_platform, list [i]);
216 873
217void 874void
218context_from_type (FUTURE properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, FUTURE notify = 0) 875context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
219 PPCODE: 876 PPCODE:
877 CONTEXT_NOTIFY_CALLBACK;
220 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (0, type, 0, 0, &res)); 878 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, pfn_notify, user_data, &res));
221 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 879 XPUSH_CLOBJ_CONTEXT;
222 880
223void 881void
224context (FUTURE properties, FUTURE devices, FUTURE notify = 0) 882context (cl_context_properties *properties, SV *devices, SV *notify = &PL_sv_undef)
225 PPCODE: 883 PPCODE:
226 /* der Gipfel der Kunst */ 884 cl_uint device_count;
885 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
886
887 CONTEXT_NOTIFY_CALLBACK;
888 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (properties, device_count, device_list, pfn_notify, user_data, &res));
889 XPUSH_CLOBJ_CONTEXT;
227 890
228void 891void
229wait_for_events (...) 892wait_for_events (...)
230 CODE: 893 CODE:
231 EVENT_LIST (0, items); 894 EVENT_LIST (0);
232 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr)); 895 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr));
233 896
234PROTOTYPES: DISABLE 897PROTOTYPES: DISABLE
235 898
236MODULE = OpenCL PACKAGE = OpenCL::Platform 899MODULE = OpenCL PACKAGE = OpenCL::Platform
237 900
238void 901void
239info (OpenCL::Platform this, cl_platform_info name) 902info (OpenCL::Platform self, cl_platform_info name)
240 PPCODE: 903 PPCODE:
241 INFO (Platform) 904 INFO (Platform)
242 905
906void
907unload_compiler (OpenCL::Platform self)
908 CODE:
909#if CL_VERSION_1_2
910 clUnloadPlatformCompiler (self);
911#endif
912
243#BEGIN:platform 913#BEGIN:platform
244 914
245void 915void
246profile (OpenCL::Platform this) 916profile (OpenCL::Platform self)
247 ALIAS: 917 ALIAS:
248 profile = CL_PLATFORM_PROFILE 918 profile = CL_PLATFORM_PROFILE
249 version = CL_PLATFORM_VERSION 919 version = CL_PLATFORM_VERSION
250 name = CL_PLATFORM_NAME 920 name = CL_PLATFORM_NAME
251 vendor = CL_PLATFORM_VENDOR 921 vendor = CL_PLATFORM_VENDOR
252 extensions = CL_PLATFORM_EXTENSIONS 922 extensions = CL_PLATFORM_EXTENSIONS
253 PPCODE: 923 PPCODE:
254 size_t size; 924 size_t size;
255 NEED_SUCCESS (GetPlatformInfo, (this, ix, 0, 0, &size)); 925 NEED_SUCCESS (GetPlatformInfo, (self, ix, 0, 0, &size));
256 char *value = tmpbuf (size); 926 char *value = tmpbuf (size);
257 NEED_SUCCESS (GetPlatformInfo, (this, ix, size, value, 0)); 927 NEED_SUCCESS (GetPlatformInfo, (self, ix, size, value, 0));
258 EXTEND (SP, 1); 928 EXTEND (SP, 1);
259 const int i = 0; 929 const int i = 0;
260 PUSHs (sv_2mortal (newSVpv (value, 0))); 930 PUSHs (sv_2mortal (newSVpv (value, 0)));
261 931
262#END:platform 932#END:platform
263 933
264void 934void
265devices (OpenCL::Platform this, cl_device_type type = CL_DEVICE_TYPE_ALL) 935devices (OpenCL::Platform self, cl_device_type type = CL_DEVICE_TYPE_ALL)
266 PPCODE: 936 PPCODE:
267 cl_device_id *list; 937 cl_device_id *list;
268 cl_uint count; 938 cl_uint count;
269 int i; 939 int i;
270 940
271 NEED_SUCCESS (GetDeviceIDs, (this, type, 0, 0, &count)); 941 NEED_SUCCESS (GetDeviceIDs, (self, type, 0, 0, &count));
272 list = tmpbuf (sizeof (*list) * count); 942 list = tmpbuf (sizeof (*list) * count);
273 NEED_SUCCESS (GetDeviceIDs, (this, type, count, list, 0)); 943 NEED_SUCCESS (GetDeviceIDs, (self, type, count, list, 0));
274 944
275 EXTEND (SP, count); 945 EXTEND (SP, count);
276 for (i = 0; i < count; ++i) 946 for (i = 0; i < count; ++i)
277 PUSHs (sv_setref_pv (sv_newmortal (), "OpenCL::Device", list [i])); 947 PUSH_CLOBJ (stash_device, list [i]);
278 948
279void 949void
280context (OpenCL::Platform this, FUTURE properties, SV *devices, FUTURE notify = 0) 950context (OpenCL::Platform self, SV *properties, SV *devices, SV *notify = &PL_sv_undef)
281 PPCODE: 951 PPCODE:
952 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
953 cl_context_properties *props = SvCONTEXTPROPERTIES (cv, "properties", properties, extra, 2);
954
955 cl_uint device_count;
956 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
957
958 CONTEXT_NOTIFY_CALLBACK;
959 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (props, device_count, device_list, pfn_notify, user_data, &res));
960 XPUSH_CLOBJ_CONTEXT;
961
962void
963context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
964 PPCODE:
965 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
966 cl_context_properties *props = SvCONTEXTPROPERTIES (cv, "properties", properties, extra, 2);
967
968 CONTEXT_NOTIFY_CALLBACK;
969 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, pfn_notify, user_data, &res));
970 XPUSH_CLOBJ_CONTEXT;
971
972MODULE = OpenCL PACKAGE = OpenCL::Device
973
974void
975info (OpenCL::Device self, cl_device_info name)
976 PPCODE:
977 INFO (Device)
978
979#if CL_VERSION_1_2
980
981void
982sub_devices (OpenCL::Device self, SV *properties)
983 PPCODE:
282 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV) 984 if (!SvROK (properties) || SvTYPE (SvRV (properties)) != SVt_PVAV)
283 croak ("OpenCL::Platform argument 'device' must be an arrayref with device objects, in call"); 985 croak ("OpenCL::Device::sub_devices: properties must be specified as reference to an array of property-value pairs");
284 986
285 AV *av = (AV *)SvRV (devices); 987 properties = SvRV (properties);
286 cl_uint num_devices = av_len (av) + 1; 988
287 cl_device_id *device_list = tmpbuf (sizeof (cl_device_id) * num_devices); 989 cl_uint count = av_len ((AV *)properties) + 1;
990 cl_device_partition_property *props = tmpbuf (sizeof (*props) * count + 1);
991
288 int i; 992 int i;
993 for (i = 0; i < count; ++i)
994 props [i] = (cl_device_partition_property)SvIV (*av_fetch ((AV *)properties, i, 0));
289 995
290 for (i = num_devices; i--; ) 996 props [count] = 0;
291 device_list [i] = SvPTROBJ ("clCreateContext", "devices", *av_fetch (av, i, 0), "OpenCL::Device");
292 997
293 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (0, num_devices, device_list, 0, 0, &res)); 998 cl_uint num_devices;
294 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 999 NEED_SUCCESS (CreateSubDevices, (self, props, 0, 0, &num_devices));
1000 cl_device_id *list = tmpbuf (sizeof (*list) * num_devices);
1001 NEED_SUCCESS (CreateSubDevices, (self, props, num_devices, list, 0));
295 1002
296void 1003 EXTEND (SP, num_devices);
297context_from_type (OpenCL::Platform this, FUTURE properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, FUTURE notify = 0) 1004 for (i = 0; i < count; ++i)
298 PPCODE: 1005 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 1006
303MODULE = OpenCL PACKAGE = OpenCL::Device 1007#endif
304
305void
306info (OpenCL::Device this, cl_device_info name)
307 PPCODE:
308 INFO (Device)
309 1008
310#BEGIN:device 1009#BEGIN:device
311 1010
312void 1011void
313type (OpenCL::Device this) 1012type (OpenCL::Device self)
314 PPCODE: 1013 PPCODE:
315 cl_device_type value [1]; 1014 cl_device_type value [1];
316 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_TYPE, sizeof (value), value, 0)); 1015 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_TYPE, sizeof (value), value, 0));
317 EXTEND (SP, 1); 1016 EXTEND (SP, 1);
318 const int i = 0; 1017 const int i = 0;
319 PUSHs (sv_2mortal (newSViv (value [i]))); 1018 PUSHs (sv_2mortal (newSVuv (value [i])));
320 1019
321void 1020void
322vendor_id (OpenCL::Device this) 1021vendor_id (OpenCL::Device self)
323 ALIAS: 1022 ALIAS:
324 vendor_id = CL_DEVICE_VENDOR_ID 1023 vendor_id = CL_DEVICE_VENDOR_ID
325 max_compute_units = CL_DEVICE_MAX_COMPUTE_UNITS 1024 max_compute_units = CL_DEVICE_MAX_COMPUTE_UNITS
326 max_work_item_dimensions = CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS 1025 max_work_item_dimensions = CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS
327 preferred_vector_width_char = CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR 1026 preferred_vector_width_char = CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR
345 native_vector_width_int = CL_DEVICE_NATIVE_VECTOR_WIDTH_INT 1044 native_vector_width_int = CL_DEVICE_NATIVE_VECTOR_WIDTH_INT
346 native_vector_width_long = CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG 1045 native_vector_width_long = CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG
347 native_vector_width_float = CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT 1046 native_vector_width_float = CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT
348 native_vector_width_double = CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE 1047 native_vector_width_double = CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE
349 native_vector_width_half = CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF 1048 native_vector_width_half = CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF
350 reference_count_ext = CL_DEVICE_REFERENCE_COUNT_EXT 1049 reference_count_ext = CL_DEVICE_REFERENCE_COUNT_EXT
351 PPCODE: 1050 PPCODE:
352 cl_uint value [1]; 1051 cl_uint value [1];
353 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1052 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
354 EXTEND (SP, 1); 1053 EXTEND (SP, 1);
355 const int i = 0; 1054 const int i = 0;
356 PUSHs (sv_2mortal (newSVuv (value [i]))); 1055 PUSHs (sv_2mortal (newSVuv (value [i])));
357 1056
358void 1057void
359max_work_group_size (OpenCL::Device this) 1058max_work_group_size (OpenCL::Device self)
360 ALIAS: 1059 ALIAS:
361 max_work_group_size = CL_DEVICE_MAX_WORK_GROUP_SIZE 1060 max_work_group_size = CL_DEVICE_MAX_WORK_GROUP_SIZE
362 image2d_max_width = CL_DEVICE_IMAGE2D_MAX_WIDTH 1061 image2d_max_width = CL_DEVICE_IMAGE2D_MAX_WIDTH
363 image2d_max_height = CL_DEVICE_IMAGE2D_MAX_HEIGHT 1062 image2d_max_height = CL_DEVICE_IMAGE2D_MAX_HEIGHT
364 image3d_max_width = CL_DEVICE_IMAGE3D_MAX_WIDTH 1063 image3d_max_width = CL_DEVICE_IMAGE3D_MAX_WIDTH
366 image3d_max_depth = CL_DEVICE_IMAGE3D_MAX_DEPTH 1065 image3d_max_depth = CL_DEVICE_IMAGE3D_MAX_DEPTH
367 max_parameter_size = CL_DEVICE_MAX_PARAMETER_SIZE 1066 max_parameter_size = CL_DEVICE_MAX_PARAMETER_SIZE
368 profiling_timer_resolution = CL_DEVICE_PROFILING_TIMER_RESOLUTION 1067 profiling_timer_resolution = CL_DEVICE_PROFILING_TIMER_RESOLUTION
369 PPCODE: 1068 PPCODE:
370 size_t value [1]; 1069 size_t value [1];
371 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1070 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
372 EXTEND (SP, 1); 1071 EXTEND (SP, 1);
373 const int i = 0; 1072 const int i = 0;
374 PUSHs (sv_2mortal (newSVuv (value [i]))); 1073 PUSHs (sv_2mortal (newSVuv (value [i])));
375 1074
376void 1075void
377max_work_item_sizes (OpenCL::Device this) 1076max_work_item_sizes (OpenCL::Device self)
378 PPCODE: 1077 PPCODE:
379 size_t size; 1078 size_t size;
380 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_MAX_WORK_ITEM_SIZES, 0, 0, &size)); 1079 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_MAX_WORK_ITEM_SIZES, 0, 0, &size));
381 size_t *value = tmpbuf (size); 1080 size_t *value = tmpbuf (size);
382 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_MAX_WORK_ITEM_SIZES, size, value, 0)); 1081 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_MAX_WORK_ITEM_SIZES, size, value, 0));
383 int i, n = size / sizeof (*value); 1082 int i, n = size / sizeof (*value);
384 EXTEND (SP, n); 1083 EXTEND (SP, n);
385 for (i = 0; i < n; ++i) 1084 for (i = 0; i < n; ++i)
386 PUSHs (sv_2mortal (newSVuv (value [i]))); 1085 PUSHs (sv_2mortal (newSVuv (value [i])));
387 1086
388void 1087void
389address_bits (OpenCL::Device this) 1088address_bits (OpenCL::Device self)
390 PPCODE: 1089 PPCODE:
391 cl_bitfield value [1]; 1090 cl_bitfield value [1];
392 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_ADDRESS_BITS, sizeof (value), value, 0)); 1091 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_ADDRESS_BITS, sizeof (value), value, 0));
393 EXTEND (SP, 1); 1092 EXTEND (SP, 1);
394 const int i = 0; 1093 const int i = 0;
395 PUSHs (sv_2mortal (newSVuv (value [i]))); 1094 PUSHs (sv_2mortal (newSVuv (value [i])));
396 1095
397void 1096void
398max_mem_alloc_size (OpenCL::Device this) 1097max_mem_alloc_size (OpenCL::Device self)
399 ALIAS: 1098 ALIAS:
400 max_mem_alloc_size = CL_DEVICE_MAX_MEM_ALLOC_SIZE 1099 max_mem_alloc_size = CL_DEVICE_MAX_MEM_ALLOC_SIZE
401 global_mem_cache_size = CL_DEVICE_GLOBAL_MEM_CACHE_SIZE 1100 global_mem_cache_size = CL_DEVICE_GLOBAL_MEM_CACHE_SIZE
402 global_mem_size = CL_DEVICE_GLOBAL_MEM_SIZE 1101 global_mem_size = CL_DEVICE_GLOBAL_MEM_SIZE
403 max_constant_buffer_size = CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE 1102 max_constant_buffer_size = CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE
404 local_mem_size = CL_DEVICE_LOCAL_MEM_SIZE 1103 local_mem_size = CL_DEVICE_LOCAL_MEM_SIZE
405 PPCODE: 1104 PPCODE:
406 cl_ulong value [1]; 1105 cl_ulong value [1];
407 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1106 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
408 EXTEND (SP, 1); 1107 EXTEND (SP, 1);
409 const int i = 0; 1108 const int i = 0;
410 PUSHs (sv_2mortal (newSVuv (value [i]))); 1109 PUSHs (sv_2mortal (newSVuv (value [i])));
411 1110
412void 1111void
413single_fp_config (OpenCL::Device this) 1112single_fp_config (OpenCL::Device self)
414 ALIAS: 1113 ALIAS:
415 single_fp_config = CL_DEVICE_SINGLE_FP_CONFIG 1114 single_fp_config = CL_DEVICE_SINGLE_FP_CONFIG
416 double_fp_config = CL_DEVICE_DOUBLE_FP_CONFIG 1115 double_fp_config = CL_DEVICE_DOUBLE_FP_CONFIG
417 half_fp_config = CL_DEVICE_HALF_FP_CONFIG 1116 half_fp_config = CL_DEVICE_HALF_FP_CONFIG
418 PPCODE: 1117 PPCODE:
419 cl_device_fp_config value [1]; 1118 cl_device_fp_config value [1];
420 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1119 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
421 EXTEND (SP, 1); 1120 EXTEND (SP, 1);
422 const int i = 0; 1121 const int i = 0;
423 PUSHs (sv_2mortal (newSVuv (value [i]))); 1122 PUSHs (sv_2mortal (newSVuv (value [i])));
424 1123
425void 1124void
426global_mem_cache_type (OpenCL::Device this) 1125global_mem_cache_type (OpenCL::Device self)
427 PPCODE: 1126 PPCODE:
428 cl_device_mem_cache_type value [1]; 1127 cl_device_mem_cache_type value [1];
429 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, sizeof (value), value, 0)); 1128 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, sizeof (value), value, 0));
430 EXTEND (SP, 1); 1129 EXTEND (SP, 1);
431 const int i = 0; 1130 const int i = 0;
432 PUSHs (sv_2mortal (newSVuv (value [i]))); 1131 PUSHs (sv_2mortal (newSVuv (value [i])));
433 1132
434void 1133void
435local_mem_type (OpenCL::Device this) 1134local_mem_type (OpenCL::Device self)
436 PPCODE: 1135 PPCODE:
437 cl_device_local_mem_type value [1]; 1136 cl_device_local_mem_type value [1];
438 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_LOCAL_MEM_TYPE, sizeof (value), value, 0)); 1137 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_LOCAL_MEM_TYPE, sizeof (value), value, 0));
439 EXTEND (SP, 1); 1138 EXTEND (SP, 1);
440 const int i = 0; 1139 const int i = 0;
441 PUSHs (sv_2mortal (newSVuv (value [i]))); 1140 PUSHs (sv_2mortal (newSVuv (value [i])));
442 1141
443void 1142void
444error_correction_support (OpenCL::Device this) 1143error_correction_support (OpenCL::Device self)
445 ALIAS: 1144 ALIAS:
446 error_correction_support = CL_DEVICE_ERROR_CORRECTION_SUPPORT 1145 error_correction_support = CL_DEVICE_ERROR_CORRECTION_SUPPORT
447 endian_little = CL_DEVICE_ENDIAN_LITTLE 1146 endian_little = CL_DEVICE_ENDIAN_LITTLE
448 available = CL_DEVICE_AVAILABLE 1147 available = CL_DEVICE_AVAILABLE
449 compiler_available = CL_DEVICE_COMPILER_AVAILABLE 1148 compiler_available = CL_DEVICE_COMPILER_AVAILABLE
450 host_unified_memory = CL_DEVICE_HOST_UNIFIED_MEMORY 1149 host_unified_memory = CL_DEVICE_HOST_UNIFIED_MEMORY
451 PPCODE: 1150 PPCODE:
452 cl_bool value [1]; 1151 cl_bool value [1];
453 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1152 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
454 EXTEND (SP, 1); 1153 EXTEND (SP, 1);
455 const int i = 0; 1154 const int i = 0;
456 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no)); 1155 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no));
457 1156
458void 1157void
459execution_capabilities (OpenCL::Device this) 1158execution_capabilities (OpenCL::Device self)
460 PPCODE: 1159 PPCODE:
461 cl_device_exec_capabilities value [1]; 1160 cl_device_exec_capabilities value [1];
462 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_EXECUTION_CAPABILITIES, sizeof (value), value, 0)); 1161 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_EXECUTION_CAPABILITIES, sizeof (value), value, 0));
463 EXTEND (SP, 1); 1162 EXTEND (SP, 1);
464 const int i = 0; 1163 const int i = 0;
465 PUSHs (sv_2mortal (newSVuv (value [i]))); 1164 PUSHs (sv_2mortal (newSVuv (value [i])));
466 1165
467void 1166void
468properties (OpenCL::Device this) 1167properties (OpenCL::Device self)
469 PPCODE: 1168 PPCODE:
470 cl_command_queue_properties value [1]; 1169 cl_command_queue_properties value [1];
471 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0)); 1170 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0));
472 EXTEND (SP, 1); 1171 EXTEND (SP, 1);
473 const int i = 0; 1172 const int i = 0;
474 PUSHs (sv_2mortal (newSViv (value [i]))); 1173 PUSHs (sv_2mortal (newSVuv (value [i])));
475 1174
476void 1175void
477platform (OpenCL::Device this) 1176platform (OpenCL::Device self)
478 PPCODE: 1177 PPCODE:
479 cl_platform_id value [1]; 1178 cl_platform_id value [1];
480 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_PLATFORM, sizeof (value), value, 0)); 1179 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0));
481 EXTEND (SP, 1); 1180 EXTEND (SP, 1);
482 const int i = 0; 1181 const int i = 0;
483 { 1182 {
484 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Platform", value [i])); 1183 PUSH_CLOBJ (stash_platform, value [i]);
485 } 1184 }
486 1185
487void 1186void
488name (OpenCL::Device this) 1187name (OpenCL::Device self)
489 ALIAS: 1188 ALIAS:
490 name = CL_DEVICE_NAME 1189 name = CL_DEVICE_NAME
491 vendor = CL_DEVICE_VENDOR 1190 vendor = CL_DEVICE_VENDOR
492 driver_version = CL_DRIVER_VERSION 1191 driver_version = CL_DRIVER_VERSION
493 profile = CL_DEVICE_PROFILE 1192 profile = CL_DEVICE_PROFILE
494 version = CL_DEVICE_VERSION 1193 version = CL_DEVICE_VERSION
495 extensions = CL_DEVICE_EXTENSIONS 1194 extensions = CL_DEVICE_EXTENSIONS
496 PPCODE: 1195 PPCODE:
497 size_t size; 1196 size_t size;
498 NEED_SUCCESS (GetDeviceInfo, (this, ix, 0, 0, &size)); 1197 NEED_SUCCESS (GetDeviceInfo, (self, ix, 0, 0, &size));
499 char *value = tmpbuf (size); 1198 char *value = tmpbuf (size);
500 NEED_SUCCESS (GetDeviceInfo, (this, ix, size, value, 0)); 1199 NEED_SUCCESS (GetDeviceInfo, (self, ix, size, value, 0));
501 EXTEND (SP, 1); 1200 EXTEND (SP, 1);
502 const int i = 0; 1201 const int i = 0;
503 PUSHs (sv_2mortal (newSVpv (value, 0))); 1202 PUSHs (sv_2mortal (newSVpv (value, 0)));
504 1203
505void 1204void
506parent_device_ext (OpenCL::Device this) 1205parent_device_ext (OpenCL::Device self)
507 PPCODE: 1206 PPCODE:
508 cl_device_id value [1]; 1207 cl_device_id value [1];
509 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0)); 1208 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0));
510 EXTEND (SP, 1); 1209 EXTEND (SP, 1);
511 const int i = 0; 1210 const int i = 0;
512 { 1211 {
513 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 1212 PUSH_CLOBJ (stash_device, value [i]);
514 } 1213 }
515 1214
516void 1215void
517partition_types_ext (OpenCL::Device this) 1216partition_types_ext (OpenCL::Device self)
518 ALIAS: 1217 ALIAS:
519 partition_types_ext = CL_DEVICE_PARTITION_TYPES_EXT 1218 partition_types_ext = CL_DEVICE_PARTITION_TYPES_EXT
520 affinity_domains_ext = CL_DEVICE_AFFINITY_DOMAINS_EXT 1219 affinity_domains_ext = CL_DEVICE_AFFINITY_DOMAINS_EXT
521 partition_style_ext = CL_DEVICE_PARTITION_STYLE_EXT 1220 partition_style_ext = CL_DEVICE_PARTITION_STYLE_EXT
522 PPCODE: 1221 PPCODE:
523 size_t size; 1222 size_t size;
524 NEED_SUCCESS (GetDeviceInfo, (this, ix, 0, 0, &size)); 1223 NEED_SUCCESS (GetDeviceInfo, (self, ix, 0, 0, &size));
525 cl_device_partition_property_ext *value = tmpbuf (size); 1224 cl_device_partition_property_ext *value = tmpbuf (size);
526 NEED_SUCCESS (GetDeviceInfo, (this, ix, size, value, 0)); 1225 NEED_SUCCESS (GetDeviceInfo, (self, ix, size, value, 0));
527 int i, n = size / sizeof (*value); 1226 int i, n = size / sizeof (*value);
528 EXTEND (SP, n); 1227 EXTEND (SP, n);
529 for (i = 0; i < n; ++i) 1228 for (i = 0; i < n; ++i)
530 PUSHs (sv_2mortal (newSVuv (value [i]))); 1229 PUSHs (sv_2mortal (newSVuv (value [i])));
531 1230
532#END:device 1231#END:device
533 1232
1233MODULE = OpenCL PACKAGE = OpenCL::SubDevice
1234
1235#if CL_VERSION_1_2
1236
1237void
1238DESTROY (OpenCL::SubDevice self)
1239 CODE:
1240 clReleaseDevice (self);
1241
1242#endif
1243
534MODULE = OpenCL PACKAGE = OpenCL::Context 1244MODULE = OpenCL PACKAGE = OpenCL::Context
535 1245
536void 1246void
537DESTROY (OpenCL::Context context) 1247DESTROY (OpenCL::Context self)
538 CODE: 1248 CODE:
539 clReleaseContext (context); 1249 clReleaseContext (self);
540 1250
541void 1251void
542info (OpenCL::Context this, cl_context_info name) 1252info (OpenCL::Context self, cl_context_info name)
543 PPCODE: 1253 PPCODE:
544 INFO (Context) 1254 INFO (Context)
545 1255
546void 1256void
547queue (OpenCL::Context this, OpenCL::Device device, cl_command_queue_properties properties = 0) 1257queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0)
548 PPCODE: 1258 PPCODE:
549 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (this, device, properties, &res)); 1259 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res));
550 XPUSH_NEW_OBJ ("OpenCL::Queue", queue); 1260 XPUSH_CLOBJ (stash_queue, queue);
551 1261
552void 1262void
553user_event (OpenCL::Context this) 1263user_event (OpenCL::Context self)
554 PPCODE: 1264 PPCODE:
555 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (this, &res)); 1265 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res));
556 XPUSH_NEW_OBJ ("OpenCL::UserEvent", ev); 1266 XPUSH_CLOBJ (stash_userevent, ev);
557 1267
558void 1268void
559buffer (OpenCL::Context this, cl_mem_flags flags, size_t len) 1269buffer (OpenCL::Context self, cl_mem_flags flags, size_t len)
560 PPCODE: 1270 PPCODE:
561 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) 1271 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?"); 1272 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?");
563 1273
564 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (this, flags, len, 0, &res)); 1274 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res));
565 XPUSH_NEW_OBJ ("OpenCL::Buffer", mem); 1275 XPUSH_CLOBJ (stash_bufferobj, mem);
566 1276
567void 1277void
568buffer_sv (OpenCL::Context this, cl_mem_flags flags, SV *data) 1278buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data)
569 PPCODE: 1279 PPCODE:
570 STRLEN len; 1280 STRLEN len;
571 char *ptr = SvPVbyte (data, len); 1281 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
572
573 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) 1282 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?"); 1283 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)); 1284 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res));
577 XPUSH_NEW_OBJ ("OpenCL::Buffer", mem); 1285 XPUSH_CLOBJ (stash_bufferobj, mem);
578 1286
1287#if CL_VERSION_1_2
1288
579void 1289void
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) 1290image (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: 1291 PPCODE:
582 STRLEN len; 1292 STRLEN len;
583 char *ptr = SvPVbyte (data, len); 1293 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
584 const cl_image_format format = { channel_order, channel_type }; 1294 const cl_image_format format = { channel_order, channel_type };
1295 const cl_image_desc desc = {
1296 type,
1297 width, height, depth,
1298 array_size, row_pitch, slice_pitch,
1299 num_mip_level, num_samples,
1300 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ (cv, "data", data, "OpenCL::Buffer") : 0
1301 };
585 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (this, flags, &format, width, height, row_pitch, ptr, &res)); 1302 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
586 XPUSH_NEW_OBJ ("OpenCL::Image2D", mem); 1303 HV *stash = stash_image;
1304 switch (type)
1305 {
1306 case CL_MEM_OBJECT_IMAGE1D_BUFFER: stash = stash_image1dbuffer; break;
1307 case CL_MEM_OBJECT_IMAGE1D: stash = stash_image1d; break;
1308 case CL_MEM_OBJECT_IMAGE1D_ARRAY: stash = stash_image2darray; break;
1309 case CL_MEM_OBJECT_IMAGE2D: stash = stash_image2d; break;
1310 case CL_MEM_OBJECT_IMAGE2D_ARRAY: stash = stash_image2darray; break;
1311 case CL_MEM_OBJECT_IMAGE3D: stash = stash_image3d; break;
1312 }
1313 XPUSH_CLOBJ (stash, mem);
587 1314
1315#endif
1316
588void 1317void
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) 1318image2d (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: 1319 PPCODE:
591 STRLEN len; 1320 STRLEN len;
592 char *ptr = SvPVbyte (data, len); 1321 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
593 const cl_image_format format = { channel_order, channel_type }; 1322 const cl_image_format format = { channel_order, channel_type };
1323#if PREFER_1_1
1324 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res));
1325#else
1326 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, row_pitch, 0, 0, 0, 0 };
1327 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1328#endif
1329 XPUSH_CLOBJ (stash_image2d, mem);
1330
1331void
1332image3d (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)
1333 PPCODE:
1334 STRLEN len;
1335 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
1336 const cl_image_format format = { channel_order, channel_type };
1337#if PREFER_1_1
594 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (this, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res)); 1338 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); 1339#else
1340 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, row_pitch, slice_pitch, 0, 0, 0 };
1341 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1342#endif
1343 XPUSH_CLOBJ (stash_image3d, mem);
596 1344
1345#if cl_apple_gl_sharing || cl_khr_gl_sharing
1346
597void 1347void
1348gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj)
1349 PPCODE:
1350 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res));
1351 XPUSH_CLOBJ (stash_bufferobj, mem);
1352
1353void
1354gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer)
1355 PPCODE:
1356 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res));
1357 XPUSH_CLOBJ (stash_image2d, mem);
1358
1359#if CL_VERSION_1_2
1360
1361void
1362gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1363 ALIAS:
1364 PPCODE:
1365 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture, (self, flags, target, miplevel, texture, &res));
1366 cl_gl_object_type type;
1367 NEED_SUCCESS (GetGLObjectInfo, (mem, &type, 0)); // TODO: use target instead?
1368 HV *stash = stash_memory;
1369 switch (type)
1370 {
1371 case CL_GL_OBJECT_TEXTURE_BUFFER: stash = stash_image1dbuffer; break;
1372 case CL_GL_OBJECT_TEXTURE1D: stash = stash_image1d; break;
1373 case CL_GL_OBJECT_TEXTURE1D_ARRAY: stash = stash_image2darray; break;
1374 case CL_GL_OBJECT_TEXTURE2D: stash = stash_image2d; break;
1375 case CL_GL_OBJECT_TEXTURE2D_ARRAY: stash = stash_image2darray; break;
1376 case CL_GL_OBJECT_TEXTURE3D: stash = stash_image3d; break;
1377 }
1378 XPUSH_CLOBJ (stash, mem);
1379
1380#endif
1381
1382void
1383gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1384 PPCODE:
1385#if PREFER_1_1
1386 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res));
1387#else
1388 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1389#endif
1390 XPUSH_CLOBJ (stash_image2d, mem);
1391
1392void
1393gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1394 PPCODE:
1395#if PREFER_1_1
1396 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res));
1397#else
1398 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1399#endif
1400 XPUSH_CLOBJ (stash_image3d, mem);
1401
1402#endif
1403
1404void
598supported_image_formats (OpenCL::Context this, cl_mem_flags flags, cl_mem_object_type image_type) 1405supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type)
599 PPCODE: 1406 PPCODE:
600{ 1407{
601 cl_uint count; 1408 cl_uint count;
602 cl_image_format *list; 1409 cl_image_format *list;
603 int i; 1410 int i;
604 1411
605 NEED_SUCCESS (GetSupportedImageFormats, (this, flags, image_type, 0, 0, &count)); 1412 NEED_SUCCESS (GetSupportedImageFormats, (self, flags, image_type, 0, 0, &count));
606 Newx (list, count, cl_image_format); 1413 Newx (list, count, cl_image_format);
607 NEED_SUCCESS (GetSupportedImageFormats, (this, flags, image_type, count, list, 0)); 1414 NEED_SUCCESS (GetSupportedImageFormats, (self, flags, image_type, count, list, 0));
608 1415
609 EXTEND (SP, count); 1416 EXTEND (SP, count);
610 for (i = 0; i < count; ++i) 1417 for (i = 0; i < count; ++i)
611 { 1418 {
612 AV *av = newAV (); 1419 AV *av = newAV ();
615 PUSHs (sv_2mortal (newRV_noinc ((SV *)av))); 1422 PUSHs (sv_2mortal (newRV_noinc ((SV *)av)));
616 } 1423 }
617} 1424}
618 1425
619void 1426void
620sampler (OpenCL::Context this, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode) 1427sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode)
621 PPCODE: 1428 PPCODE:
622 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (this, normalized_coords, addressing_mode, filter_mode, &res)); 1429 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res));
623 XPUSH_NEW_OBJ ("OpenCL::Sampler", sampler); 1430 XPUSH_CLOBJ (stash_sampler, sampler);
624 1431
625void 1432void
626program_with_source (OpenCL::Context this, SV *program) 1433program_with_source (OpenCL::Context self, SV *program)
627 PPCODE: 1434 PPCODE:
628 STRLEN len; 1435 STRLEN len;
629 size_t len2; 1436 size_t len2;
630 const char *ptr = SvPVbyte (program, len); 1437 const char *ptr = SvPVbyte (program, len);
631 1438
632 len2 = len; 1439 len2 = len;
633 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (this, 1, &ptr, &len2, &res)); 1440 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res));
634 XPUSH_NEW_OBJ ("OpenCL::Program", prog); 1441 XPUSH_CLOBJ (stash_program, prog);
1442
1443void
1444program_with_binary (OpenCL::Context self, SV *devices, SV *binaries)
1445 PPCODE:
1446 cl_uint device_count;
1447 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
1448
1449 if (!SvROK (binaries) || SvTYPE (SvRV (binaries)) != SVt_PVAV)
1450 croak ("OpenCL::Context::program_with_binary: binaries must be specified as reference to an array of strings");
1451
1452 binaries = SvRV (binaries);
1453
1454 if (device_count != av_len ((AV *)binaries) + 1)
1455 croak ("OpenCL::Context::program_with_binary: differing numbers of devices and binaries are not allowed");
1456
1457 size_t *length_list = tmpbuf (sizeof (*length_list) * device_count);
1458 const unsigned char **binary_list = tmpbuf (sizeof (*binary_list) * device_count);
1459 cl_int *status_list = tmpbuf (sizeof (*status_list) * device_count);
1460
1461 int i;
1462 for (i = 0; i < device_count; ++i)
1463 {
1464 STRLEN len;
1465 binary_list [i] = (const unsigned char *)SvPVbyte (*av_fetch ((AV *)binaries, i, 0), len);
1466 length_list [i] = len;
1467 }
1468
1469 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithBinary, (self, device_count, device_list,
1470 length_list, binary_list,
1471 GIMME_V == G_ARRAY ? status_list : 0, &res));
1472
1473 EXTEND (SP, 2);
1474 PUSH_CLOBJ (stash_program, prog);
1475
1476 if (GIMME_V == G_ARRAY)
1477 {
1478 AV *av = newAV ();
1479 PUSHs (sv_2mortal (newRV_noinc ((SV *)av)));
1480
1481 for (i = device_count; i--; )
1482 av_store (av, i, newSViv (status_list [i]));
1483 }
1484
1485#if CL_VERSION_1_2
1486
1487void
1488program_with_built_in_kernels (OpenCL::Context self, SV *devices, SV *kernel_names)
1489 PPCODE:
1490 cl_uint device_count;
1491 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
1492
1493 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithBuiltInKernels, (self, device_count, device_list, SvPVbyte_nolen (kernel_names), &res));
1494
1495 XPUSH_CLOBJ (stash_program, prog);
1496
1497void
1498link_program (OpenCL::Context self, SV *devices, SV *options, SV *programs, SV *notify = &PL_sv_undef)
1499 CODE:
1500 cl_uint device_count = 0;
1501 cl_device_id *device_list = 0;
1502
1503 if (SvOK (devices))
1504 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
1505
1506 cl_uint program_count;
1507 cl_program *program_list = object_list (cv, 0, "programs", programs, "OpenCL::Program", &program_count);
1508
1509 void *user_data;
1510 program_callback pfn_notify = make_program_callback (notify, &user_data);
1511
1512 NEED_SUCCESS_ARG (cl_program prog, LinkProgram, (self, device_count, device_list, SvPVbyte_nolen (options),
1513 program_count, program_list, pfn_notify, user_data, &res));
1514
1515 XPUSH_CLOBJ (stash_program, prog);
1516
1517#endif
635 1518
636#BEGIN:context 1519#BEGIN:context
637 1520
638void 1521void
639reference_count (OpenCL::Context this) 1522reference_count (OpenCL::Context self)
640 ALIAS: 1523 ALIAS:
641 reference_count = CL_CONTEXT_REFERENCE_COUNT 1524 reference_count = CL_CONTEXT_REFERENCE_COUNT
642 num_devices = CL_CONTEXT_NUM_DEVICES 1525 num_devices = CL_CONTEXT_NUM_DEVICES
643 PPCODE: 1526 PPCODE:
644 cl_uint value [1]; 1527 cl_uint value [1];
645 NEED_SUCCESS (GetContextInfo, (this, ix, sizeof (value), value, 0)); 1528 NEED_SUCCESS (GetContextInfo, (self, ix, sizeof (value), value, 0));
646 EXTEND (SP, 1); 1529 EXTEND (SP, 1);
647 const int i = 0; 1530 const int i = 0;
648 PUSHs (sv_2mortal (newSVuv (value [i]))); 1531 PUSHs (sv_2mortal (newSVuv (value [i])));
649 1532
650void 1533void
651devices (OpenCL::Context this) 1534devices (OpenCL::Context self)
652 PPCODE: 1535 PPCODE:
653 size_t size; 1536 size_t size;
654 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_DEVICES, 0, 0, &size)); 1537 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, 0, 0, &size));
655 cl_device_id *value = tmpbuf (size); 1538 cl_device_id *value = tmpbuf (size);
656 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_DEVICES, size, value, 0)); 1539 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0));
657 int i, n = size / sizeof (*value); 1540 int i, n = size / sizeof (*value);
658 EXTEND (SP, n); 1541 EXTEND (SP, n);
659 for (i = 0; i < n; ++i) 1542 for (i = 0; i < n; ++i)
660 { 1543 {
661 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 1544 PUSH_CLOBJ (stash_device, value [i]);
662 } 1545 }
663 1546
664void 1547void
665properties (OpenCL::Context this) 1548properties (OpenCL::Context self)
666 PPCODE: 1549 PPCODE:
667 size_t size; 1550 size_t size;
668 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_PROPERTIES, 0, 0, &size)); 1551 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_PROPERTIES, 0, 0, &size));
669 cl_context_properties *value = tmpbuf (size); 1552 cl_context_properties *value = tmpbuf (size);
670 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_PROPERTIES, size, value, 0)); 1553 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_PROPERTIES, size, value, 0));
671 int i, n = size / sizeof (*value); 1554 int i, n = size / sizeof (*value);
672 EXTEND (SP, n); 1555 EXTEND (SP, n);
673 for (i = 0; i < n; ++i) 1556 for (i = 0; i < n; ++i)
674 PUSHs (sv_2mortal (newSVuv ((UV)value [i]))); 1557 PUSHs (sv_2mortal (newSVuv ((UV)value [i])));
675 1558
676#END:context 1559#END:context
677 1560
678MODULE = OpenCL PACKAGE = OpenCL::Queue 1561MODULE = OpenCL PACKAGE = OpenCL::Queue
679 1562
680void 1563void
681DESTROY (OpenCL::Queue this) 1564DESTROY (OpenCL::Queue self)
682 CODE: 1565 CODE:
683 clReleaseCommandQueue (this); 1566 clReleaseCommandQueue (self);
684 1567
685void 1568void
686enqueue_read_buffer (OpenCL::Queue this, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...) 1569read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...)
1570 ALIAS:
1571 enqueue_read_buffer = 0
687 PPCODE: 1572 PPCODE:
688 cl_event ev = 0;
689 EVENT_LIST (6, items - 6); 1573 EVENT_LIST (6);
690 1574
691 SvUPGRADE (data, SVt_PV); 1575 SvUPGRADE (data, SVt_PV);
692 SvGROW (data, len); 1576 SvGROW (data, len);
693 SvPOK_only (data); 1577 SvPOK_only (data);
694 SvCUR_set (data, len); 1578 SvCUR_set (data, len);
1579
1580 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)); 1581 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
696 1582
697 if (ev) 1583 if (ev)
698 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1584 XPUSH_CLOBJ (stash_event, ev);
699 1585
700void 1586void
701enqueue_write_buffer (OpenCL::Queue this, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...) 1587write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...)
1588 ALIAS:
1589 enqueue_write_buffer = 0
702 PPCODE: 1590 PPCODE:
703 cl_event ev = 0; 1591 EVENT_LIST (5);
1592
704 STRLEN len; 1593 STRLEN len;
705 char *ptr = SvPVbyte (data, len); 1594 char *ptr = SvPVbyte (data, len);
706 EVENT_LIST (5, items - 5);
707 1595
1596 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)); 1597 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
709 1598
710 if (ev) 1599 if (ev)
711 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1600 XPUSH_CLOBJ (stash_event, ev);
712 1601
1602#if CL_VERSION_1_2
1603
713void 1604void
714enqueue_copy_buffer (OpenCL::Queue this, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...) 1605fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...)
1606 ALIAS:
1607 enqueue_fill_buffer = 0
715 PPCODE: 1608 PPCODE:
1609 EVENT_LIST (5);
1610
1611 STRLEN len;
1612 char *ptr = SvPVbyte (data, len);
1613
716 cl_event ev = 0; 1614 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)); 1615 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
720 1616
721 if (ev) 1617 if (ev)
722 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1618 XPUSH_CLOBJ (stash_event, ev);
723 1619
724void 1620void
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, ...) 1621fill_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, ...)
1622 ALIAS:
1623 enqueue_fill_image = 0
726 PPCODE: 1624 PPCODE:
1625 EVENT_LIST (12);
1626
1627 const size_t origin [3] = { x, y, z };
1628 const size_t region [3] = { width, height, depth };
1629
1630 const cl_float c_f [4] = { r, g, b, a };
1631 const cl_uint c_u [4] = { r, g, b, a };
1632 const cl_int c_s [4] = { r, g, b, a };
1633 const void *c_fus [3] = { &c_f, &c_u, &c_s };
1634 static const char fus [] = { 0, 0, 0, 0, 0, 0, 0, 2, 2, 2, 1, 1, 1, 0, 0 };
1635 cl_image_format format;
1636 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
1637 assert (sizeof (fus) == CL_FLOAT + 1 - CL_SNORM_INT8);
1638 if (format.image_channel_data_type < CL_SNORM_INT8 || CL_FLOAT < format.image_channel_data_type)
1639 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported.");
1640
727 cl_event ev = 0; 1641 cl_event ev = 0;
1642 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type - CL_SNORM_INT8]],
1643 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1644
1645 if (ev)
1646 XPUSH_CLOBJ (stash_event, ev);
1647
1648#endif
1649
1650void
1651copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...)
1652 ALIAS:
1653 enqueue_copy_buffer = 0
1654 PPCODE:
1655 EVENT_LIST (6);
1656
1657 cl_event ev = 0;
1658 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1659
1660 if (ev)
1661 XPUSH_CLOBJ (stash_event, ev);
1662
1663void
1664read_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, ...)
1665 ALIAS:
1666 enqueue_read_buffer_rect = 0
1667 PPCODE:
1668 EVENT_LIST (17);
1669
728 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1670 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 }; 1671 const size_t host_origin[3] = { host_x, host_y, host_z };
730 const size_t region[3] = { width, height, depth }; 1672 const size_t region[3] = { width, height, depth };
731 EVENT_LIST (17, items - 17);
732 1673
733 if (!buf_row_pitch) 1674 if (!buf_row_pitch)
734 buf_row_pitch = region [0]; 1675 buf_row_pitch = region [0];
735 1676
736 if (!buf_slice_pitch) 1677 if (!buf_slice_pitch)
746 1687
747 SvUPGRADE (data, SVt_PV); 1688 SvUPGRADE (data, SVt_PV);
748 SvGROW (data, len); 1689 SvGROW (data, len);
749 SvPOK_only (data); 1690 SvPOK_only (data);
750 SvCUR_set (data, len); 1691 SvCUR_set (data, len);
1692
1693 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)); 1694 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 1695
753 if (ev) 1696 if (ev)
754 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1697 XPUSH_CLOBJ (stash_event, ev);
755 1698
756void 1699void
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, ...) 1700write_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, ...)
1701 ALIAS:
1702 enqueue_write_buffer_rect = 0
758 PPCODE: 1703 PPCODE:
759 cl_event ev = 0; 1704 EVENT_LIST (17);
1705
760 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1706 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 }; 1707 const size_t host_origin[3] = { host_x, host_y, host_z };
762 const size_t region[3] = { width, height, depth }; 1708 const size_t region[3] = { width, height, depth };
763 STRLEN len; 1709 STRLEN len;
764 char *ptr = SvPVbyte (data, len); 1710 char *ptr = SvPVbyte (data, len);
765 EVENT_LIST (17, items - 17);
766 1711
767 if (!buf_row_pitch) 1712 if (!buf_row_pitch)
768 buf_row_pitch = region [0]; 1713 buf_row_pitch = region [0];
769 1714
770 if (!buf_slice_pitch) 1715 if (!buf_slice_pitch)
779 size_t min_len = host_row_pitch * host_slice_pitch * region [2]; 1724 size_t min_len = host_row_pitch * host_slice_pitch * region [2];
780 1725
781 if (len < min_len) 1726 if (len < min_len)
782 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1727 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
783 1728
1729 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)); 1730 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 1731
786 if (ev) 1732 if (ev)
787 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1733 XPUSH_CLOBJ (stash_event, ev);
788 1734
789void 1735void
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, ...) 1736copy_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, ...)
1737 ALIAS:
1738 enqueue_copy_buffer_rect = 0
791 PPCODE: 1739 PPCODE:
1740 EVENT_LIST (16);
1741
1742 const size_t src_origin[3] = { src_x, src_y, src_z };
1743 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1744 const size_t region[3] = { width, height, depth };
1745
792 cl_event ev = 0; 1746 cl_event ev = 0;
1747 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));
1748
1749 if (ev)
1750 XPUSH_CLOBJ (stash_event, ev);
1751
1752void
1753read_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, ...)
1754 ALIAS:
1755 enqueue_read_image = 0
1756 PPCODE:
1757 EVENT_LIST (12);
1758
793 const size_t src_origin[3] = { src_x, src_y, src_z }; 1759 const size_t src_origin[3] = { src_x, src_y, src_z };
794 const size_t region[3] = { width, height, depth }; 1760 const size_t region[3] = { width, height, depth };
795 EVENT_LIST (12, items - 12);
796 1761
797 if (!row_pitch) 1762 if (!row_pitch)
798 row_pitch = img_row_pitch (src); 1763 row_pitch = img_row_pitch (src);
799 1764
800 if (depth > 1 && !slice_pitch) 1765 if (depth > 1 && !slice_pitch)
804 1769
805 SvUPGRADE (data, SVt_PV); 1770 SvUPGRADE (data, SVt_PV);
806 SvGROW (data, len); 1771 SvGROW (data, len);
807 SvPOK_only (data); 1772 SvPOK_only (data);
808 SvCUR_set (data, len); 1773 SvCUR_set (data, len);
1774
1775 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)); 1776 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 1777
811 if (ev) 1778 if (ev)
812 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1779 XPUSH_CLOBJ (stash_event, ev);
813 1780
814void 1781void
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, ...) 1782write_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, ...)
1783 ALIAS:
1784 enqueue_write_image = 0
816 PPCODE: 1785 PPCODE:
817 cl_event ev = 0; 1786 EVENT_LIST (12);
1787
818 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1788 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
819 const size_t region[3] = { width, height, depth }; 1789 const size_t region[3] = { width, height, depth };
820 STRLEN len; 1790 STRLEN len;
821 char *ptr = SvPVbyte (data, len); 1791 char *ptr = SvPVbyte (data, len);
822 EVENT_LIST (12, items - 12);
823 1792
824 if (!row_pitch) 1793 if (!row_pitch)
825 row_pitch = img_row_pitch (dst); 1794 row_pitch = img_row_pitch (dst);
826 1795
827 if (depth > 1 && !slice_pitch) 1796 if (depth > 1 && !slice_pitch)
830 size_t min_len = slice_pitch ? slice_pitch * depth : row_pitch * height; 1799 size_t min_len = slice_pitch ? slice_pitch * depth : row_pitch * height;
831 1800
832 if (len < min_len) 1801 if (len < min_len)
833 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1802 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
834 1803
1804 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)); 1805 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 1806
837 if (ev) 1807 if (ev)
838 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1808 XPUSH_CLOBJ (stash_event, ev);
839 1809
840void 1810void
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, ...) 1811copy_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, ...)
1812 ALIAS:
1813 enqueue_copy_image = 0
842 PPCODE: 1814 PPCODE:
843 cl_event ev = 0; 1815 EVENT_LIST (12);
1816
844 const size_t src_origin[3] = { src_x, src_y, src_z }; 1817 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 }; 1818 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
846 const size_t region[3] = { width, height, depth }; 1819 const size_t region[3] = { width, height, depth };
847 EVENT_LIST (16, items - 16);
848 1820
1821 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)); 1822 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
850 1823
851 if (ev) 1824 if (ev)
852 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1825 XPUSH_CLOBJ (stash_event, ev);
853 1826
854void 1827void
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, ...) 1828copy_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, ...)
1829 ALIAS:
1830 enqueue_copy_image_to_buffer = 0
856 PPCODE: 1831 PPCODE:
1832 EVENT_LIST (10);
1833
1834 const size_t src_origin[3] = { src_x, src_y, src_z };
1835 const size_t region [3] = { width, height, depth };
1836
857 cl_event ev = 0; 1837 cl_event ev = 0;
1838 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1839
1840 if (ev)
1841 XPUSH_CLOBJ (stash_event, ev);
1842
1843void
1844copy_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, ...)
1845 ALIAS:
1846 enqueue_copy_buffer_to_image = 0
1847 PPCODE:
1848 EVENT_LIST (10);
1849
858 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1850 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1851 const size_t region [3] = { width, height, depth };
1852
1853 cl_event ev = 0;
1854 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1855
1856 if (ev)
1857 XPUSH_CLOBJ (stash_event, ev);
1858
1859void
1860map_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, ...)
1861 ALIAS:
1862 enqueue_map_buffer = 0
1863 PPCODE:
1864 EVENT_LIST (6);
1865
1866 size_t cb = SvIV (cb_);
1867
1868 if (!SvOK (cb_))
1869 {
1870 NEED_SUCCESS (GetMemObjectInfo, (buf, CL_MEM_SIZE, sizeof (cb), &cb, 0));
1871 cb -= offset;
1872 }
1873
1874 cl_event ev;
1875 NEED_SUCCESS_ARG (void *ptr, EnqueueMapBuffer, (self, buf, blocking, map_flags, offset, cb, event_list_count, event_list_ptr, &ev, &res));
1876 XPUSHs (mapped_new (stash_mappedbuffer, self, buf, map_flags, ptr, cb, ev, 0, 0));
1877
1878void
1879map_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, ...)
1880 ALIAS:
1881 enqueue_map_image = 0
1882 PPCODE:
1883 size_t width = SvIV (width_);
1884 if (!SvOK (width_))
1885 {
1886 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_WIDTH, sizeof (width), &width, 0));
1887 width -= x;
1888 }
1889
1890 size_t height = SvIV (width_);
1891 if (!SvOK (height_))
1892 {
1893 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_HEIGHT, sizeof (height), &height, 0));
1894 height -= y;
1895 }
1896
1897 size_t depth = SvIV (width_);
1898 if (!SvOK (depth_))
1899 {
1900 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_DEPTH, sizeof (depth), &depth, 0));
1901 depth -= z;
1902
1903 // stupid opencl returns 0 for depth, but requires 1 for 2d images
1904 if (!depth)
1905 depth = 1;
1906 }
1907
1908 const size_t origin[3] = { x, y, z };
859 const size_t region[3] = { width, height, depth }; 1909 const size_t region[3] = { width, height, depth };
1910 size_t row_pitch, slice_pitch;
860 EVENT_LIST (10, items - 10); 1911 EVENT_LIST (10);
861 1912
862 NEED_SUCCESS (EnqueueCopyBufferToImage, (this, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1913 cl_event ev;
1914 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));
1915
1916 size_t cb = slice_pitch ? slice_pitch * region [2]
1917 : row_pitch ? row_pitch * region [1]
1918 : region [0];
1919
1920 XPUSHs (mapped_new (stash_mappedimage, self, img, map_flags, ptr, cb, ev, row_pitch, slice_pitch));
1921
1922void
1923unmap (OpenCL::Queue self, OpenCL::Mapped mapped, ...)
1924 PPCODE:
1925 mapped_unmap (cv, ST (1), mapped, self, &ST (2), items - 2);
1926 if (GIMME_V != G_VOID)
1927 {
1928 clRetainEvent (mapped->event);
1929 XPUSH_CLOBJ (stash_event, mapped->event);
1930 }
1931
1932void
1933task (OpenCL::Queue self, OpenCL::Kernel kernel, ...)
1934 ALIAS:
1935 enqueue_task = 0
1936 PPCODE:
1937 EVENT_LIST (2);
1938
1939 cl_event ev = 0;
1940 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
863 1941
864 if (ev) 1942 if (ev)
865 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1943 XPUSH_CLOBJ (stash_event, ev);
866 1944
867void 1945void
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, ...) 1946nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...)
1947 ALIAS:
1948 enqueue_nd_range_kernel = 0
907 PPCODE: 1949 PPCODE:
908 cl_event ev = 0; 1950 EVENT_LIST (5);
1951
909 size_t *gwo = 0, *gws, *lws = 0; 1952 size_t *gwo = 0, *gws, *lws = 0;
910 int gws_len; 1953 int gws_len;
911 size_t *lists; 1954 size_t *lists;
912 int i; 1955 int i;
913 EVENT_LIST (5, items - 5);
914 1956
915 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV) 1957 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV)
916 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference"); 1958 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference");
917 1959
918 gws_len = AvFILLp (SvRV (global_work_size)) + 1; 1960 gws_len = AvFILLp (SvRV (global_work_size)) + 1;
919 1961
920 lists = tmpbuf (sizeof (size_t) * 3 * gws_len); 1962 lists = tmpbuf (sizeof (size_t) * 3 * gws_len);
921 1963
922 gws = lists + gws_len * 0; 1964 gws = lists + gws_len * 0;
923 for (i = 0; i < gws_len; ++i) 1965 for (i = 0; i < gws_len; ++i)
1966 {
924 gws [i] = SvIV (AvARRAY (SvRV (global_work_size))[i]); 1967 gws [i] = SvIV (AvARRAY (SvRV (global_work_size))[i]);
1968 // at least nvidia crashes for 0-sized work group sizes, work around
1969 if (!gws [i])
1970 croak ("clEnqueueNDRangeKernel: global_work_size[%d] is zero, must be non-zero", i);
1971 }
925 1972
926 if (SvOK (global_work_offset)) 1973 if (SvOK (global_work_offset))
927 { 1974 {
928 if (!SvROK (global_work_offset) || SvTYPE (SvRV (global_work_offset)) != SVt_PVAV) 1975 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"); 1976 croak ("clEnqueueNDRangeKernel: global_work_offset must be undef or an array reference");
936 gwo [i] = SvIV (AvARRAY (SvRV (global_work_offset))[i]); 1983 gwo [i] = SvIV (AvARRAY (SvRV (global_work_offset))[i]);
937 } 1984 }
938 1985
939 if (SvOK (local_work_size)) 1986 if (SvOK (local_work_size))
940 { 1987 {
941 if (SvOK (local_work_size) && !SvROK (local_work_size) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV) 1988 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"); 1989 croak ("clEnqueueNDRangeKernel: local_work_size must be undef or an array reference");
943 1990
944 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len) 1991 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"); 1992 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size");
946 1993
947 lws = lists + gws_len * 2; 1994 lws = lists + gws_len * 2;
948 for (i = 0; i < gws_len; ++i) 1995 for (i = 0; i < gws_len; ++i)
1996 {
949 lws [i] = SvIV (AvARRAY (SvRV (local_work_size))[i]); 1997 lws [i] = SvIV (AvARRAY (SvRV (local_work_size))[i]);
1998 // at least nvidia crashes for 0-sized work group sizes, work around
1999 if (!lws [i])
2000 croak ("clEnqueueNDRangeKernel: local_work_size[%d] is zero, must be non-zero", i);
2001 }
950 } 2002 }
951 2003
2004 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)); 2005 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
953 2006
954 if (ev) 2007 if (ev)
955 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 2008 XPUSH_CLOBJ (stash_event, ev);
956 2009
2010#if CL_VERSION_1_2
2011
957void 2012void
958enqueue_marker (OpenCL::Queue this) 2013migrate_mem_objects (OpenCL::Queue self, SV *objects, cl_mem_migration_flags flags, ...)
2014 ALIAS:
2015 enqueue_migrate_mem_objects = 0
959 PPCODE: 2016 PPCODE:
2017 EVENT_LIST (3);
2018
2019 cl_uint obj_count;
2020 cl_mem *obj_list = object_list (cv, 0, "objects", objects, "OpenCL::Memory", &obj_count);
2021
2022 cl_event ev = 0;
2023 NEED_SUCCESS (EnqueueMigrateMemObjects, (self, obj_count, obj_list, flags, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2024
2025 if (ev)
2026 XPUSH_CLOBJ (stash_event, ev);
2027
2028#endif
2029
2030#if cl_apple_gl_sharing || cl_khr_gl_sharing
2031
2032void
2033acquire_gl_objects (OpenCL::Queue self, SV *objects, ...)
2034 ALIAS:
2035 release_gl_objects = 1
2036 enqueue_acquire_gl_objects = 0
2037 enqueue_release_gl_objects = 1
2038 PPCODE:
2039 EVENT_LIST (2);
2040
2041 cl_uint obj_count;
2042 cl_mem *obj_list = object_list (cv, 0, "objects", objects, "OpenCL::Memory", &obj_count);
2043
960 cl_event ev; 2044 cl_event ev = 0;
961 NEED_SUCCESS (EnqueueMarker, (this, &ev));
962 XPUSH_NEW_OBJ ("OpenCL::Event", ev);
963 2045
2046 if (ix)
2047 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, obj_count, obj_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2048 else
2049 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, obj_count, obj_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2050
2051 if (ev)
2052 XPUSH_CLOBJ (stash_event, ev);
2053
2054#endif
2055
964void 2056void
965enqueue_wait_for_events (OpenCL::Queue this, ...) 2057wait_for_events (OpenCL::Queue self, ...)
2058 ALIAS:
2059 enqueue_wait_for_events = 0
966 CODE: 2060 CODE:
967 EVENT_LIST (1, items - 1); 2061 EVENT_LIST (1);
2062#if PREFER_1_1
968 NEED_SUCCESS (EnqueueWaitForEvents, (this, event_list_count, event_list_ptr)); 2063 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
2064#else
2065 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0));
2066#endif
969 2067
970void 2068void
971enqueue_barrier (OpenCL::Queue this) 2069marker (OpenCL::Queue self, ...)
2070 ALIAS:
2071 enqueue_marker = 0
972 CODE: 2072 PPCODE:
2073 EVENT_LIST (1);
2074 cl_event ev = 0;
2075#if PREFER_1_1
2076 if (!event_list_count)
2077 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
2078 else
2079#if CL_VERSION_1_2
2080 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2081#else
2082 {
2083 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr)); // also a barrier
2084 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
2085 }
2086#endif
2087#else
2088 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2089#endif
2090 if (ev)
2091 XPUSH_CLOBJ (stash_event, ev);
2092
2093void
2094barrier (OpenCL::Queue self, ...)
2095 ALIAS:
2096 enqueue_barrier = 0
2097 PPCODE:
2098 EVENT_LIST (1);
2099 cl_event ev = 0;
2100#if PREFER_1_1
2101 if (!event_list_count && GIMME_V == G_VOID)
973 NEED_SUCCESS (EnqueueBarrier, (this)); 2102 NEED_SUCCESS (EnqueueBarrier, (self));
2103 else
2104#if CL_VERSION_1_2
2105 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2106#else
2107 {
2108 if (event_list_count)
2109 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
974 2110
2111 if (GIMME_V != G_VOID)
2112 NEED_SUCCESS (EnqueueMarker, (self, &ev));
2113 }
2114#endif
2115#else
2116 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2117#endif
2118 if (ev)
2119 XPUSH_CLOBJ (stash_event, ev);
2120
975void 2121void
976flush (OpenCL::Queue this) 2122flush (OpenCL::Queue self)
977 CODE: 2123 CODE:
978 NEED_SUCCESS (Flush, (this)); 2124 NEED_SUCCESS (Flush, (self));
979 2125
980void 2126void
981finish (OpenCL::Queue this) 2127finish (OpenCL::Queue self)
982 CODE: 2128 CODE:
983 NEED_SUCCESS (Finish, (this)); 2129 NEED_SUCCESS (Finish, (self));
984 2130
985void 2131void
986info (OpenCL::Queue this, cl_command_queue_info name) 2132info (OpenCL::Queue self, cl_command_queue_info name)
987 PPCODE: 2133 PPCODE:
988 INFO (CommandQueue) 2134 INFO (CommandQueue)
989 2135
990#BEGIN:command_queue 2136#BEGIN:command_queue
991 2137
992void 2138void
993context (OpenCL::Queue this) 2139context (OpenCL::Queue self)
994 PPCODE: 2140 PPCODE:
995 cl_context value [1]; 2141 cl_context value [1];
996 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_CONTEXT, sizeof (value), value, 0)); 2142 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0));
997 EXTEND (SP, 1); 2143 EXTEND (SP, 1);
998 const int i = 0; 2144 const int i = 0;
999 { 2145 {
1000 NEED_SUCCESS (RetainContext, (value [i])); 2146 NEED_SUCCESS (RetainContext, (value [i]));
1001 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2147 PUSH_CLOBJ (stash_context, value [i]);
1002 } 2148 }
1003 2149
1004void 2150void
1005device (OpenCL::Queue this) 2151device (OpenCL::Queue self)
1006 PPCODE: 2152 PPCODE:
1007 cl_device_id value [1]; 2153 cl_device_id value [1];
1008 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_DEVICE, sizeof (value), value, 0)); 2154 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0));
1009 EXTEND (SP, 1); 2155 EXTEND (SP, 1);
1010 const int i = 0; 2156 const int i = 0;
1011 { 2157 {
1012 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 2158 PUSH_CLOBJ (stash_device, value [i]);
1013 } 2159 }
1014 2160
1015void 2161void
1016reference_count (OpenCL::Queue this) 2162reference_count (OpenCL::Queue self)
1017 PPCODE: 2163 PPCODE:
1018 cl_uint value [1]; 2164 cl_uint value [1];
1019 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_REFERENCE_COUNT, sizeof (value), value, 0)); 2165 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_REFERENCE_COUNT, sizeof (value), value, 0));
1020 EXTEND (SP, 1); 2166 EXTEND (SP, 1);
1021 const int i = 0; 2167 const int i = 0;
1022 PUSHs (sv_2mortal (newSVuv (value [i]))); 2168 PUSHs (sv_2mortal (newSVuv (value [i])));
1023 2169
1024void 2170void
1025properties (OpenCL::Queue this) 2171properties (OpenCL::Queue self)
1026 PPCODE: 2172 PPCODE:
1027 cl_command_queue_properties value [1]; 2173 cl_command_queue_properties value [1];
1028 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_PROPERTIES, sizeof (value), value, 0)); 2174 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_PROPERTIES, sizeof (value), value, 0));
1029 EXTEND (SP, 1); 2175 EXTEND (SP, 1);
1030 const int i = 0; 2176 const int i = 0;
1031 PUSHs (sv_2mortal (newSViv (value [i]))); 2177 PUSHs (sv_2mortal (newSVuv (value [i])));
1032 2178
1033#END:command_queue 2179#END:command_queue
1034 2180
1035MODULE = OpenCL PACKAGE = OpenCL::Memory 2181MODULE = OpenCL PACKAGE = OpenCL::Memory
1036 2182
1037void 2183void
1038DESTROY (OpenCL::Memory this) 2184DESTROY (OpenCL::Memory self)
1039 CODE: 2185 CODE:
1040 clReleaseMemObject (this); 2186 clReleaseMemObject (self);
1041 2187
1042void 2188void
1043info (OpenCL::Memory this, cl_mem_info name) 2189info (OpenCL::Memory self, cl_mem_info name)
1044 PPCODE: 2190 PPCODE:
1045 INFO (MemObject) 2191 INFO (MemObject)
1046 2192
1047#BEGIN:mem 2193#BEGIN:mem
1048 2194
1049void 2195void
1050type (OpenCL::Memory this) 2196type (OpenCL::Memory self)
1051 PPCODE: 2197 PPCODE:
1052 cl_mem_object_type value [1]; 2198 cl_mem_object_type value [1];
1053 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_TYPE, sizeof (value), value, 0)); 2199 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_TYPE, sizeof (value), value, 0));
1054 EXTEND (SP, 1); 2200 EXTEND (SP, 1);
1055 const int i = 0; 2201 const int i = 0;
1056 PUSHs (sv_2mortal (newSViv (value [i]))); 2202 PUSHs (sv_2mortal (newSVuv (value [i])));
1057 2203
1058void 2204void
1059flags (OpenCL::Memory this) 2205flags (OpenCL::Memory self)
1060 PPCODE: 2206 PPCODE:
1061 cl_mem_flags value [1]; 2207 cl_mem_flags value [1];
1062 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_FLAGS, sizeof (value), value, 0)); 2208 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_FLAGS, sizeof (value), value, 0));
1063 EXTEND (SP, 1); 2209 EXTEND (SP, 1);
1064 const int i = 0; 2210 const int i = 0;
1065 PUSHs (sv_2mortal (newSViv (value [i]))); 2211 PUSHs (sv_2mortal (newSVuv (value [i])));
1066 2212
1067void 2213void
1068size (OpenCL::Memory this) 2214size (OpenCL::Memory self)
1069 ALIAS: 2215 ALIAS:
1070 size = CL_MEM_SIZE 2216 size = CL_MEM_SIZE
1071 offset = CL_MEM_OFFSET 2217 offset = CL_MEM_OFFSET
1072 PPCODE: 2218 PPCODE:
1073 size_t value [1]; 2219 size_t value [1];
1074 NEED_SUCCESS (GetMemObjectInfo, (this, ix, sizeof (value), value, 0)); 2220 NEED_SUCCESS (GetMemObjectInfo, (self, ix, sizeof (value), value, 0));
1075 EXTEND (SP, 1); 2221 EXTEND (SP, 1);
1076 const int i = 0; 2222 const int i = 0;
1077 PUSHs (sv_2mortal (newSVuv (value [i]))); 2223 PUSHs (sv_2mortal (newSVuv (value [i])));
1078 2224
1079void 2225void
1080host_ptr (OpenCL::Memory this) 2226host_ptr (OpenCL::Memory self)
1081 PPCODE: 2227 PPCODE:
1082 void * value [1]; 2228 void * value [1];
1083 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_HOST_PTR, sizeof (value), value, 0)); 2229 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_HOST_PTR, sizeof (value), value, 0));
1084 EXTEND (SP, 1); 2230 EXTEND (SP, 1);
1085 const int i = 0; 2231 const int i = 0;
1086 PUSHs (sv_2mortal (newSVuv ((IV)(intptr_t)value [i]))); 2232 PUSHs (sv_2mortal (newSVuv ((IV)(intptr_t)value [i])));
1087 2233
1088void 2234void
1089map_count (OpenCL::Memory this) 2235map_count (OpenCL::Memory self)
1090 ALIAS: 2236 ALIAS:
1091 map_count = CL_MEM_MAP_COUNT 2237 map_count = CL_MEM_MAP_COUNT
1092 reference_count = CL_MEM_REFERENCE_COUNT 2238 reference_count = CL_MEM_REFERENCE_COUNT
1093 PPCODE: 2239 PPCODE:
1094 cl_uint value [1]; 2240 cl_uint value [1];
1095 NEED_SUCCESS (GetMemObjectInfo, (this, ix, sizeof (value), value, 0)); 2241 NEED_SUCCESS (GetMemObjectInfo, (self, ix, sizeof (value), value, 0));
1096 EXTEND (SP, 1); 2242 EXTEND (SP, 1);
1097 const int i = 0; 2243 const int i = 0;
1098 PUSHs (sv_2mortal (newSVuv (value [i]))); 2244 PUSHs (sv_2mortal (newSVuv (value [i])));
1099 2245
1100void 2246void
1101context (OpenCL::Memory this) 2247context (OpenCL::Memory self)
1102 PPCODE: 2248 PPCODE:
1103 cl_context value [1]; 2249 cl_context value [1];
1104 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_CONTEXT, sizeof (value), value, 0)); 2250 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0));
1105 EXTEND (SP, 1); 2251 EXTEND (SP, 1);
1106 const int i = 0; 2252 const int i = 0;
1107 { 2253 {
1108 NEED_SUCCESS (RetainContext, (value [i])); 2254 NEED_SUCCESS (RetainContext, (value [i]));
1109 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2255 PUSH_CLOBJ (stash_context, value [i]);
1110 } 2256 }
1111 2257
1112void 2258void
1113associated_memobject (OpenCL::Memory this) 2259associated_memobject (OpenCL::Memory self)
1114 PPCODE: 2260 PPCODE:
1115 cl_mem value [1]; 2261 cl_mem value [1];
1116 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0)); 2262 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0));
1117 EXTEND (SP, 1); 2263 EXTEND (SP, 1);
1118 const int i = 0; 2264 const int i = 0;
1119 { 2265 {
1120 NEED_SUCCESS (RetainMemObject, (value [i])); 2266 NEED_SUCCESS (RetainMemObject, (value [i]));
1121 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Memory", value [i])); 2267 PUSH_CLOBJ (stash_memory, value [i]);
1122 } 2268 }
1123 2269
1124#END:mem 2270#END:mem
1125 2271
2272#if cl_apple_gl_sharing || cl_khr_gl_sharing
2273
2274void
2275gl_object_info (OpenCL::Memory self)
2276 PPCODE:
2277 cl_gl_object_type type;
2278 cl_GLuint name;
2279 NEED_SUCCESS (GetGLObjectInfo, (self, &type, &name));
2280 EXTEND (SP, 2);
2281 PUSHs (sv_2mortal (newSVuv (type)));
2282 PUSHs (sv_2mortal (newSVuv (name)));
2283
2284#endif
2285
2286MODULE = OpenCL PACKAGE = OpenCL::BufferObj
2287
2288void
2289sub_buffer_region (OpenCL::BufferObj self, cl_mem_flags flags, size_t origin, size_t size)
2290 PPCODE:
2291 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR | CL_MEM_ALLOC_HOST_PTR))
2292 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!");
2293
2294 cl_buffer_region crdata = { origin, size };
2295
2296 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res));
2297 XPUSH_CLOBJ (stash_buffer, mem);
2298
1126MODULE = OpenCL PACKAGE = OpenCL::Image 2299MODULE = OpenCL PACKAGE = OpenCL::Image
1127 2300
1128void 2301void
1129image_info (OpenCL::Image this, cl_image_info name) 2302image_info (OpenCL::Image self, cl_image_info name)
1130 PPCODE: 2303 PPCODE:
1131 INFO (Image) 2304 INFO (Image)
1132 2305
2306void
2307format (OpenCL::Image self)
2308 PPCODE:
2309 cl_image_format format;
2310 NEED_SUCCESS (GetImageInfo, (self, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
2311 EXTEND (SP, 2);
2312 PUSHs (sv_2mortal (newSVuv (format.image_channel_order)));
2313 PUSHs (sv_2mortal (newSVuv (format.image_channel_data_type)));
2314
1133#BEGIN:image 2315#BEGIN:image
1134 2316
1135void 2317void
1136element_size (OpenCL::Image this) 2318element_size (OpenCL::Image self)
1137 ALIAS: 2319 ALIAS:
1138 element_size = CL_IMAGE_ELEMENT_SIZE 2320 element_size = CL_IMAGE_ELEMENT_SIZE
1139 row_pitch = CL_IMAGE_ROW_PITCH 2321 row_pitch = CL_IMAGE_ROW_PITCH
1140 slice_pitch = CL_IMAGE_SLICE_PITCH 2322 slice_pitch = CL_IMAGE_SLICE_PITCH
1141 width = CL_IMAGE_WIDTH 2323 width = CL_IMAGE_WIDTH
1142 height = CL_IMAGE_HEIGHT 2324 height = CL_IMAGE_HEIGHT
1143 depth = CL_IMAGE_DEPTH 2325 depth = CL_IMAGE_DEPTH
1144 PPCODE: 2326 PPCODE:
1145 size_t value [1]; 2327 size_t value [1];
1146 NEED_SUCCESS (GetImageInfo, (this, ix, sizeof (value), value, 0)); 2328 NEED_SUCCESS (GetImageInfo, (self, ix, sizeof (value), value, 0));
1147 EXTEND (SP, 1); 2329 EXTEND (SP, 1);
1148 const int i = 0; 2330 const int i = 0;
1149 PUSHs (sv_2mortal (newSVuv (value [i]))); 2331 PUSHs (sv_2mortal (newSVuv (value [i])));
1150 2332
1151#END:image 2333#END:image
1152 2334
2335#if cl_apple_gl_sharing || cl_khr_gl_sharing
2336
2337#BEGIN:gl_texture
2338
2339void
2340target (OpenCL::Image self)
2341 PPCODE:
2342 cl_GLenum value [1];
2343 NEED_SUCCESS (GetGLTextureInfo, (self, CL_GL_TEXTURE_TARGET, sizeof (value), value, 0));
2344 EXTEND (SP, 1);
2345 const int i = 0;
2346 PUSHs (sv_2mortal (newSVuv (value [i])));
2347
2348void
2349gl_mipmap_level (OpenCL::Image self)
2350 PPCODE:
2351 cl_GLint value [1];
2352 NEED_SUCCESS (GetGLTextureInfo, (self, CL_GL_MIPMAP_LEVEL, sizeof (value), value, 0));
2353 EXTEND (SP, 1);
2354 const int i = 0;
2355 PUSHs (sv_2mortal (newSViv (value [i])));
2356
2357#END:gl_texture
2358
2359#endif
2360
1153MODULE = OpenCL PACKAGE = OpenCL::Sampler 2361MODULE = OpenCL PACKAGE = OpenCL::Sampler
1154 2362
1155void 2363void
1156DESTROY (OpenCL::Sampler this) 2364DESTROY (OpenCL::Sampler self)
1157 CODE: 2365 CODE:
1158 clReleaseSampler (this); 2366 clReleaseSampler (self);
1159 2367
1160void 2368void
1161info (OpenCL::Sampler this, cl_sampler_info name) 2369info (OpenCL::Sampler self, cl_sampler_info name)
1162 PPCODE: 2370 PPCODE:
1163 INFO (Sampler) 2371 INFO (Sampler)
1164 2372
1165#BEGIN:sampler 2373#BEGIN:sampler
1166 2374
1167void 2375void
1168reference_count (OpenCL::Sampler this) 2376reference_count (OpenCL::Sampler self)
1169 PPCODE: 2377 PPCODE:
1170 cl_uint value [1]; 2378 cl_uint value [1];
1171 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_REFERENCE_COUNT, sizeof (value), value, 0)); 2379 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_REFERENCE_COUNT, sizeof (value), value, 0));
1172 EXTEND (SP, 1); 2380 EXTEND (SP, 1);
1173 const int i = 0; 2381 const int i = 0;
1174 PUSHs (sv_2mortal (newSVuv (value [i]))); 2382 PUSHs (sv_2mortal (newSVuv (value [i])));
1175 2383
1176void 2384void
1177context (OpenCL::Sampler this) 2385context (OpenCL::Sampler self)
1178 PPCODE: 2386 PPCODE:
1179 cl_context value [1]; 2387 cl_context value [1];
1180 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_CONTEXT, sizeof (value), value, 0)); 2388 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0));
1181 EXTEND (SP, 1); 2389 EXTEND (SP, 1);
1182 const int i = 0; 2390 const int i = 0;
1183 { 2391 {
1184 NEED_SUCCESS (RetainContext, (value [i])); 2392 NEED_SUCCESS (RetainContext, (value [i]));
1185 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2393 PUSH_CLOBJ (stash_context, value [i]);
1186 } 2394 }
1187 2395
1188void 2396void
1189normalized_coords (OpenCL::Sampler this) 2397normalized_coords (OpenCL::Sampler self)
1190 PPCODE: 2398 PPCODE:
1191 cl_addressing_mode value [1]; 2399 cl_addressing_mode value [1];
1192 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0)); 2400 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0));
1193 EXTEND (SP, 1); 2401 EXTEND (SP, 1);
1194 const int i = 0; 2402 const int i = 0;
1195 PUSHs (sv_2mortal (newSViv (value [i]))); 2403 PUSHs (sv_2mortal (newSVuv (value [i])));
1196 2404
1197void 2405void
1198addressing_mode (OpenCL::Sampler this) 2406addressing_mode (OpenCL::Sampler self)
1199 PPCODE: 2407 PPCODE:
1200 cl_filter_mode value [1]; 2408 cl_filter_mode value [1];
1201 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0)); 2409 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0));
1202 EXTEND (SP, 1); 2410 EXTEND (SP, 1);
1203 const int i = 0; 2411 const int i = 0;
1204 PUSHs (sv_2mortal (newSViv (value [i]))); 2412 PUSHs (sv_2mortal (newSVuv (value [i])));
1205 2413
1206void 2414void
1207filter_mode (OpenCL::Sampler this) 2415filter_mode (OpenCL::Sampler self)
1208 PPCODE: 2416 PPCODE:
1209 cl_bool value [1]; 2417 cl_bool value [1];
1210 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_FILTER_MODE, sizeof (value), value, 0)); 2418 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_FILTER_MODE, sizeof (value), value, 0));
1211 EXTEND (SP, 1); 2419 EXTEND (SP, 1);
1212 const int i = 0; 2420 const int i = 0;
1213 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no)); 2421 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no));
1214 2422
1215#END:sampler 2423#END:sampler
1216 2424
1217MODULE = OpenCL PACKAGE = OpenCL::Program 2425MODULE = OpenCL PACKAGE = OpenCL::Program
1218 2426
1219void 2427void
1220DESTROY (OpenCL::Program this) 2428DESTROY (OpenCL::Program self)
1221 CODE: 2429 CODE:
1222 clReleaseProgram (this); 2430 clReleaseProgram (self);
1223 2431
1224void 2432void
1225build (OpenCL::Program this, OpenCL::Device device, SV *options = &PL_sv_undef) 2433build (OpenCL::Program self, SV *devices = &PL_sv_undef, SV *options = &PL_sv_undef, SV *notify = &PL_sv_undef)
2434 ALIAS:
2435 build_async = 1
1226 CODE: 2436 CODE:
2437 cl_uint device_count = 0;
2438 cl_device_id *device_list = 0;
2439
2440 if (SvOK (devices))
2441 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
2442
2443 void *user_data;
2444 program_callback pfn_notify = make_program_callback (notify, &user_data);
2445
2446 if (ix)
2447 build_program_async (self, device_count, device_list, SvPVbyte_nolen (options), user_data);
2448 else
2449 NEED_SUCCESS (BuildProgram, (self, device_count, device_list, SvPVbyte_nolen (options), pfn_notify, user_data));
2450
2451#if CL_VERSION_1_2
2452
2453void
2454compile (OpenCL::Program self, SV *devices, SV *options = &PL_sv_undef, SV *headers = &PL_sv_undef, SV *notify = &PL_sv_undef)
2455 CODE:
2456 cl_uint device_count = 0;
2457 cl_device_id *device_list = 0;
2458
2459 if (SvOK (devices))
2460 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
2461
2462 cl_uint header_count = 0;
2463 cl_program *header_list = 0;
2464 const char **header_name = 0;
2465
2466 if (SvOK (headers))
2467 {
2468 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVHV)
2469 croak ("clCompileProgram: headers must be undef or a hashref of name => OpenCL::Program pairs");
2470
2471 HV *hv = (HV *)SvRV (devices);
2472
2473 header_count = hv_iterinit (hv);
2474 header_list = tmpbuf (sizeof (*header_list) * header_count);
2475 header_name = tmpbuf (sizeof (*header_name) * header_count);
2476
2477 HE *he;
2478 int i = 0;
2479 while (he = hv_iternext (hv))
2480 {
2481 header_name [i] = SvPVbyte_nolen (HeSVKEY_force (he));
2482 header_list [i] = SvCLOBJ (cv, "headers", HeVAL (he), "OpenCL::Program");
2483 ++i;
2484 }
2485 }
2486
2487 void *user_data;
2488 program_callback pfn_notify = make_program_callback (notify, &user_data);
2489
1227 NEED_SUCCESS (BuildProgram, (this, 1, &device, SvPVbyte_nolen (options), 0, 0)); 2490 NEED_SUCCESS (CompileProgram, (self, device_count, device_list, SvPVbyte_nolen (options),
2491 header_count, header_list, header_name, pfn_notify, user_data));
1228 2492
2493#endif
2494
1229void 2495void
1230build_info (OpenCL::Program this, OpenCL::Device device, cl_program_build_info name) 2496build_info (OpenCL::Program self, OpenCL::Device device, cl_program_build_info name)
1231 PPCODE: 2497 PPCODE:
1232 size_t size; 2498 size_t size;
1233 NEED_SUCCESS (GetProgramBuildInfo, (this, device, name, 0, 0, &size)); 2499 NEED_SUCCESS (GetProgramBuildInfo, (self, device, name, 0, 0, &size));
1234 SV *sv = sv_2mortal (newSV (size)); 2500 SV *sv = sv_2mortal (newSV (size));
1235 SvUPGRADE (sv, SVt_PV); 2501 SvUPGRADE (sv, SVt_PV);
1236 SvPOK_only (sv); 2502 SvPOK_only (sv);
1237 SvCUR_set (sv, size); 2503 SvCUR_set (sv, size);
1238 NEED_SUCCESS (GetProgramBuildInfo, (this, device, name, size, SvPVX (sv), 0)); 2504 NEED_SUCCESS (GetProgramBuildInfo, (self, device, name, size, SvPVX (sv), 0));
1239 XPUSHs (sv); 2505 XPUSHs (sv);
1240 2506
1241#BEGIN:program_build 2507#BEGIN:program_build
1242 2508
1243void 2509void
1244build_status (OpenCL::Program this, OpenCL::Device device) 2510build_status (OpenCL::Program self, OpenCL::Device device)
1245 PPCODE: 2511 PPCODE:
1246 cl_build_status value [1]; 2512 cl_build_status value [1];
1247 NEED_SUCCESS (GetProgramBuildInfo, (this, device, CL_PROGRAM_BUILD_STATUS, sizeof (value), value, 0)); 2513 NEED_SUCCESS (GetProgramBuildInfo, (self, device, CL_PROGRAM_BUILD_STATUS, sizeof (value), value, 0));
1248 EXTEND (SP, 1); 2514 EXTEND (SP, 1);
1249 const int i = 0; 2515 const int i = 0;
1250 PUSHs (sv_2mortal (newSViv (value [i]))); 2516 PUSHs (sv_2mortal (newSViv (value [i])));
1251 2517
1252void 2518void
1253build_options (OpenCL::Program this, OpenCL::Device device) 2519build_options (OpenCL::Program self, OpenCL::Device device)
1254 ALIAS: 2520 ALIAS:
1255 build_options = CL_PROGRAM_BUILD_OPTIONS 2521 build_options = CL_PROGRAM_BUILD_OPTIONS
1256 build_log = CL_PROGRAM_BUILD_LOG 2522 build_log = CL_PROGRAM_BUILD_LOG
1257 PPCODE: 2523 PPCODE:
1258 size_t size; 2524 size_t size;
1259 NEED_SUCCESS (GetProgramBuildInfo, (this, device, ix, 0, 0, &size)); 2525 NEED_SUCCESS (GetProgramBuildInfo, (self, device, ix, 0, 0, &size));
1260 char *value = tmpbuf (size); 2526 char *value = tmpbuf (size);
1261 NEED_SUCCESS (GetProgramBuildInfo, (this, device, ix, size, value, 0)); 2527 NEED_SUCCESS (GetProgramBuildInfo, (self, device, ix, size, value, 0));
1262 EXTEND (SP, 1); 2528 EXTEND (SP, 1);
1263 const int i = 0; 2529 const int i = 0;
1264 PUSHs (sv_2mortal (newSVpv (value, 0))); 2530 PUSHs (sv_2mortal (newSVpv (value, 0)));
1265 2531
2532void
2533binary_type (OpenCL::Program self, OpenCL::Device device)
2534 PPCODE:
2535 cl_program_binary_type value [1];
2536 NEED_SUCCESS (GetProgramBuildInfo, (self, device, CL_PROGRAM_BINARY_TYPE, sizeof (value), value, 0));
2537 EXTEND (SP, 1);
2538 const int i = 0;
2539 PUSHs (sv_2mortal (newSVuv ((UV)value [i])));
2540
1266#END:program_build 2541#END:program_build
1267 2542
1268void 2543void
1269kernel (OpenCL::Program program, SV *function) 2544kernel (OpenCL::Program program, SV *function)
1270 PPCODE: 2545 PPCODE:
1271 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res)); 2546 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res));
1272 XPUSH_NEW_OBJ ("OpenCL::Kernel", kernel); 2547 XPUSH_CLOBJ (stash_kernel, kernel);
1273 2548
1274void 2549void
2550kernels_in_program (OpenCL::Program program)
2551 PPCODE:
2552 cl_uint num_kernels;
2553 NEED_SUCCESS (CreateKernelsInProgram, (program, 0, 0, &num_kernels));
2554 cl_kernel *kernels = tmpbuf (sizeof (cl_kernel) * num_kernels);
2555 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0));
2556
2557 int i;
2558 EXTEND (SP, num_kernels);
2559 for (i = 0; i < num_kernels; ++i)
2560 PUSH_CLOBJ (stash_kernel, kernels [i]);
2561
2562void
1275info (OpenCL::Program this, cl_program_info name) 2563info (OpenCL::Program self, cl_program_info name)
1276 PPCODE: 2564 PPCODE:
1277 INFO (Program) 2565 INFO (Program)
1278 2566
1279void 2567void
1280binaries (OpenCL::Program this) 2568binaries (OpenCL::Program self)
1281 PPCODE: 2569 PPCODE:
1282 cl_uint n, i; 2570 cl_uint n, i;
1283 size_t size; 2571 size_t size;
1284 2572
1285 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_NUM_DEVICES , sizeof (n) , &n , 0)); 2573 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_NUM_DEVICES , sizeof (n) , &n , 0));
1286 if (!n) XSRETURN_EMPTY; 2574 if (!n) XSRETURN_EMPTY;
1287 2575
1288 size_t *sizes = tmpbuf (sizeof (*sizes) * n); 2576 size_t *sizes = tmpbuf (sizeof (*sizes) * n);
1289 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARY_SIZES, sizeof (*sizes) * n, sizes, &size)); 2577 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARY_SIZES, sizeof (*sizes) * n, sizes, &size));
1290 if (size != sizeof (*sizes) * n) XSRETURN_EMPTY; 2578 if (size != sizeof (*sizes) * n) XSRETURN_EMPTY;
1291 unsigned char **ptrs = tmpbuf (sizeof (*ptrs) * n); 2579 unsigned char **ptrs = tmpbuf (sizeof (*ptrs) * n);
1292 2580
1293 EXTEND (SP, n); 2581 EXTEND (SP, n);
1294 for (i = 0; i < n; ++i) 2582 for (i = 0; i < n; ++i)
1295 { 2583 {
1296 SV *sv = sv_2mortal (newSV (sizes [i])); 2584 SV *sv = sv_2mortal (newSV (sizes [i]));
1297 SvUPGRADE (sv, SVt_PV); 2585 SvUPGRADE (sv, SVt_PV);
1298 SvPOK_only (sv); 2586 SvPOK_only (sv);
1299 SvCUR_set (sv, sizes [i]); 2587 SvCUR_set (sv, sizes [i]);
1300 ptrs [i] = SvPVX (sv); 2588 ptrs [i] = (void *)SvPVX (sv);
1301 PUSHs (sv); 2589 PUSHs (sv);
1302 } 2590 }
1303 2591
1304 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARIES , sizeof (*ptrs ) * n, ptrs , &size)); 2592 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARIES , sizeof (*ptrs ) * n, ptrs , &size));
1305 if (size != sizeof (*ptrs) * n) XSRETURN_EMPTY; 2593 if (size != sizeof (*ptrs) * n) XSRETURN_EMPTY;
1306 2594
1307#BEGIN:program 2595#BEGIN:program
1308 2596
1309void 2597void
1310reference_count (OpenCL::Program this) 2598reference_count (OpenCL::Program self)
1311 ALIAS: 2599 ALIAS:
1312 reference_count = CL_PROGRAM_REFERENCE_COUNT 2600 reference_count = CL_PROGRAM_REFERENCE_COUNT
1313 num_devices = CL_PROGRAM_NUM_DEVICES 2601 num_devices = CL_PROGRAM_NUM_DEVICES
1314 PPCODE: 2602 PPCODE:
1315 cl_uint value [1]; 2603 cl_uint value [1];
1316 NEED_SUCCESS (GetProgramInfo, (this, ix, sizeof (value), value, 0)); 2604 NEED_SUCCESS (GetProgramInfo, (self, ix, sizeof (value), value, 0));
1317 EXTEND (SP, 1); 2605 EXTEND (SP, 1);
1318 const int i = 0; 2606 const int i = 0;
1319 PUSHs (sv_2mortal (newSVuv (value [i]))); 2607 PUSHs (sv_2mortal (newSVuv (value [i])));
1320 2608
1321void 2609void
1322context (OpenCL::Program this) 2610context (OpenCL::Program self)
1323 PPCODE: 2611 PPCODE:
1324 cl_context value [1]; 2612 cl_context value [1];
1325 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_CONTEXT, sizeof (value), value, 0)); 2613 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0));
1326 EXTEND (SP, 1); 2614 EXTEND (SP, 1);
1327 const int i = 0; 2615 const int i = 0;
1328 { 2616 {
1329 NEED_SUCCESS (RetainContext, (value [i])); 2617 NEED_SUCCESS (RetainContext, (value [i]));
1330 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2618 PUSH_CLOBJ (stash_context, value [i]);
1331 } 2619 }
1332 2620
1333void 2621void
1334devices (OpenCL::Program this) 2622devices (OpenCL::Program self)
1335 PPCODE: 2623 PPCODE:
1336 size_t size; 2624 size_t size;
1337 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_DEVICES, 0, 0, &size)); 2625 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, 0, 0, &size));
1338 cl_device_id *value = tmpbuf (size); 2626 cl_device_id *value = tmpbuf (size);
1339 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_DEVICES, size, value, 0)); 2627 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0));
1340 int i, n = size / sizeof (*value); 2628 int i, n = size / sizeof (*value);
1341 EXTEND (SP, n); 2629 EXTEND (SP, n);
1342 for (i = 0; i < n; ++i) 2630 for (i = 0; i < n; ++i)
1343 { 2631 {
1344 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 2632 PUSH_CLOBJ (stash_device, value [i]);
1345 } 2633 }
1346 2634
1347void 2635void
1348source (OpenCL::Program this) 2636source (OpenCL::Program self)
1349 PPCODE: 2637 PPCODE:
1350 size_t size; 2638 size_t size;
1351 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_SOURCE, 0, 0, &size)); 2639 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_SOURCE, 0, 0, &size));
1352 char *value = tmpbuf (size); 2640 char *value = tmpbuf (size);
1353 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_SOURCE, size, value, 0)); 2641 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_SOURCE, size, value, 0));
1354 EXTEND (SP, 1); 2642 EXTEND (SP, 1);
1355 const int i = 0; 2643 const int i = 0;
1356 PUSHs (sv_2mortal (newSVpv (value, 0))); 2644 PUSHs (sv_2mortal (newSVpv (value, 0)));
1357 2645
1358void 2646void
1359binary_sizes (OpenCL::Program this) 2647binary_sizes (OpenCL::Program self)
1360 PPCODE: 2648 PPCODE:
1361 size_t size; 2649 size_t size;
1362 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARY_SIZES, 0, 0, &size)); 2650 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARY_SIZES, 0, 0, &size));
1363 size_t *value = tmpbuf (size); 2651 size_t *value = tmpbuf (size);
1364 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARY_SIZES, size, value, 0)); 2652 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARY_SIZES, size, value, 0));
1365 int i, n = size / sizeof (*value); 2653 int i, n = size / sizeof (*value);
1366 EXTEND (SP, n); 2654 EXTEND (SP, n);
1367 for (i = 0; i < n; ++i) 2655 for (i = 0; i < n; ++i)
1368 PUSHs (sv_2mortal (newSVuv (value [i]))); 2656 PUSHs (sv_2mortal (newSVuv (value [i])));
1369 2657
1370#END:program 2658#END:program
1371 2659
1372MODULE = OpenCL PACKAGE = OpenCL::Kernel 2660MODULE = OpenCL PACKAGE = OpenCL::Kernel
1373 2661
1374void 2662void
1375DESTROY (OpenCL::Kernel this) 2663DESTROY (OpenCL::Kernel self)
1376 CODE: 2664 CODE:
1377 clReleaseKernel (this); 2665 clReleaseKernel (self);
1378 2666
1379void 2667void
2668setf (OpenCL::Kernel self, const char *format, ...)
2669 CODE:
2670 int i;
2671 for (i = 2; ; ++i)
2672 {
2673 while (*format == ' ')
2674 ++format;
2675
2676 char type = *format++;
2677
2678 if (!type)
2679 break;
2680
2681 if (i >= items)
2682 croak ("OpenCL::Kernel::setf format string too long (not enough arguments)");
2683
2684 SV *sv = ST (i);
2685
2686 union
2687 {
2688 cl_char cc; cl_uchar cC; cl_short cs; cl_ushort cS;
2689 cl_int ci; cl_uint cI; cl_long cl; cl_ulong cL;
2690 cl_half ch; cl_float cf; cl_double cd;
2691 cl_mem cm;
2692 cl_sampler ca;
2693 size_t cz;
2694 cl_event ce;
2695 } arg;
2696 size_t size;
2697 int nullarg = 0;
2698
2699 switch (type)
2700 {
2701 case 'c': arg.cc = SvIV (sv); size = sizeof (arg.cc); break;
2702 case 'C': arg.cC = SvUV (sv); size = sizeof (arg.cC); break;
2703 case 's': arg.cs = SvIV (sv); size = sizeof (arg.cs); break;
2704 case 'S': arg.cS = SvUV (sv); size = sizeof (arg.cS); break;
2705 case 'i': arg.ci = SvIV (sv); size = sizeof (arg.ci); break;
2706 case 'I': arg.cI = SvUV (sv); size = sizeof (arg.cI); break;
2707 case 'l': arg.cl = SvIV (sv); size = sizeof (arg.cl); break;
2708 case 'L': arg.cL = SvUV (sv); size = sizeof (arg.cL); break;
2709
2710 case 'h': arg.ch = SvUV (sv); size = sizeof (arg.ch); break;
2711 case 'f': arg.cf = SvNV (sv); size = sizeof (arg.cf); break;
2712 case 'd': arg.cd = SvNV (sv); size = sizeof (arg.cd); break;
2713
2714 case 'z': nullarg = 1; size = SvIV (sv); break;
2715
2716 case 'm': nullarg = !SvOK (sv); arg.cm = SvCLOBJ (cv, "m", sv, "OpenCL::Memory" ); size = sizeof (arg.cm); break;
2717 case 'a': nullarg = !SvOK (sv); arg.ca = SvCLOBJ (cv, "a", sv, "OpenCL::Sampler"); size = sizeof (arg.ca); break;
2718 case 'e': nullarg = !SvOK (sv); arg.ca = SvCLOBJ (cv, "e", sv, "OpenCL::Event" ); size = sizeof (arg.ce); break;
2719
2720 default:
2721 croak ("OpenCL::Kernel::setf format character '%c' not supported", type);
2722 }
2723
2724 res = clSetKernelArg (self, i - 2, size, nullarg ? 0 : &arg);
2725 if (res)
2726 croak ("OpenCL::Kernel::setf kernel parameter '%c' (#%d): %s", type, i - 2, err2str (res));
2727 }
2728
2729 if (i != items)
2730 croak ("OpenCL::Kernel::setf format string too short (too many arguments)");
2731
2732void
1380set_char (OpenCL::Kernel this, cl_uint idx, cl_char value) 2733set_char (OpenCL::Kernel self, cl_uint idx, cl_char value)
1381 CODE: 2734 CODE:
1382 clSetKernelArg (this, idx, sizeof (value), &value); 2735 clSetKernelArg (self, idx, sizeof (value), &value);
1383 2736
1384void 2737void
1385set_uchar (OpenCL::Kernel this, cl_uint idx, cl_uchar value) 2738set_uchar (OpenCL::Kernel self, cl_uint idx, cl_uchar value)
1386 CODE: 2739 CODE:
1387 clSetKernelArg (this, idx, sizeof (value), &value); 2740 clSetKernelArg (self, idx, sizeof (value), &value);
1388 2741
1389void 2742void
1390set_short (OpenCL::Kernel this, cl_uint idx, cl_short value) 2743set_short (OpenCL::Kernel self, cl_uint idx, cl_short value)
1391 CODE: 2744 CODE:
1392 clSetKernelArg (this, idx, sizeof (value), &value); 2745 clSetKernelArg (self, idx, sizeof (value), &value);
1393 2746
1394void 2747void
1395set_ushort (OpenCL::Kernel this, cl_uint idx, cl_ushort value) 2748set_ushort (OpenCL::Kernel self, cl_uint idx, cl_ushort value)
1396 CODE: 2749 CODE:
1397 clSetKernelArg (this, idx, sizeof (value), &value); 2750 clSetKernelArg (self, idx, sizeof (value), &value);
1398 2751
1399void 2752void
1400set_int (OpenCL::Kernel this, cl_uint idx, cl_int value) 2753set_int (OpenCL::Kernel self, cl_uint idx, cl_int value)
1401 CODE: 2754 CODE:
1402 clSetKernelArg (this, idx, sizeof (value), &value); 2755 clSetKernelArg (self, idx, sizeof (value), &value);
1403 2756
1404void 2757void
1405set_uint (OpenCL::Kernel this, cl_uint idx, cl_uint value) 2758set_uint (OpenCL::Kernel self, cl_uint idx, cl_uint value)
1406 CODE: 2759 CODE:
1407 clSetKernelArg (this, idx, sizeof (value), &value); 2760 clSetKernelArg (self, idx, sizeof (value), &value);
1408 2761
1409void 2762void
1410set_long (OpenCL::Kernel this, cl_uint idx, cl_long value) 2763set_long (OpenCL::Kernel self, cl_uint idx, cl_long value)
1411 CODE: 2764 CODE:
1412 clSetKernelArg (this, idx, sizeof (value), &value); 2765 clSetKernelArg (self, idx, sizeof (value), &value);
1413 2766
1414void 2767void
1415set_ulong (OpenCL::Kernel this, cl_uint idx, cl_ulong value) 2768set_ulong (OpenCL::Kernel self, cl_uint idx, cl_ulong value)
1416 CODE: 2769 CODE:
1417 clSetKernelArg (this, idx, sizeof (value), &value); 2770 clSetKernelArg (self, idx, sizeof (value), &value);
1418 2771
1419void 2772void
1420set_half (OpenCL::Kernel this, cl_uint idx, cl_half value) 2773set_half (OpenCL::Kernel self, cl_uint idx, cl_half value)
1421 CODE: 2774 CODE:
1422 clSetKernelArg (this, idx, sizeof (value), &value); 2775 clSetKernelArg (self, idx, sizeof (value), &value);
1423 2776
1424void 2777void
1425set_float (OpenCL::Kernel this, cl_uint idx, cl_float value) 2778set_float (OpenCL::Kernel self, cl_uint idx, cl_float value)
1426 CODE: 2779 CODE:
1427 clSetKernelArg (this, idx, sizeof (value), &value); 2780 clSetKernelArg (self, idx, sizeof (value), &value);
1428 2781
1429void 2782void
1430set_double (OpenCL::Kernel this, cl_uint idx, cl_double value) 2783set_double (OpenCL::Kernel self, cl_uint idx, cl_double value)
1431 CODE: 2784 CODE:
1432 clSetKernelArg (this, idx, sizeof (value), &value); 2785 clSetKernelArg (self, idx, sizeof (value), &value);
1433 2786
1434void 2787void
1435set_memory (OpenCL::Kernel this, cl_uint idx, OpenCL::Memory_ornull value) 2788set_memory (OpenCL::Kernel self, cl_uint idx, OpenCL::Memory_ornull value)
1436 CODE: 2789 CODE:
1437 clSetKernelArg (this, idx, sizeof (value), &value); 2790 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
1438 2791
1439void 2792void
1440set_buffer (OpenCL::Kernel this, cl_uint idx, OpenCL::Buffer_ornull value) 2793set_buffer (OpenCL::Kernel self, cl_uint idx, OpenCL::Buffer_ornull value)
1441 CODE: 2794 CODE:
1442 clSetKernelArg (this, idx, sizeof (value), &value); 2795 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
1443 2796
1444void 2797void
1445set_image2d (OpenCL::Kernel this, cl_uint idx, OpenCL::Image2D_ornull value) 2798set_image (OpenCL::Kernel self, cl_uint idx, OpenCL::Image_ornull value)
2799 ALIAS:
2800 set_image2d = 0
2801 set_image3d = 0
1446 CODE: 2802 CODE:
1447 clSetKernelArg (this, idx, sizeof (value), &value); 2803 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
1448 2804
1449void 2805void
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) 2806set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value)
1456 CODE: 2807 CODE:
1457 clSetKernelArg (this, idx, sizeof (value), &value); 2808 clSetKernelArg (self, idx, sizeof (value), &value);
1458 2809
1459void 2810void
2811set_local (OpenCL::Kernel self, cl_uint idx, size_t size)
2812 CODE:
2813 clSetKernelArg (self, idx, size, 0);
2814
2815void
1460set_event (OpenCL::Kernel this, cl_uint idx, OpenCL::Event value) 2816set_event (OpenCL::Kernel self, cl_uint idx, OpenCL::Event value)
1461 CODE: 2817 CODE:
1462 clSetKernelArg (this, idx, sizeof (value), &value); 2818 clSetKernelArg (self, idx, sizeof (value), &value);
1463 2819
1464void 2820void
1465info (OpenCL::Kernel this, cl_kernel_info name) 2821info (OpenCL::Kernel self, cl_kernel_info name)
1466 PPCODE: 2822 PPCODE:
1467 INFO (Kernel) 2823 INFO (Kernel)
1468 2824
1469#BEGIN:kernel 2825#BEGIN:kernel
1470 2826
1471void 2827void
1472function_name (OpenCL::Kernel this) 2828function_name (OpenCL::Kernel self)
1473 PPCODE: 2829 PPCODE:
1474 size_t size; 2830 size_t size;
1475 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_FUNCTION_NAME, 0, 0, &size)); 2831 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_FUNCTION_NAME, 0, 0, &size));
1476 char *value = tmpbuf (size); 2832 char *value = tmpbuf (size);
1477 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_FUNCTION_NAME, size, value, 0)); 2833 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_FUNCTION_NAME, size, value, 0));
1478 EXTEND (SP, 1); 2834 EXTEND (SP, 1);
1479 const int i = 0; 2835 const int i = 0;
1480 PUSHs (sv_2mortal (newSVpv (value, 0))); 2836 PUSHs (sv_2mortal (newSVpv (value, 0)));
1481 2837
1482void 2838void
1483num_args (OpenCL::Kernel this) 2839num_args (OpenCL::Kernel self)
1484 ALIAS: 2840 ALIAS:
1485 num_args = CL_KERNEL_NUM_ARGS 2841 num_args = CL_KERNEL_NUM_ARGS
1486 reference_count = CL_KERNEL_REFERENCE_COUNT 2842 reference_count = CL_KERNEL_REFERENCE_COUNT
1487 PPCODE: 2843 PPCODE:
1488 cl_uint value [1]; 2844 cl_uint value [1];
1489 NEED_SUCCESS (GetKernelInfo, (this, ix, sizeof (value), value, 0)); 2845 NEED_SUCCESS (GetKernelInfo, (self, ix, sizeof (value), value, 0));
1490 EXTEND (SP, 1); 2846 EXTEND (SP, 1);
1491 const int i = 0; 2847 const int i = 0;
1492 PUSHs (sv_2mortal (newSVuv (value [i]))); 2848 PUSHs (sv_2mortal (newSVuv (value [i])));
1493 2849
1494void 2850void
1495context (OpenCL::Kernel this) 2851context (OpenCL::Kernel self)
1496 PPCODE: 2852 PPCODE:
1497 cl_context value [1]; 2853 cl_context value [1];
1498 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_CONTEXT, sizeof (value), value, 0)); 2854 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0));
1499 EXTEND (SP, 1); 2855 EXTEND (SP, 1);
1500 const int i = 0; 2856 const int i = 0;
1501 { 2857 {
1502 NEED_SUCCESS (RetainContext, (value [i])); 2858 NEED_SUCCESS (RetainContext, (value [i]));
1503 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2859 PUSH_CLOBJ (stash_context, value [i]);
1504 } 2860 }
1505 2861
1506void 2862void
1507program (OpenCL::Kernel this) 2863program (OpenCL::Kernel self)
1508 PPCODE: 2864 PPCODE:
1509 cl_program value [1]; 2865 cl_program value [1];
1510 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_PROGRAM, sizeof (value), value, 0)); 2866 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0));
1511 EXTEND (SP, 1); 2867 EXTEND (SP, 1);
1512 const int i = 0; 2868 const int i = 0;
1513 { 2869 {
1514 NEED_SUCCESS (RetainProgram, (value [i])); 2870 NEED_SUCCESS (RetainProgram, (value [i]));
1515 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Program", value [i])); 2871 PUSH_CLOBJ (stash_program, value [i]);
1516 } 2872 }
1517 2873
1518#END:kernel 2874#END:kernel
1519 2875
1520void 2876void
1521work_group_info (OpenCL::Kernel this, OpenCL::Device device, cl_kernel_work_group_info name) 2877work_group_info (OpenCL::Kernel self, OpenCL::Device device, cl_kernel_work_group_info name)
1522 PPCODE: 2878 PPCODE:
1523 size_t size; 2879 size_t size;
1524 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, name, 0, 0, &size)); 2880 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, name, 0, 0, &size));
1525 SV *sv = sv_2mortal (newSV (size)); 2881 SV *sv = sv_2mortal (newSV (size));
1526 SvUPGRADE (sv, SVt_PV); 2882 SvUPGRADE (sv, SVt_PV);
1527 SvPOK_only (sv); 2883 SvPOK_only (sv);
1528 SvCUR_set (sv, size); 2884 SvCUR_set (sv, size);
1529 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, name, size, SvPVX (sv), 0)); 2885 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, name, size, SvPVX (sv), 0));
1530 XPUSHs (sv); 2886 XPUSHs (sv);
1531 2887
1532#BEGIN:kernel_work_group 2888#BEGIN:kernel_work_group
1533 2889
1534void 2890void
1535work_group_size (OpenCL::Kernel this, OpenCL::Device device) 2891work_group_size (OpenCL::Kernel self, OpenCL::Device device)
1536 ALIAS: 2892 ALIAS:
1537 work_group_size = CL_KERNEL_WORK_GROUP_SIZE 2893 work_group_size = CL_KERNEL_WORK_GROUP_SIZE
1538 preferred_work_group_size_multiple = CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE 2894 preferred_work_group_size_multiple = CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE
1539 PPCODE: 2895 PPCODE:
1540 size_t value [1]; 2896 size_t value [1];
1541 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, ix, sizeof (value), value, 0)); 2897 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, ix, sizeof (value), value, 0));
1542 EXTEND (SP, 1); 2898 EXTEND (SP, 1);
1543 const int i = 0; 2899 const int i = 0;
1544 PUSHs (sv_2mortal (newSVuv (value [i]))); 2900 PUSHs (sv_2mortal (newSVuv (value [i])));
1545 2901
1546void 2902void
1547compile_work_group_size (OpenCL::Kernel this, OpenCL::Device device) 2903compile_work_group_size (OpenCL::Kernel self, OpenCL::Device device)
1548 PPCODE: 2904 PPCODE:
1549 size_t size; 2905 size_t size;
1550 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, 0, 0, &size)); 2906 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, 0, 0, &size));
1551 size_t *value = tmpbuf (size); 2907 size_t *value = tmpbuf (size);
1552 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, size, value, 0)); 2908 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, size, value, 0));
1553 int i, n = size / sizeof (*value); 2909 int i, n = size / sizeof (*value);
1554 EXTEND (SP, n); 2910 EXTEND (SP, n);
1555 for (i = 0; i < n; ++i) 2911 for (i = 0; i < n; ++i)
1556 PUSHs (sv_2mortal (newSVuv (value [i]))); 2912 PUSHs (sv_2mortal (newSVuv (value [i])));
1557 2913
1558void 2914void
1559local_mem_size (OpenCL::Kernel this, OpenCL::Device device) 2915local_mem_size (OpenCL::Kernel self, OpenCL::Device device)
1560 ALIAS: 2916 ALIAS:
1561 local_mem_size = CL_KERNEL_LOCAL_MEM_SIZE 2917 local_mem_size = CL_KERNEL_LOCAL_MEM_SIZE
1562 private_mem_size = CL_KERNEL_PRIVATE_MEM_SIZE 2918 private_mem_size = CL_KERNEL_PRIVATE_MEM_SIZE
1563 PPCODE: 2919 PPCODE:
1564 cl_ulong value [1]; 2920 cl_ulong value [1];
1565 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, ix, sizeof (value), value, 0)); 2921 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, ix, sizeof (value), value, 0));
1566 EXTEND (SP, 1); 2922 EXTEND (SP, 1);
1567 const int i = 0; 2923 const int i = 0;
1568 PUSHs (sv_2mortal (newSVuv (value [i]))); 2924 PUSHs (sv_2mortal (newSVuv (value [i])));
1569 2925
1570#END:kernel_work_group 2926#END:kernel_work_group
1571 2927
2928#if CL_VERSION_1_2
2929
2930void
2931arg_info (OpenCL::Kernel self, cl_uint idx, cl_kernel_arg_info name)
2932 PPCODE:
2933 size_t size;
2934 NEED_SUCCESS (GetKernelArgInfo, (self, idx, name, 0, 0, &size));
2935 SV *sv = sv_2mortal (newSV (size));
2936 SvUPGRADE (sv, SVt_PV);
2937 SvPOK_only (sv);
2938 SvCUR_set (sv, size);
2939 NEED_SUCCESS (GetKernelArgInfo, (self, idx, name, size, SvPVX (sv), 0));
2940 XPUSHs (sv);
2941
2942#BEGIN:kernel_arg
2943
2944void
2945arg_address_qualifier (OpenCL::Kernel self, cl_uint idx)
2946 PPCODE:
2947 cl_kernel_arg_address_qualifier value [1];
2948 NEED_SUCCESS (GetKernelArgInfo, (self, idx, CL_KERNEL_ARG_ADDRESS_QUALIFIER, sizeof (value), value, 0));
2949 EXTEND (SP, 1);
2950 const int i = 0;
2951 PUSHs (sv_2mortal (newSVuv (value [i])));
2952
2953void
2954arg_access_qualifier (OpenCL::Kernel self, cl_uint idx)
2955 PPCODE:
2956 cl_kernel_arg_access_qualifier value [1];
2957 NEED_SUCCESS (GetKernelArgInfo, (self, idx, CL_KERNEL_ARG_ACCESS_QUALIFIER, sizeof (value), value, 0));
2958 EXTEND (SP, 1);
2959 const int i = 0;
2960 PUSHs (sv_2mortal (newSVuv (value [i])));
2961
2962void
2963arg_type_name (OpenCL::Kernel self, cl_uint idx)
2964 ALIAS:
2965 arg_type_name = CL_KERNEL_ARG_TYPE_NAME
2966 arg_name = CL_KERNEL_ARG_NAME
2967 PPCODE:
2968 size_t size;
2969 NEED_SUCCESS (GetKernelArgInfo, (self, idx, ix, 0, 0, &size));
2970 char *value = tmpbuf (size);
2971 NEED_SUCCESS (GetKernelArgInfo, (self, idx, ix, size, value, 0));
2972 EXTEND (SP, 1);
2973 const int i = 0;
2974 PUSHs (sv_2mortal (newSVpv (value, 0)));
2975
2976void
2977arg_type_qualifier (OpenCL::Kernel self, cl_uint idx)
2978 PPCODE:
2979 cl_kernel_arg_type_qualifier value [1];
2980 NEED_SUCCESS (GetKernelArgInfo, (self, idx, CL_KERNEL_ARG_TYPE_QUALIFIER, sizeof (value), value, 0));
2981 EXTEND (SP, 1);
2982 const int i = 0;
2983 PUSHs (sv_2mortal (newSVuv (value [i])));
2984
2985#END:kernel_arg
2986
2987#endif
2988
1572MODULE = OpenCL PACKAGE = OpenCL::Event 2989MODULE = OpenCL PACKAGE = OpenCL::Event
1573 2990
1574void 2991void
1575DESTROY (OpenCL::Event this) 2992DESTROY (OpenCL::Event self)
1576 CODE: 2993 CODE:
1577 clReleaseEvent (this); 2994 clReleaseEvent (self);
1578 2995
1579void 2996void
1580wait (OpenCL::Event this) 2997wait (OpenCL::Event self)
1581 CODE: 2998 CODE:
1582 clWaitForEvents (1, &this); 2999 clWaitForEvents (1, &self);
1583 3000
1584void 3001void
3002cb (OpenCL::Event self, cl_int command_exec_callback_type, SV *cb)
3003 CODE:
3004 clSetEventCallback (self, command_exec_callback_type, eq_event_notify, SvREFCNT_inc (s_get_cv (cb)));
3005
3006void
1585info (OpenCL::Event this, cl_event_info name) 3007info (OpenCL::Event self, cl_event_info name)
1586 PPCODE: 3008 PPCODE:
1587 INFO (Event) 3009 INFO (Event)
1588 3010
1589#BEGIN:event 3011#BEGIN:event
1590 3012
1591void 3013void
1592command_queue (OpenCL::Event this) 3014command_queue (OpenCL::Event self)
1593 PPCODE: 3015 PPCODE:
1594 cl_command_queue value [1]; 3016 cl_command_queue value [1];
1595 NEED_SUCCESS (GetEventInfo, (this, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0)); 3017 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0));
1596 EXTEND (SP, 1); 3018 EXTEND (SP, 1);
1597 const int i = 0; 3019 const int i = 0;
1598 { 3020 {
1599 NEED_SUCCESS (RetainCommandQueue, (value [i])); 3021 NEED_SUCCESS (RetainCommandQueue, (value [i]));
1600 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Queue", value [i])); 3022 PUSH_CLOBJ (stash_queue, value [i]);
1601 } 3023 }
1602 3024
1603void 3025void
1604command_type (OpenCL::Event this) 3026command_type (OpenCL::Event self)
1605 PPCODE: 3027 PPCODE:
1606 cl_command_type value [1]; 3028 cl_command_type value [1];
1607 NEED_SUCCESS (GetEventInfo, (this, CL_EVENT_COMMAND_TYPE, sizeof (value), value, 0)); 3029 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_TYPE, sizeof (value), value, 0));
1608 EXTEND (SP, 1); 3030 EXTEND (SP, 1);
1609 const int i = 0; 3031 const int i = 0;
1610 PUSHs (sv_2mortal (newSVuv (value [i]))); 3032 PUSHs (sv_2mortal (newSVuv (value [i])));
1611 3033
1612void 3034void
1613reference_count (OpenCL::Event this) 3035reference_count (OpenCL::Event self)
1614 ALIAS: 3036 ALIAS:
1615 reference_count = CL_EVENT_REFERENCE_COUNT 3037 reference_count = CL_EVENT_REFERENCE_COUNT
1616 command_execution_status = CL_EVENT_COMMAND_EXECUTION_STATUS 3038 command_execution_status = CL_EVENT_COMMAND_EXECUTION_STATUS
1617 PPCODE: 3039 PPCODE:
1618 cl_uint value [1]; 3040 cl_uint value [1];
1619 NEED_SUCCESS (GetEventInfo, (this, ix, sizeof (value), value, 0)); 3041 NEED_SUCCESS (GetEventInfo, (self, ix, sizeof (value), value, 0));
1620 EXTEND (SP, 1); 3042 EXTEND (SP, 1);
1621 const int i = 0; 3043 const int i = 0;
1622 PUSHs (sv_2mortal (newSVuv (value [i]))); 3044 PUSHs (sv_2mortal (newSVuv (value [i])));
1623 3045
1624void 3046void
1625context (OpenCL::Event this) 3047context (OpenCL::Event self)
1626 PPCODE: 3048 PPCODE:
1627 cl_context value [1]; 3049 cl_context value [1];
1628 NEED_SUCCESS (GetEventInfo, (this, CL_EVENT_CONTEXT, sizeof (value), value, 0)); 3050 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0));
1629 EXTEND (SP, 1); 3051 EXTEND (SP, 1);
1630 const int i = 0; 3052 const int i = 0;
1631 { 3053 {
1632 NEED_SUCCESS (RetainContext, (value [i])); 3054 NEED_SUCCESS (RetainContext, (value [i]));
1633 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 3055 PUSH_CLOBJ (stash_context, value [i]);
1634 } 3056 }
1635 3057
1636#END:event 3058#END:event
1637 3059
1638void 3060void
1639profiling_info (OpenCL::Event this, cl_profiling_info name) 3061profiling_info (OpenCL::Event self, cl_profiling_info name)
1640 PPCODE: 3062 PPCODE:
1641 INFO (EventProfiling) 3063 INFO (EventProfiling)
1642 3064
1643#BEGIN:profiling 3065#BEGIN:profiling
1644 3066
1645void 3067void
1646profiling_command_queued (OpenCL::Event this) 3068profiling_command_queued (OpenCL::Event self)
1647 ALIAS: 3069 ALIAS:
1648 profiling_command_queued = CL_PROFILING_COMMAND_QUEUED 3070 profiling_command_queued = CL_PROFILING_COMMAND_QUEUED
1649 profiling_command_submit = CL_PROFILING_COMMAND_SUBMIT 3071 profiling_command_submit = CL_PROFILING_COMMAND_SUBMIT
1650 profiling_command_start = CL_PROFILING_COMMAND_START 3072 profiling_command_start = CL_PROFILING_COMMAND_START
1651 profiling_command_end = CL_PROFILING_COMMAND_END 3073 profiling_command_end = CL_PROFILING_COMMAND_END
1652 PPCODE: 3074 PPCODE:
1653 cl_ulong value [1]; 3075 cl_ulong value [1];
1654 NEED_SUCCESS (GetEventProfilingInfo, (this, ix, sizeof (value), value, 0)); 3076 NEED_SUCCESS (GetEventProfilingInfo, (self, ix, sizeof (value), value, 0));
1655 EXTEND (SP, 1); 3077 EXTEND (SP, 1);
1656 const int i = 0; 3078 const int i = 0;
1657 PUSHs (sv_2mortal (newSVuv (value [i]))); 3079 PUSHs (sv_2mortal (newSVuv (value [i])));
1658 3080
1659#END:profiling 3081#END:profiling
1660 3082
1661MODULE = OpenCL PACKAGE = OpenCL::UserEvent 3083MODULE = OpenCL PACKAGE = OpenCL::UserEvent
1662 3084
1663void 3085void
1664set_status (OpenCL::UserEvent this, cl_int execution_status) 3086set_status (OpenCL::UserEvent self, cl_int execution_status)
1665 CODE: 3087 CODE:
1666 clSetUserEventStatus (this, execution_status); 3088 clSetUserEventStatus (self, execution_status);
1667 3089
3090MODULE = OpenCL PACKAGE = OpenCL::Mapped
3091
3092void
3093DESTROY (SV *self)
3094 CODE:
3095 OpenCL__Mapped mapped = SvMAPPED (self);
3096
3097 clEnqueueUnmapMemObject (mapped->queue, mapped->memobj, mapped->ptr, 1, &mapped->event, 0);
3098 mapped_detach (self, mapped);
3099
3100 clReleaseCommandQueue (mapped->queue);
3101 clReleaseEvent (mapped->event);
3102 Safefree (mapped);
3103
3104void
3105unmap (OpenCL::Mapped self, ...)
3106 CODE:
3107 mapped_unmap (cv, ST (0), self, self->queue, &ST (1), items - 1);
3108
3109bool
3110mapped (OpenCL::Mapped self)
3111 CODE:
3112 RETVAL = !!self->ptr;
3113 OUTPUT:
3114 RETVAL
3115
3116void
3117wait (OpenCL::Mapped self)
3118 PPCODE:
3119 if (self->event)
3120 NEED_SUCCESS (WaitForEvents, (1, &self->event));
3121
3122void
3123event (OpenCL::Mapped self)
3124 PPCODE:
3125 if (!self->event)
3126 XSRETURN_UNDEF;
3127
3128 clRetainEvent (self->event);
3129 XPUSH_CLOBJ (stash_event, self->event);
3130
3131size_t
3132size (OpenCL::Mapped self)
3133 CODE:
3134 RETVAL = self->cb;
3135 OUTPUT:
3136 RETVAL
3137
3138IV
3139ptr (OpenCL::Mapped self)
3140 CODE:
3141 RETVAL = PTR2IV (self->ptr);
3142 OUTPUT:
3143 RETVAL
3144
3145void
3146set (OpenCL::Mapped self, size_t offset, SV *data)
3147 CODE:
3148 STRLEN len;
3149 const char *ptr = SvPVbyte (data, len);
3150
3151 if (offset + len > self->cb)
3152 croak ("OpenCL::Mapped::set out of bound condition detected");
3153
3154 memcpy (offset + (char *)self->ptr, ptr, len);
3155
3156MODULE = OpenCL PACKAGE = OpenCL::MappedBuffer
3157
3158MODULE = OpenCL PACKAGE = OpenCL::MappedImage
3159
3160IV
3161row_pitch (OpenCL::Mapped self)
3162 ALIAS:
3163 slice_pitch = 1
3164 CODE:
3165 RETVAL = ix ? self->slice_pitch : self->row_pitch;
3166 OUTPUT:
3167 RETVAL
3168
3169
3170

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines