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

Comparing OpenCL/OpenCL.pm (file contents):
Revision 1.2 by root, Tue Nov 15 09:24:40 2011 UTC vs.
Revision 1.88 by root, Tue Oct 22 17:25:38 2013 UTC

1=head1 NAME 1=head1 NAME
2 2
3OpenCL - bindings to, well, OpenCL 3OpenCL - Open Computing Language Bindings
4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 use OpenCL; 7 use OpenCL;
8 8
9=head1 DESCRIPTION 9=head1 DESCRIPTION
10 10
11This is an early release which is not useful yet. 11This is an early release which might be useful, but hasn't seen much testing.
12 12
13=head2 OpenCL FROM 10000 FEET HEIGHT
14
15Here is a high level overview of OpenCL:
16
17First you need to find one or more OpenCL::Platforms (kind of like
18vendors) - usually there is only one.
19
20Each platform gives you access to a number of OpenCL::Device objects, e.g.
21your graphics card.
22
23From a platform and some device(s), you create an OpenCL::Context, which is
24a very central object in OpenCL: Once you have a context you can create
25most other objects:
26
27OpenCL::Program objects, which store source code and, after building for a
28specific device ("compiling and linking"), also binary programs. For each
29kernel function in a program you can then create an OpenCL::Kernel object
30which represents basically a function call with argument values.
31
32OpenCL::Memory objects of various flavours: OpenCL::Buffer objects (flat
33memory areas, think arrays or structs) and OpenCL::Image objects (think 2D
34or 3D array) for bulk data and input and output for kernels.
35
36OpenCL::Sampler objects, which are kind of like texture filter modes in
37OpenGL.
38
39OpenCL::Queue objects - command queues, which allow you to submit memory
40reads, writes and copies, as well as kernel calls to your devices. They
41also offer a variety of methods to synchronise request execution, for
42example with barriers or OpenCL::Event objects.
43
44OpenCL::Event objects are used to signal when something is complete.
45
46=head2 HELPFUL RESOURCES
47
48The OpenCL specs used to develop this module - download these and keept
49hema round, they are required reference material:
50
51 http://www.khronos.org/registry/cl/specs/opencl-1.1.pdf
52 http://www.khronos.org/registry/cl/specs/opencl-1.2.pdf
53 http://www.khronos.org/registry/cl/specs/opencl-1.2-extensions.pdf
54
55OpenCL manpages:
56
57 http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/
58 http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/
59
60If you are into UML class diagrams, the following diagram might help - if
61not, it will be mildly confusing (also, the class hierarchy of this module
62is much more fine-grained):
63
64 http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/classDiagram.html
65
66Here's a tutorial from AMD (very AMD-centric, too), not sure how useful it
67is, but at least it's free of charge:
68
69 http://developer.amd.com/zones/OpenCLZone/courses/Documents/Introduction_to_OpenCL_Programming%20Training_Guide%20%28201005%29.pdf
70
71And here's NVIDIA's OpenCL Best Practises Guide:
72
73 http://developer.download.nvidia.com/compute/cuda/3_2/toolkit/docs/OpenCL_Best_Practices_Guide.pdf
74
75=head1 BASIC WORKFLOW
76
77To get something done, you basically have to do this once (refer to the
78examples below for actual code, this is just a high-level description):
79
80Find some platform (e.g. the first one) and some device(s) (e.g. the first
81device of the platform), and create a context from those.
82
83Create program objects from your OpenCL source code, then build (compile)
84the programs for each device you want to run them on.
85
86Create kernel objects for all kernels you want to use (surprisingly, these
87are not device-specific).
88
89Then, to execute stuff, you repeat these steps, possibly resuing or
90sharing some buffers:
91
92Create some input and output buffers from your context. Set these as
93arguments to your kernel.
94
95Enqueue buffer writes to initialise your input buffers (when not
96initialised at creation time).
97
98Enqueue the kernel execution.
99
100Enqueue buffer reads for your output buffer to read results.
101
102=head1 EXAMPLES
103
13Enumerate all devices and get contexts for them; 104=head2 Enumerate all devices and get contexts for them.
105
106Best run this once to get a feel for the platforms and devices in your
107system.
14 108
15 for my $platform (OpenCL::platforms) { 109 for my $platform (OpenCL::platforms) {
16 warn $platform->info (OpenCL::PLATFORM_NAME); 110 printf "platform: %s\n", $platform->name;
17 warn $platform->info (OpenCL::PLATFORM_EXTENSIONS); 111 printf "extensions: %s\n", $platform->extensions;
18 for my $device ($platform->devices) { 112 for my $device ($platform->devices) {
19 warn $device->info (OpenCL::DEVICE_NAME); 113 printf "+ device: %s\n", $device->name;
20 my $ctx = $device->context_simple; 114 my $ctx = $platform->context (undef, [$device]);
21 # do stuff 115 # do stuff
22 } 116 }
23 } 117 }
24 118
25Get a useful context and a command queue: 119=head2 Get a useful context and a command queue.
26 120
27 my $dev = ((OpenCL::platforms)[0]->devices)[0]; 121This is a useful boilerplate for any OpenCL program that only wants to use
28 my $ctx = $dev->context_simple; 122one device,
123
124 my ($platform) = OpenCL::platforms; # find first platform
125 my ($dev) = $platform->devices; # find first device of platform
126 my $ctx = $platform->context (undef, [$dev]); # create context out of those
127 my $queue = $ctx->queue ($dev); # create a command queue for the device
128
129=head2 Print all supported image formats of a context.
130
131Best run this once for your context, to see whats available and how to
132gather information.
133
134 for my $type (OpenCL::MEM_OBJECT_IMAGE2D, OpenCL::MEM_OBJECT_IMAGE3D) {
135 print "supported image formats for ", OpenCL::enum2str $type, "\n";
136
137 for my $f ($ctx->supported_image_formats (0, $type)) {
138 printf " %-10s %-20s\n", OpenCL::enum2str $f->[0], OpenCL::enum2str $f->[1];
139 }
140 }
141
142=head2 Create a buffer with some predefined data, read it back synchronously,
143then asynchronously.
144
145 my $buf = $ctx->buffer_sv (OpenCL::MEM_COPY_HOST_PTR, "helmut");
146
147 $queue->read_buffer ($buf, 1, 1, 3, my $data);
148 print "$data\n";
149
150 my $ev = $queue->read_buffer ($buf, 0, 1, 3, my $data);
151 $ev->wait;
152 print "$data\n"; # prints "elm"
153
154=head2 Create and build a program, then create a kernel out of one of its
155functions.
156
157 my $src = '
158 kernel void
159 squareit (global float *input, global float *output)
160 {
161 $id = get_global_id (0);
162 output [id] = input [id] * input [id];
163 }
164 ';
165
166 my $prog = $ctx->build_program ($src);
167 my $kernel = $prog->kernel ("squareit");
168
169=head2 Create some input and output float buffers, then call the
170'squareit' kernel on them.
171
172 my $input = $ctx->buffer_sv (OpenCL::MEM_COPY_HOST_PTR, pack "f*", 1, 2, 3, 4.5);
173 my $output = $ctx->buffer (0, OpenCL::SIZEOF_FLOAT * 5);
174
175 # set buffer
176 $kernel->set_buffer (0, $input);
177 $kernel->set_buffer (1, $output);
178
179 # execute it for all 4 numbers
180 $queue->nd_range_kernel ($kernel, undef, [4], undef);
181
182 # enqueue a synchronous read
183 $queue->read_buffer ($output, 1, 0, OpenCL::SIZEOF_FLOAT * 4, my $data);
184
185 # print the results:
186 printf "%s\n", join ", ", unpack "f*", $data;
187
188=head2 The same enqueue operations as before, but assuming an out-of-order queue,
189showing off barriers.
190
191 # execute it for all 4 numbers
192 $queue->nd_range_kernel ($kernel, undef, [4], undef);
193
194 # enqueue a barrier to ensure in-order execution
195 $queue->barrier;
196
197 # enqueue an async read
198 $queue->read_buffer ($output, 0, 0, OpenCL::SIZEOF_FLOAT * 4, my $data);
199
200 # wait for all requests to finish
201 $queue->finish;
202
203=head2 The same enqueue operations as before, but assuming an out-of-order queue,
204showing off event objects and wait lists.
205
206 # execute it for all 4 numbers
207 my $ev = $queue->nd_range_kernel ($kernel, undef, [4], undef);
208
209 # enqueue an async read
210 $ev = $queue->read_buffer ($output, 0, 0, OpenCL::SIZEOF_FLOAT * 4, my $data, $ev);
211
212 # wait for the last event to complete
213 $ev->wait;
214
215=head2 Use the OpenGL module to share a texture between OpenCL and OpenGL and draw some julia
216set flight effect.
217
218This is quite a long example to get you going - you can also download it
219from L<http://cvs.schmorp.de/OpenCL/examples/juliaflight>.
220
221 use OpenGL ":all";
222 use OpenCL;
223
224 my $S = $ARGV[0] || 256; # window/texture size, smaller is faster
225
226 # open a window and create a gl texture
227 OpenGL::glpOpenWindow width => $S, height => $S;
228 my $texid = glGenTextures_p 1;
229 glBindTexture GL_TEXTURE_2D, $texid;
230 glTexImage2D_c GL_TEXTURE_2D, 0, GL_RGBA8, $S, $S, 0, GL_RGBA, GL_UNSIGNED_BYTE, 0;
231
232 # find and use the first opencl device that let's us get a shared opengl context
233 my $platform;
234 my $dev;
235 my $ctx;
236
237 sub get_context {
238 for (OpenCL::platforms) {
239 $platform = $_;
240 for ($platform->devices) {
241 $dev = $_;
242 $ctx = eval { $platform->context ([OpenCL::GLX_DISPLAY_KHR, undef, OpenCL::GL_CONTEXT_KHR, undef], [$dev]) }
243 and return;
244 }
245 }
246
247 die "cannot find suitable OpenCL device\n";
248 }
249
250 get_context;
251
29 my $queue = $ctx->command_queue_simple ($dev); 252 my $queue = $ctx->queue ($dev);
253
254 # now attach an opencl image2d object to the opengl texture
255 my $tex = $ctx->gl_texture2d (OpenCL::MEM_WRITE_ONLY, GL_TEXTURE_2D, 0, $texid);
256
257 # now the boring opencl code
258 my $src = <<EOF;
259 kernel void
260 juliatunnel (write_only image2d_t img, float time)
261 {
262 int2 xy = (int2)(get_global_id (0), get_global_id (1));
263 float2 p = convert_float2 (xy) / $S.f * 2.f - 1.f;
264
265 float2 m = (float2)(1.f, p.y) / fabs (p.x); // tunnel
266 m.x = fabs (fmod (m.x + time * 0.05f, 4.f) - 2.f);
267
268 float2 z = m;
269 float2 c = (float2)(sin (time * 0.01133f), cos (time * 0.02521f));
270
271 for (int i = 0; i < 25 && dot (z, z) < 4.f; ++i) // standard julia
272 z = (float2)(z.x * z.x - z.y * z.y, 2.f * z.x * z.y) + c;
273
274 float3 colour = (float3)(z.x, z.y, atan2 (z.y, z.x));
275 write_imagef (img, xy, (float4)(colour * p.x * p.x, 1.));
276 }
277 EOF
278
279 my $prog = $ctx->build_program ($src);
280 my $kernel = $prog->kernel ("juliatunnel");
281
282 # program compiled, kernel ready, now draw and loop
283
284 for (my $time; ; ++$time) {
285 # acquire objects from opengl
286 $queue->acquire_gl_objects ([$tex]);
287
288 # configure and run our kernel
289 $kernel->setf ("mf", $tex, $time*2); # mf = memory object, float
290 $queue->nd_range_kernel ($kernel, undef, [$S, $S], undef);
291
292 # release objects to opengl again
293 $queue->release_gl_objects ([$tex]);
294
295 # wait
296 $queue->finish;
297
298 # now draw the texture, the defaults should be all right
299 glTexParameterf GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST;
300
301 glEnable GL_TEXTURE_2D;
302 glBegin GL_QUADS;
303 glTexCoord2f 0, 1; glVertex3i -1, -1, -1;
304 glTexCoord2f 0, 0; glVertex3i 1, -1, -1;
305 glTexCoord2f 1, 0; glVertex3i 1, 1, -1;
306 glTexCoord2f 1, 1; glVertex3i -1, 1, -1;
307 glEnd;
308
309 glXSwapBuffers;
310
311 select undef, undef, undef, 1/60;
312 }
313
314=head2 How to modify the previous example to not rely on GL sharing.
315
316For those poor souls with only a sucky CPU OpenCL implementation, you
317currently have to read the image into some perl scalar, and then modify a
318texture or use glDrawPixels or so).
319
320First, when you don't need gl sharing, you can create the context much simpler:
321
322 $ctx = $platform->context (undef, [$dev])
323
324To use a texture, you would modify the above example by creating an
325OpenCL::Image manually instead of deriving it from a texture:
326
327 my $tex = $ctx->image2d (OpenCL::MEM_WRITE_ONLY, OpenCL::RGBA, OpenCL::UNORM_INT8, $S, $S);
328
329And in the draw loop, intead of acquire_gl_objects/release_gl_objects, you
330would read the image2d after the kernel has written it:
331
332 $queue->read_image ($tex, 0, 0, 0, 0, $S, $S, 1, 0, 0, my $data);
333
334And then you would upload the pixel data to the texture (or use glDrawPixels):
335
336 glTexSubImage2D_s GL_TEXTURE_2D, 0, 0, 0, $S, $S, GL_RGBA, GL_UNSIGNED_BYTE, $data;
337
338The fully modified example can be found at
339L<http://cvs.schmorp.de/OpenCL/examples/juliaflight-nosharing>.
340
341=head2 Julia sets look soooo 80ies.
342
343Then colour them differently, e.g. using orbit traps! Replace the loop and
344colour calculation from the previous examples by this:
345
346 float2 dm = (float2)(1.f, 1.f);
347
348 for (int i = 0; i < 25; ++i)
349 {
350 z = (float2)(z.x * z.x - z.y * z.y, 2.f * z.x * z.y) + c;
351 dm = fmin (dm, (float2)(fabs (dot (z, z) - 1.f), fabs (z.x - 1.f)));
352 }
353
354 float3 colour = (float3)(dm.x * dm.y, dm.x * dm.y, dm.x);
355
356Also try C<-10.f> instead of C<-1.f>.
357
358=head1 DOCUMENTATION
359
360=head2 BASIC CONVENTIONS
361
362This is not a one-to-one C-style translation of OpenCL to Perl - instead
363I attempted to make the interface as type-safe as possible by introducing
364object syntax where it makes sense. There are a number of important
365differences between the OpenCL C API and this module:
30 366
31=over 4 367=over 4
32 368
369=item * Object lifetime managament is automatic - there is no need
370to free objects explicitly (C<clReleaseXXX>), the release function
371is called automatically once all Perl references to it go away.
372
373=item * OpenCL uses CamelCase for function names
374(e.g. C<clGetPlatformIDs>, C<clGetPlatformInfo>), while this module
375uses underscores as word separator and often leaves out prefixes
376(C<OpenCL::platforms>, C<< $platform->info >>).
377
378=item * OpenCL often specifies fixed vector function arguments as short
379arrays (C<size_t origin[3]>), while this module explicitly expects the
380components as separate arguments (C<$orig_x, $orig_y, $orig_z>) in
381function calls.
382
383=item * Structures are often specified by flattening out their components
384as with short vectors, and returned as arrayrefs.
385
386=item * When enqueuing commands, the wait list is specified by adding
387extra arguments to the function - anywhere a C<$wait_events...> argument
388is documented this can be any number of event objects. As an extsnion
389implemented by this module, C<undef> values will be ignored in the event
390list.
391
392=item * When enqueuing commands, if the enqueue method is called in void
393context, no event is created. In all other contexts an event is returned
394by the method.
395
396=item * This module expects all functions to return C<OpenCL::SUCCESS>. If any
397other status is returned the function will throw an exception, so you
398don't normally have to to any error checking.
399
400=back
401
402=head2 CONSTANTS
403
404All C<CL_xxx> constants that this module supports are always available
405in the C<OpenCL> namespace as C<OpenCL::xxx> (i.e. without the C<CL_>
406prefix). Constants which are not defined in the header files used during
407compilation, or otherwise are not available, will have the value C<0> (in
408some cases, this will make them indistinguishable from real constants,
409sorry).
410
411The latest version of this module knows and exports the constants
412listed in L<http://cvs.schmorp.de/OpenCL/constiv.h>.
413
414=head2 OPENCL 1.1 VS. OPENCL 1.2
415
416This module supports both OpenCL version 1.1 and 1.2, although the OpenCL
4171.2 interface hasn't been tested much for lack of availability of an
418actual implementation.
419
420Every function or method in this manual page that interfaces to a
421particular OpenCL function has a link to the its C manual page.
422
423If the link contains a F<1.1>, then this function is an OpenCL 1.1
424function. Most but not all also exist in OpenCL 1.2, and this module
425tries to emulate the missing ones for you, when told to do so at
426compiletime. You can check whether a function was removed in OpenCL 1.2 by
427replacing the F<1.1> component in the URL by F<1.2>.
428
429If the link contains a F<1.2>, then this is a OpenCL 1.2-only
430function. Even if the module was compiled with OpenCL 1.2 header files
431and has an 1.2 OpenCL library, calling such a function on a platform that
432doesn't implement 1.2 causes undefined behaviour, usually a crash (But
433this is not guaranteed).
434
435You can find out whether this module was compiled to prefer 1.1
436functionality by ooking at C<OpenCL::PREFER_1_1> - if it is true, then
4371.1 functions generally are implemented using 1.1 OpenCL functions. If it
438is false, then 1.1 functions missing from 1.2 are emulated by calling 1.2
439fucntions.
440
441This is a somewhat sorry state of affairs, but the Khronos group choose to
442make every release of OpenCL source and binary incompatible with previous
443releases.
444
445=head2 PERL AND OPENCL TYPES
446
447This handy(?) table lists OpenCL types and their perl, PDL and pack/unpack
448format equivalents:
449
450 OpenCL perl PDL pack/unpack
451 char IV - c
452 uchar IV byte C
453 short IV short s
454 ushort IV ushort S
455 int IV long? l
456 uint IV - L
457 long IV longlong q
458 ulong IV - Q
459 float NV float f
460 half IV ushort S
461 double NV double d
462
463=head2 GLX SUPPORT
464
465Due to the sad state that OpenGL support is in in Perl (mostly the OpenGL
466module, which has little to no documentation and has little to no support
467for glX), this module, as a special extension, treats context creation
468properties C<OpenCL::GLX_DISPLAY_KHR> and C<OpenCL::GL_CONTEXT_KHR>
469specially: If either or both of these are C<undef>, then the OpenCL
470module tries to dynamically resolve C<glXGetCurrentDisplay> and
471C<glXGetCurrentContext>, call these functions and use their return values
472instead.
473
474For this to work, the OpenGL library must be loaded, a GLX context must
475have been created and be made current, and C<dlsym> must be available and
476capable of finding the function via C<RTLD_DEFAULT>.
477
478=head2 EVENT SYSTEM
479
480OpenCL can generate a number of (potentially) asynchronous events, for
481example, after compiling a program, to signal a context-related error or,
482perhaps most important, to signal completion of queued jobs (by setting
483callbacks on OpenCL::Event objects).
484
485The OpenCL module converts all these callbacks into events - you can
486still register callbacks, but they are not executed when your OpenCL
487implementation calls the actual callback, but only later. Therefore, none
488of the limitations of OpenCL callbacks apply to the perl implementation:
489it is perfectly safe to make blocking operations from event callbacks, and
490enqueued operations don't need to be flushed.
491
492To facilitate this, this module maintains an event queue - each
493time an asynchronous event happens, it is queued, and perl will be
494interrupted. This is implemented via the L<Async::Interrupt> module. In
495addition, this module has L<AnyEvent> support, so it can seamlessly
496integrate itself into many event loops.
497
498Since L<Async::Interrupt> is a bit hard to understand, here are some case examples:
499
500=head3 Don't use callbacks.
501
502When your program never uses any callbacks, then there will never be any
503notifications you need to take care of, and therefore no need to worry
504about all this.
505
506You can achieve a great deal by explicitly waiting for events, or using
507barriers and flush calls. In many programs, there is no need at all to
508tinker with asynchronous events.
509
510=head3 Use AnyEvent
511
512This module automatically registers a watcher that invokes all outstanding
513event callbacks when AnyEvent is initialised (and block asynchronous
514interruptions). Using this mode of operations is the safest and most
515recommended one.
516
517To use this, simply use AnyEvent and this module normally, make sure you
518have an event loop running:
519
520 use Gtk2 -init;
521 use AnyEvent;
522
523 # initialise AnyEvent, by creating a watcher, or:
524 AnyEvent::detect;
525
526 my $e = $queue->marker;
527 $e->cb (sub {
528 warn "opencl is finished\n";
529 })
530
531 main Gtk2;
532
533Note that this module will not initialise AnyEvent for you. Before
534AnyEvent is initialised, the module will asynchronously interrupt perl
535instead. To avoid any surprises, it's best to explicitly initialise
536AnyEvent.
537
538You can temporarily enable asynchronous interruptions (see next paragraph)
539by calling C<$OpenCL::INTERRUPT->unblock> and disable them again by
540calling C<$OpenCL::INTERRUPT->block>.
541
542=head3 Let yourself be interrupted at any time
543
544This mode is the default unless AnyEvent is loaded and initialised. In
545this mode, OpenCL asynchronously interrupts a running perl program. The
546emphasis is on both I<asynchronously> and I<running> here.
547
548Asynchronously means that perl might execute your callbacks at any
549time. For example, in the following code (I<THAT YOU SHOULD NOT COPY>),
550the C<until> loop following the marker call will be interrupted by the
551callback:
552
553 my $e = $queue->marker;
554 my $flag;
555 $e->cb (sub { $flag = 1 });
556 1 until $flag;
557 # $flag is now 1
558
559The reason why you shouldn't blindly copy the above code is that
560busy waiting is a really really bad thing, and really really bad for
561performance.
562
563While at first this asynchronous business might look exciting, it can be
564really hard, because you need to be prepared for the callback code to be
565executed at any time, which limits the amount of things the callback code
566can do safely.
567
568This can be mitigated somewhat by using C<<
569$OpenCL::INTERRUPT->scope_block >> (see the L<Async::Interrupt>
570documentation for details).
571
572The other problem is that your program must be actively I<running> to be
573interrupted. When you calculate stuff, your program is running. When you
574hang in some C functions or other block execution (by calling C<sleep>,
575C<select>, running an event loop and so on), your program is waiting, not
576running.
577
578One way around that would be to attach a read watcher to your event loop,
579listening for events on C<< $OpenCL::INTERRUPT->pipe_fileno >>, using a
580dummy callback (C<sub { }>) to temporarily execute some perl code.
581
582That is then awfully close to using the built-in AnyEvent support above,
583though, so consider that one instead.
584
585=head3 Be creative
586
587OpenCL exports the L<Async::Interrupt> object it uses in the global
588variable C<$OpenCL::INTERRUPT>. You can configure it in any way you like.
589
590So if you want to feel like a real pro, err, wait, if you feel no risk
591menas no fun, you can experiment by implementing your own mode of
592operations.
593
33=cut 594=cut
34 595
35package OpenCL; 596package OpenCL;
36 597
37use common::sense; 598use common::sense;
599use Carp ();
600use Async::Interrupt ();
601
602our $POLL_FUNC; # set by XS
38 603
39BEGIN { 604BEGIN {
40 our $VERSION = '0.01'; 605 our $VERSION = '1.01';
41 606
42 require XSLoader; 607 require XSLoader;
43 XSLoader::load (__PACKAGE__, $VERSION); 608 XSLoader::load (__PACKAGE__, $VERSION);
609
610 @OpenCL::Platform::ISA =
611 @OpenCL::Device::ISA =
612 @OpenCL::Context::ISA =
613 @OpenCL::Queue::ISA =
614 @OpenCL::Memory::ISA =
615 @OpenCL::Sampler::ISA =
616 @OpenCL::Program::ISA =
617 @OpenCL::Kernel::ISA =
618 @OpenCL::Event::ISA = OpenCL::Object::;
619
620 @OpenCL::SubDevice::ISA = OpenCL::Device::;
621
622 @OpenCL::Buffer::ISA =
623 @OpenCL::Image::ISA = OpenCL::Memory::;
624
625 @OpenCL::BufferObj::ISA = OpenCL::Buffer::;
626
627 @OpenCL::Image2D::ISA =
628 @OpenCL::Image3D::ISA =
629 @OpenCL::Image2DArray::ISA =
630 @OpenCL::Image1D::ISA =
631 @OpenCL::Image1DArray::ISA =
632 @OpenCL::Image1DBuffer::ISA = OpenCL::Image::;
633
634 @OpenCL::UserEvent::ISA = OpenCL::Event::;
635
636 @OpenCL::MappedBuffer::ISA =
637 @OpenCL::MappedImage::ISA = OpenCL::Mapped::;
44} 638}
45 639
640=head2 THE OpenCL PACKAGE
641
642=over 4
643
644=item $int = OpenCL::errno
645
646The last error returned by a function - it's only valid after an error occured
647and before calling another OpenCL function.
648
649=item $str = OpenCL::err2str [$errval]
650
651Converts an error value into a human readable string. If no error value is
652given, then the last error will be used (as returned by OpenCL::errno).
653
654The latest version of this module knows the error constants
655listed in L<http://cvs.schmorp.de/OpenCL/errstr.h>.
656
657=item $str = OpenCL::enum2str $enum
658
659Converts most enum values (of parameter names, image format constants,
660object types, addressing and filter modes, command types etc.) into a
661human readable string. When confronted with some random integer it can be
662very helpful to pass it through this function to maybe get some readable
663string out of it.
664
665The latest version of this module knows the enumaration constants
666listed in L<http://cvs.schmorp.de/OpenCL/enumstr.h>.
667
668=item @platforms = OpenCL::platforms
669
670Returns all available OpenCL::Platform objects.
671
672L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetPlatformIDs.html>
673
674=item $ctx = OpenCL::context_from_type $properties, $type = OpenCL::DEVICE_TYPE_DEFAULT, $callback->($err, $pvt) = $print_stderr
675
676Tries to create a context from a default device and platform type - never worked for me.
677Consider using C<< $platform->context_from_type >> instead.
678
679type: OpenCL::DEVICE_TYPE_DEFAULT, OpenCL::DEVICE_TYPE_CPU, OpenCL::DEVICE_TYPE_GPU,
680OpenCL::DEVICE_TYPE_ACCELERATOR, OpenCL::DEVICE_TYPE_CUSTOM, OpenCL::DEVICE_TYPE_ALL.
681
682L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContextFromType.html>
683
684=item $ctx = OpenCL::context $properties, \@devices, $callback->($err, $pvt) = $print_stderr)
685
686Create a new OpenCL::Context object using the given device object(s).
687Consider using C<< $platform->context >> instead.
688
689L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContext.html>
690
691=item OpenCL::wait_for_events $wait_events...
692
693Waits for all events to complete.
694
695L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clWaitForEvents.html>
696
697=item OpenCL::poll
698
699Checks if there are any outstanding events (see L<EVENT SYSTEM>) and
700invokes their callbacks.
701
702=item $OpenCL::INTERRUPT
703
704The L<Async::Interrupt> object used to signal asynchronous events (see
705L<EVENT SYSTEM>).
706
707=cut
708
709our $INTERRUPT = new Async::Interrupt c_cb => [$POLL_FUNC, 0];
710
711&_eq_initialise ($INTERRUPT->signal_func);
712
713=item $OpenCL::WATCHER
714
715The L<AnyEvent> watcher object used to watch for asynchronous events (see
716L<EVENT SYSTEM>). This variable is C<undef> until L<AnyEvent> has been
717loaded I<and> initialised (e.g. by calling C<AnyEvent::detect>).
718
719=cut
720
721our $WATCHER;
722
723sub _init_anyevent {
724 $INTERRUPT->block;
725 $WATCHER = AE::io ($INTERRUPT->pipe_fileno, 0, sub { $INTERRUPT->handle });
726}
727
728if (defined $AnyEvent::MODEL) {
729 _init_anyevent;
730} else {
731 push @AnyEvent::post_detect, \&_init_anyevent;
732}
733
734=back
735
736=head2 THE OpenCL::Object CLASS
737
738This is the base class for all objects in the OpenCL module. The only
739method it implements is the C<id> method, which is only useful if you want
740to interface to OpenCL on the C level.
741
742=over 4
743
744=item $iv = $obj->id
745
746OpenCL objects are represented by pointers or integers on the C level. If
747you want to interface to an OpenCL object directly on the C level, then
748you need this value, which is returned by this method. You should use an
749C<IV> type in your code and cast that to the correct type.
750
751=cut
752
753sub OpenCL::Object::id {
754 ref $_[0] eq "SCALAR"
755 ? ${ $_[0] }
756 : $_[0][0]
757}
758
759=back
760
761=head2 THE OpenCL::Platform CLASS
762
763=over 4
764
765=item @devices = $platform->devices ($type = OpenCL::DEVICE_TYPE_ALL)
766
767Returns a list of matching OpenCL::Device objects.
768
769=item $ctx = $platform->context_from_type ($properties, $type = OpenCL::DEVICE_TYPE_DEFAULT, $callback->($err, $pvt) = $print_stderr)
770
771Tries to create a context. Never worked for me, and you need devices explicitly anyway.
772
773L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContextFromType.html>
774
775=item $ctx = $platform->context ($properties, \@devices, $callback->($err, $pvt) = $print_stderr)
776
777Create a new OpenCL::Context object using the given device object(s)- a
778OpenCL::CONTEXT_PLATFORM property is supplied automatically.
779
780L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContext.html>
781
782=item $packed_value = $platform->info ($name)
783
784Calls C<clGetPlatformInfo> and returns the packed, raw value - for
785strings, this will be the string (possibly including terminating \0), for
786other values you probably need to use the correct C<unpack>.
787
788It's best to avoid this method and use one of the following convenience
789wrappers.
790
791L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetPlatformInfo.html>
792
793=item $platform->unload_compiler
794
795Attempts to unload the compiler for this platform, for endless
796profit. Does nothing on OpenCL 1.1.
797
798L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clUnloadPlatformCompiler.html>
799
800=for gengetinfo begin platform
801
802=item $string = $platform->profile
803
804Calls C<clGetPlatformInfo> with C<OpenCL::PLATFORM_PROFILE> and returns the result.
805
806=item $string = $platform->version
807
808Calls C<clGetPlatformInfo> with C<OpenCL::PLATFORM_VERSION> and returns the result.
809
810=item $string = $platform->name
811
812Calls C<clGetPlatformInfo> with C<OpenCL::PLATFORM_NAME> and returns the result.
813
814=item $string = $platform->vendor
815
816Calls C<clGetPlatformInfo> with C<OpenCL::PLATFORM_VENDOR> and returns the result.
817
818=item $string = $platform->extensions
819
820Calls C<clGetPlatformInfo> with C<OpenCL::PLATFORM_EXTENSIONS> and returns the result.
821
822=for gengetinfo end platform
823
824=back
825
826=head2 THE OpenCL::Device CLASS
827
828=over 4
829
830=item $packed_value = $device->info ($name)
831
832See C<< $platform->info >> for details.
833
834type: OpenCL::DEVICE_TYPE_DEFAULT, OpenCL::DEVICE_TYPE_CPU,
835OpenCL::DEVICE_TYPE_GPU, OpenCL::DEVICE_TYPE_ACCELERATOR,
836OpenCL::DEVICE_TYPE_CUSTOM, OpenCL::DEVICE_TYPE_ALL.
837
838fp_config: OpenCL::FP_DENORM, OpenCL::FP_INF_NAN, OpenCL::FP_ROUND_TO_NEAREST,
839OpenCL::FP_ROUND_TO_ZERO, OpenCL::FP_ROUND_TO_INF, OpenCL::FP_FMA,
840OpenCL::FP_SOFT_FLOAT, OpenCL::FP_CORRECTLY_ROUNDED_DIVIDE_SQRT.
841
842mem_cache_type: OpenCL::NONE, OpenCL::READ_ONLY_CACHE, OpenCL::READ_WRITE_CACHE.
843
844local_mem_type: OpenCL::LOCAL, OpenCL::GLOBAL.
845
846exec_capabilities: OpenCL::EXEC_KERNEL, OpenCL::EXEC_NATIVE_KERNEL.
847
848command_queue_properties: OpenCL::QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE,
849OpenCL::QUEUE_PROFILING_ENABLE.
850
851partition_properties: OpenCL::DEVICE_PARTITION_EQUALLY,
852OpenCL::DEVICE_PARTITION_BY_COUNTS, OpenCL::DEVICE_PARTITION_BY_COUNTS_LIST_END,
853OpenCL::DEVICE_PARTITION_BY_AFFINITY_DOMAIN.
854
855affinity_domain: OpenCL::DEVICE_AFFINITY_DOMAIN_NUMA,
856OpenCL::DEVICE_AFFINITY_DOMAIN_L4_CACHE, OpenCL::DEVICE_AFFINITY_DOMAIN_L3_CACHE,
857OpenCL::DEVICE_AFFINITY_DOMAIN_L2_CACHE, OpenCL::DEVICE_AFFINITY_DOMAIN_L1_CACHE,
858OpenCL::DEVICE_AFFINITY_DOMAIN_NEXT_PARTITIONABLE.
859
860L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetDeviceInfo.html>
861
862=item @devices = $device->sub_devices (\@properties)
863
864Creates OpencL::SubDevice objects by partitioning an existing device.
865
866L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clCreateSubDevices.html>
867
868=for gengetinfo begin device
869
870=item $device_type = $device->type
871
872Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_TYPE> and returns the result.
873
874=item $uint = $device->vendor_id
875
876Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_VENDOR_ID> and returns the result.
877
878=item $uint = $device->max_compute_units
879
880Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_COMPUTE_UNITS> and returns the result.
881
882=item $uint = $device->max_work_item_dimensions
883
884Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_WORK_ITEM_DIMENSIONS> and returns the result.
885
886=item $int = $device->max_work_group_size
887
888Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_WORK_GROUP_SIZE> and returns the result.
889
890=item @ints = $device->max_work_item_sizes
891
892Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_WORK_ITEM_SIZES> and returns the result.
893
894=item $uint = $device->preferred_vector_width_char
895
896Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PREFERRED_VECTOR_WIDTH_CHAR> and returns the result.
897
898=item $uint = $device->preferred_vector_width_short
899
900Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PREFERRED_VECTOR_WIDTH_SHORT> and returns the result.
901
902=item $uint = $device->preferred_vector_width_int
903
904Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PREFERRED_VECTOR_WIDTH_INT> and returns the result.
905
906=item $uint = $device->preferred_vector_width_long
907
908Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PREFERRED_VECTOR_WIDTH_LONG> and returns the result.
909
910=item $uint = $device->preferred_vector_width_float
911
912Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT> and returns the result.
913
914=item $uint = $device->preferred_vector_width_double
915
916Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE> and returns the result.
917
918=item $uint = $device->max_clock_frequency
919
920Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_CLOCK_FREQUENCY> and returns the result.
921
922=item $bitfield = $device->address_bits
923
924Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_ADDRESS_BITS> and returns the result.
925
926=item $uint = $device->max_read_image_args
927
928Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_READ_IMAGE_ARGS> and returns the result.
929
930=item $uint = $device->max_write_image_args
931
932Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_WRITE_IMAGE_ARGS> and returns the result.
933
934=item $ulong = $device->max_mem_alloc_size
935
936Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_MEM_ALLOC_SIZE> and returns the result.
937
938=item $int = $device->image2d_max_width
939
940Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_IMAGE2D_MAX_WIDTH> and returns the result.
941
942=item $int = $device->image2d_max_height
943
944Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_IMAGE2D_MAX_HEIGHT> and returns the result.
945
946=item $int = $device->image3d_max_width
947
948Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_IMAGE3D_MAX_WIDTH> and returns the result.
949
950=item $int = $device->image3d_max_height
951
952Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_IMAGE3D_MAX_HEIGHT> and returns the result.
953
954=item $int = $device->image3d_max_depth
955
956Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_IMAGE3D_MAX_DEPTH> and returns the result.
957
958=item $uint = $device->image_support
959
960Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_IMAGE_SUPPORT> and returns the result.
961
962=item $int = $device->max_parameter_size
963
964Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_PARAMETER_SIZE> and returns the result.
965
966=item $uint = $device->max_samplers
967
968Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_SAMPLERS> and returns the result.
969
970=item $uint = $device->mem_base_addr_align
971
972Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MEM_BASE_ADDR_ALIGN> and returns the result.
973
974=item $uint = $device->min_data_type_align_size
975
976Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MIN_DATA_TYPE_ALIGN_SIZE> and returns the result.
977
978=item $device_fp_config = $device->single_fp_config
979
980Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_SINGLE_FP_CONFIG> and returns the result.
981
982=item $device_mem_cache_type = $device->global_mem_cache_type
983
984Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_GLOBAL_MEM_CACHE_TYPE> and returns the result.
985
986=item $uint = $device->global_mem_cacheline_size
987
988Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_GLOBAL_MEM_CACHELINE_SIZE> and returns the result.
989
990=item $ulong = $device->global_mem_cache_size
991
992Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_GLOBAL_MEM_CACHE_SIZE> and returns the result.
993
994=item $ulong = $device->global_mem_size
995
996Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_GLOBAL_MEM_SIZE> and returns the result.
997
998=item $ulong = $device->max_constant_buffer_size
999
1000Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_CONSTANT_BUFFER_SIZE> and returns the result.
1001
1002=item $uint = $device->max_constant_args
1003
1004Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_MAX_CONSTANT_ARGS> and returns the result.
1005
1006=item $device_local_mem_type = $device->local_mem_type
1007
1008Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_LOCAL_MEM_TYPE> and returns the result.
1009
1010=item $ulong = $device->local_mem_size
1011
1012Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_LOCAL_MEM_SIZE> and returns the result.
1013
1014=item $boolean = $device->error_correction_support
1015
1016Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_ERROR_CORRECTION_SUPPORT> and returns the result.
1017
1018=item $int = $device->profiling_timer_resolution
1019
1020Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PROFILING_TIMER_RESOLUTION> and returns the result.
1021
1022=item $boolean = $device->endian_little
1023
1024Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_ENDIAN_LITTLE> and returns the result.
1025
1026=item $boolean = $device->available
1027
1028Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_AVAILABLE> and returns the result.
1029
1030=item $boolean = $device->compiler_available
1031
1032Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_COMPILER_AVAILABLE> and returns the result.
1033
1034=item $device_exec_capabilities = $device->execution_capabilities
1035
1036Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_EXECUTION_CAPABILITIES> and returns the result.
1037
1038=item $command_queue_properties = $device->properties
1039
1040Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_QUEUE_PROPERTIES> and returns the result.
1041
1042=item $ = $device->platform
1043
1044Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PLATFORM> and returns the result.
1045
1046=item $string = $device->name
1047
1048Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_NAME> and returns the result.
1049
1050=item $string = $device->vendor
1051
1052Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_VENDOR> and returns the result.
1053
1054=item $string = $device->driver_version
1055
1056Calls C<clGetDeviceInfo> with C<OpenCL::DRIVER_VERSION> and returns the result.
1057
1058=item $string = $device->profile
1059
1060Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PROFILE> and returns the result.
1061
1062=item $string = $device->version
1063
1064Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_VERSION> and returns the result.
1065
1066=item $string = $device->extensions
1067
1068Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_EXTENSIONS> and returns the result.
1069
1070=item $uint = $device->preferred_vector_width_half
1071
1072Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PREFERRED_VECTOR_WIDTH_HALF> and returns the result.
1073
1074=item $uint = $device->native_vector_width_char
1075
1076Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_NATIVE_VECTOR_WIDTH_CHAR> and returns the result.
1077
1078=item $uint = $device->native_vector_width_short
1079
1080Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_NATIVE_VECTOR_WIDTH_SHORT> and returns the result.
1081
1082=item $uint = $device->native_vector_width_int
1083
1084Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_NATIVE_VECTOR_WIDTH_INT> and returns the result.
1085
1086=item $uint = $device->native_vector_width_long
1087
1088Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_NATIVE_VECTOR_WIDTH_LONG> and returns the result.
1089
1090=item $uint = $device->native_vector_width_float
1091
1092Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_NATIVE_VECTOR_WIDTH_FLOAT> and returns the result.
1093
1094=item $uint = $device->native_vector_width_double
1095
1096Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE> and returns the result.
1097
1098=item $uint = $device->native_vector_width_half
1099
1100Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_NATIVE_VECTOR_WIDTH_HALF> and returns the result.
1101
1102=item $device_fp_config = $device->double_fp_config
1103
1104Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_DOUBLE_FP_CONFIG> and returns the result.
1105
1106=item $device_fp_config = $device->half_fp_config
1107
1108Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_HALF_FP_CONFIG> and returns the result.
1109
1110=item $boolean = $device->host_unified_memory
1111
1112Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_HOST_UNIFIED_MEMORY> and returns the result.
1113
1114=item $device = $device->parent_device_ext
1115
1116Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PARENT_DEVICE_EXT> and returns the result.
1117
1118=item @device_partition_property_exts = $device->partition_types_ext
1119
1120Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PARTITION_TYPES_EXT> and returns the result.
1121
1122=item @device_partition_property_exts = $device->affinity_domains_ext
1123
1124Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_AFFINITY_DOMAINS_EXT> and returns the result.
1125
1126=item $uint = $device->reference_count_ext
1127
1128Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_REFERENCE_COUNT_EXT> and returns the result.
1129
1130=item @device_partition_property_exts = $device->partition_style_ext
1131
1132Calls C<clGetDeviceInfo> with C<OpenCL::DEVICE_PARTITION_STYLE_EXT> and returns the result.
1133
1134=for gengetinfo end device
1135
1136=back
1137
1138=head2 THE OpenCL::Context CLASS
1139
1140An OpenCL::Context is basically a container, or manager, for a number of
1141devices of a platform. It is used to create all sorts of secondary objects
1142such as buffers, queues, programs and so on.
1143
1144All context creation functions and methods take a list of properties
1145(type-value pairs). All property values can be specified as integers -
1146some additionally support other types:
1147
1148=over 4
1149
1150=item OpenCL::CONTEXT_PLATFORM
1151
1152Also accepts OpenCL::Platform objects.
1153
1154=item OpenCL::GLX_DISPLAY_KHR
1155
1156Also accepts C<undef>, in which case a deep and troubling hack is engaged
1157to find the current glx display (see L<GLX SUPPORT>).
1158
1159=item OpenCL::GL_CONTEXT_KHR
1160
1161Also accepts C<undef>, in which case a deep and troubling hack is engaged
1162to find the current glx context (see L<GLX SUPPORT>).
1163
1164=back
1165
1166=over 4
1167
1168=item $prog = $ctx->build_program ($program, $options = "")
1169
1170This convenience function tries to build the program on all devices in
1171the context. If the build fails, then the function will C<croak> with the
1172build log. Otherwise ti returns the program object.
1173
1174The C<$program> can either be a C<OpenCL::Program> object or a string
1175containing the program. In the latter case, a program objetc will be
1176created automatically.
1177
1178=cut
1179
1180sub OpenCL::Context::build_program {
1181 my ($self, $prog, $options) = @_;
1182
1183 $prog = $self->program_with_source ($prog)
1184 unless ref $prog;
1185
1186 eval { $prog->build (undef, $options); 1 }
1187 or errno == BUILD_PROGRAM_FAILURE
1188 or errno == INVALID_BINARY # workaround nvidia bug
1189 or Carp::croak "OpenCL::Context->build_program: " . err2str;
1190
1191 # we check status for all devices
1192 for my $dev ($self->devices) {
1193 $prog->build_status ($dev) == BUILD_SUCCESS
1194 or Carp::croak "Building OpenCL program for device '" . $dev->name . "' failed:\n"
1195 . $prog->build_log ($dev);
1196 }
1197
1198 $prog
1199}
1200
1201=item $queue = $ctx->queue ($device, $properties)
1202
1203Create a new OpenCL::Queue object from the context and the given device.
1204
1205L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateCommandQueue.html>
1206
1207Example: create an out-of-order queue.
1208
1209 $queue = $ctx->queue ($device, OpenCL::QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE);
1210
1211=item $ev = $ctx->user_event
1212
1213Creates a new OpenCL::UserEvent object.
1214
1215L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateUserEvent.html>
1216
1217=item $buf = $ctx->buffer ($flags, $len)
1218
1219Creates a new OpenCL::Buffer (actually OpenCL::BufferObj) object with the
1220given flags and octet-size.
1221
1222flags: OpenCL::MEM_READ_WRITE, OpenCL::MEM_WRITE_ONLY, OpenCL::MEM_READ_ONLY,
1223OpenCL::MEM_USE_HOST_PTR, OpenCL::MEM_ALLOC_HOST_PTR, OpenCL::MEM_COPY_HOST_PTR,
1224OpenCL::MEM_HOST_WRITE_ONLY, OpenCL::MEM_HOST_READ_ONLY, OpenCL::MEM_HOST_NO_ACCESS.
1225
1226L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateBuffer.html>
1227
1228A somewhat informative thread on the flags is:
1229
1230L<http://www.khronos.org/message_boards/viewtopic.php?f=28&t=2440>
1231
1232=item $buf = $ctx->buffer_sv ($flags, $data)
1233
1234Creates a new OpenCL::Buffer (actually OpenCL::BufferObj) object and
1235initialise it with the given data values.
1236
1237=item $img = $ctx->image ($self, $flags, $channel_order, $channel_type, $type, $width, $height, $depth = 0, $array_size = 0, $row_pitch = 0, $slice_pitch = 0, $num_mip_level = 0, $num_samples = 0, $*data = &PL_sv_undef)
1238
1239Creates a new OpenCL::Image object and optionally initialises it with
1240the given data values.
1241
1242channel_order: OpenCL::R, OpenCL::A, OpenCL::RG, OpenCL::RA, OpenCL::RGB,
1243OpenCL::RGBA, OpenCL::BGRA, OpenCL::ARGB, OpenCL::INTENSITY, OpenCL::LUMINANCE,
1244OpenCL::Rx, OpenCL::RGx, OpenCL::RGBx.
1245
1246channel_type: OpenCL::SNORM_INT8, OpenCL::SNORM_INT16, OpenCL::UNORM_INT8,
1247OpenCL::UNORM_INT16, OpenCL::UNORM_SHORT_565, OpenCL::UNORM_SHORT_555,
1248OpenCL::UNORM_INT_101010, OpenCL::SIGNED_INT8, OpenCL::SIGNED_INT16,
1249OpenCL::SIGNED_INT32, OpenCL::UNSIGNED_INT8, OpenCL::UNSIGNED_INT16,
1250OpenCL::UNSIGNED_INT32, OpenCL::HALF_FLOAT, OpenCL::FLOAT.
1251
1252type: OpenCL::MEM_OBJECT_BUFFER, OpenCL::MEM_OBJECT_IMAGE2D,
1253OpenCL::MEM_OBJECT_IMAGE3D, OpenCL::MEM_OBJECT_IMAGE2D_ARRAY,
1254OpenCL::MEM_OBJECT_IMAGE1D, OpenCL::MEM_OBJECT_IMAGE1D_ARRAY,
1255OpenCL::MEM_OBJECT_IMAGE1D_BUFFER.
1256
1257L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clCreateImage.html>
1258
1259=item $img = $ctx->image2d ($flags, $channel_order, $channel_type, $width, $height, $row_pitch = 0, $data = undef)
1260
1261Creates a new OpenCL::Image2D object and optionally initialises it with
1262the given data values.
1263
1264L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateImage2D.html>
1265
1266=item $img = $ctx->image3d ($flags, $channel_order, $channel_type, $width, $height, $depth, $row_pitch = 0, $slice_pitch = 0, $data = undef)
1267
1268Creates a new OpenCL::Image3D object and optionally initialises it with
1269the given data values.
1270
1271L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateImage3D.html>
1272
1273=item $buffer = $ctx->gl_buffer ($flags, $bufobj)
1274
1275Creates a new OpenCL::Buffer (actually OpenCL::BufferObj) object that refers to the given
1276OpenGL buffer object.
1277
1278flags: OpenCL::MEM_READ_WRITE, OpenCL::MEM_READ_ONLY, OpenCL::MEM_WRITE_ONLY.
1279
1280http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateFromGLBuffer.html
1281
1282=item $img = $ctx->gl_texture ($flags, $target, $miplevel, $texture)
1283
1284Creates a new OpenCL::Image object that refers to the given OpenGL
1285texture object or buffer.
1286
1287target: GL_TEXTURE_1D, GL_TEXTURE_1D_ARRAY, GL_TEXTURE_BUFFER,
1288GL_TEXTURE_2D, GL_TEXTURE_2D_ARRAY, GL_TEXTURE_3D,
1289GL_TEXTURE_CUBE_MAP_POSITIVE_X, GL_TEXTURE_CUBE_MAP_POSITIVE_Y,
1290GL_TEXTURE_CUBE_MAP_POSITIVE_Z, GL_TEXTURE_CUBE_MAP_NEGATIVE_X,
1291GL_TEXTURE_CUBE_MAP_NEGATIVE_Y, GL_TEXTURE_CUBE_MAP_NEGATIVE_Z,
1292GL_TEXTURE_RECTANGLE/GL_TEXTURE_RECTANGLE_ARB.
1293
1294http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clCreateFromGLTexture.html
1295
1296=item $img = $ctx->gl_texture2d ($flags, $target, $miplevel, $texture)
1297
1298Creates a new OpenCL::Image2D object that refers to the given OpenGL
12992D texture object.
1300
1301http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateFromGLTexture2D.html
1302
1303=item $img = $ctx->gl_texture3d ($flags, $target, $miplevel, $texture)
1304
1305Creates a new OpenCL::Image3D object that refers to the given OpenGL
13063D texture object.
1307
1308http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateFromGLTexture3D.html
1309
1310=item $ctx->gl_renderbuffer ($flags, $renderbuffer)
1311
1312Creates a new OpenCL::Image2D object that refers to the given OpenGL
1313render buffer.
1314
1315http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateFromGLRenderbuffer.html
1316
1317=item @formats = $ctx->supported_image_formats ($flags, $image_type)
1318
1319Returns a list of matching image formats - each format is an arrayref with
1320two values, $channel_order and $channel_type, in it.
1321
1322L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetSupportedImageFormats.html>
1323
1324=item $sampler = $ctx->sampler ($normalized_coords, $addressing_mode, $filter_mode)
1325
1326Creates a new OpenCL::Sampler object.
1327
1328addressing_mode: OpenCL::ADDRESS_NONE, OpenCL::ADDRESS_CLAMP_TO_EDGE,
1329OpenCL::ADDRESS_CLAMP, OpenCL::ADDRESS_REPEAT, OpenCL::ADDRESS_MIRRORED_REPEAT.
1330
1331filter_mode: OpenCL::FILTER_NEAREST, OpenCL::FILTER_LINEAR.
1332
1333L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateSampler.html>
1334
1335=item $program = $ctx->program_with_source ($string)
1336
1337Creates a new OpenCL::Program object from the given source code.
1338
1339L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateProgramWithSource.html>
1340
1341=item ($program, \@status) = $ctx->program_with_binary (\@devices, \@binaries)
1342
1343Creates a new OpenCL::Program object from the given binaries.
1344
1345L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateProgramWithBinary.html>
1346
1347Example: clone an existing program object that contains a successfully
1348compiled program, no matter how useless this is.
1349
1350 my $clone = $ctx->program_with_binary ([$prog->devices], [$prog->binaries]);
1351
1352=item $program = $ctx->program_with_built_in_kernels (\@devices, $kernel_names)
1353
1354Creates a new OpenCL::Program object from the given built-in kernel names.
1355
1356L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clCreateProgramWithBuiltInKernels.html>
1357
1358=item $program = $ctx->link_program (\@devices, $options, \@programs, $cb->($program) = undef)
1359
1360Links all (already compiled) program objects specified in C<@programs>
1361together and returns a new OpenCL::Program object with the result.
1362
1363L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clLinkProgram.html>
1364
1365=item $packed_value = $ctx->info ($name)
1366
1367See C<< $platform->info >> for details.
1368
1369L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetContextInfo.html>
1370
1371=for gengetinfo begin context
1372
1373=item $uint = $context->reference_count
1374
1375Calls C<clGetContextInfo> with C<OpenCL::CONTEXT_REFERENCE_COUNT> and returns the result.
1376
1377=item @devices = $context->devices
1378
1379Calls C<clGetContextInfo> with C<OpenCL::CONTEXT_DEVICES> and returns the result.
1380
1381=item @property_ints = $context->properties
1382
1383Calls C<clGetContextInfo> with C<OpenCL::CONTEXT_PROPERTIES> and returns the result.
1384
1385=item $uint = $context->num_devices
1386
1387Calls C<clGetContextInfo> with C<OpenCL::CONTEXT_NUM_DEVICES> and returns the result.
1388
1389=for gengetinfo end context
1390
1391=back
1392
1393=head2 THE OpenCL::Queue CLASS
1394
1395An OpenCL::Queue represents an execution queue for OpenCL. You execute
1396requests by calling their respective method and waiting for it to complete
1397in some way.
1398
1399Most methods that enqueue some request return an event object that can
1400be used to wait for completion (optionally using a callback), unless
1401the method is called in void context, in which case no event object is
1402created.
1403
1404They also allow you to specify any number of other event objects that this
1405request has to wait for before it starts executing, by simply passing the
1406event objects as extra parameters to the enqueue methods. To simplify
1407program design, this module ignores any C<undef> values in the list of
1408events. This makes it possible to code operations such as this, without
1409having to put a valid event object into C<$event> first:
1410
1411 $event = $queue->xxx (..., $event);
1412
1413Queues execute in-order by default, without any parallelism, so in most
1414cases (i.e. you use only one queue) it's not necessary to wait for or
1415create event objects, althoguh an our of order queue is often a bit
1416faster.
1417
1418=over 4
1419
1420=item $ev = $queue->read_buffer ($buffer, $blocking, $offset, $len, $data, $wait_events...)
1421
1422Reads data from buffer into the given string.
1423
1424L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReadBuffer.html>
1425
1426=item $ev = $queue->write_buffer ($buffer, $blocking, $offset, $data, $wait_events...)
1427
1428Writes data to buffer from the given string.
1429
1430L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWriteBuffer.html>
1431
1432=item $ev = $queue->copy_buffer ($src, $dst, $src_offset, $dst_offset, $len, $wait_events...)
1433
1434L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyBuffer.html>
1435
1436$eue->read_buffer_rect ($buf, cl_bool blocking, $buf_x, $buf_y, $buf_z, $host_x, $host_y, $host_z, $width, $height, $depth, $buf_row_pitch, $buf_slice_pitch, $host_row_pitch, $host_slice_pitch, $data, $wait_events...)
1437
1438http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReadBufferRect.html
1439
1440=item $ev = $queue->write_buffer_rect ($buf, $blocking, $buf_y, $host_x, $host_z, $height, $buf_row_pitch, $host_row_pitch, $data, $wait_events...)
1441
1442http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWriteBufferRect.html
1443
1444=item $ev = $queue->copy_buffer_to_image ($src_buffer, $dst_image, $src_offset, $dst_x, $dst_y, $dst_z, $width, $height, $depth, $wait_events...)
1445
1446L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyBufferToImage.html>
1447
1448=item $ev = $queue->read_image ($src, $blocking, $x, $y, $z, $width, $height, $depth, $row_pitch, $slice_pitch, $data, $wait_events...)
1449
1450C<$row_pitch> (and C<$slice_pitch>) can be C<0>, in which case the OpenCL
1451module uses the image width (and height) to supply default values.
1452
1453L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReadImage.html>
1454
1455=item $ev = $queue->write_image ($src, $blocking, $x, $y, $z, $width, $height, $depth, $row_pitch, $slice_pitch, $data, $wait_events...)
1456
1457C<$row_pitch> (and C<$slice_pitch>) can be C<0>, in which case the OpenCL
1458module uses the image width (and height) to supply default values.
1459L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWriteImage.html>
1460
1461=item $ev = $queue->copy_image ($src_image, $dst_image, $src_x, $src_y, $src_z, $dst_x, $dst_y, $dst_z, $width, $height, $depth, $wait_events...)
1462
1463L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyImage.html>
1464
1465=item $ev = $queue->copy_image_to_buffer ($src_image, $dst_image, $src_x, $src_y, $src_z, $width, $height, $depth, $dst_offset, $wait_events...)
1466
1467L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyImageToBuffer.html>
1468
1469=item $ev = $queue->copy_buffer_rect ($src, $dst, $src_x, $src_y, $src_z, $dst_x, $dst_y, $dst_z, $width, $height, $depth, $src_row_pitch, $src_slice_pitch, $dst_row_pitch, $dst_slice_pitch, $wait_event...)
1470
1471Yeah.
1472
1473L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyBufferToImage.html>.
1474
1475=item $ev = $queue->fill_buffer ($mem, $pattern, $offset, $size, ...)
1476
1477Fills the given buffer object with repeated applications of C<$pattern>,
1478starting at C<$offset> for C<$size> octets.
1479
1480L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clEnqueueFillBuffer.html>
1481
1482=item $ev = $queue->fill_image ($img, $r, $g, $b, $a, $x, $y, $z, $width, $height, $depth, ...)
1483
1484Fills the given image area with the given rgba colour components. The
1485components are normally floating point values between C<0> and C<1>,
1486except when the image channel data type is a signe dor unsigned
1487unnormalised format, in which case the range is determined by the format.
1488
1489L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clEnqueueFillImage.html>
1490
1491=item $ev = $queue->task ($kernel, $wait_events...)
1492
1493L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueTask.html>
1494
1495=item $ev = $queue->nd_range_kernel ($kernel, \@global_work_offset, \@global_work_size, \@local_work_size, $wait_events...)
1496
1497Enqueues a kernel execution.
1498
1499\@global_work_size must be specified as a reference to an array of
1500integers specifying the work sizes (element counts).
1501
1502\@global_work_offset must be either C<undef> (in which case all offsets
1503are C<0>), or a reference to an array of work offsets, with the same number
1504of elements as \@global_work_size.
1505
1506\@local_work_size must be either C<undef> (in which case the
1507implementation is supposed to choose good local work sizes), or a
1508reference to an array of local work sizes, with the same number of
1509elements as \@global_work_size.
1510
1511L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueNDRangeKernel.html>
1512
1513=item $ev = $queue->migrate_mem_objects (\@mem_objects, $flags, $wait_events...)
1514
1515Migrates a number of OpenCL::Memory objects to or from the device.
1516
1517flags: OpenCL::MIGRATE_MEM_OBJECT_HOST, OpenCL::MIGRATE_MEM_OBJECT_CONTENT_UNDEFINED
1518
1519L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clEnqueueMigrateMemObjects.html>
1520
1521=item $ev = $queue->acquire_gl_objects ([object, ...], $wait_events...)
1522
1523Enqueues a list (an array-ref of OpenCL::Memory objects) to be acquired
1524for subsequent OpenCL usage.
1525
1526L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueAcquireGLObjects.html>
1527
1528=item $ev = $queue->release_gl_objects ([object, ...], $wait_events...)
1529
1530Enqueues a list (an array-ref of OpenCL::Memory objects) to be released
1531for subsequent OpenGL usage.
1532
1533L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReleaseGLObjects.html>
1534
1535=item $ev = $queue->wait_for_events ($wait_events...)
1536
1537L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWaitForEvents.html>
1538
1539=item $ev = $queue->marker ($wait_events...)
1540
1541L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clEnqueueMarkerWithWaitList.html>
1542
1543=item $ev = $queue->barrier ($wait_events...)
1544
1545L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clEnqueueBarrierWithWaitList.html>
1546
1547=item $queue->flush
1548
1549L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clFlush.html>
1550
1551=item $queue->finish
1552
1553L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clFinish.html>
1554
1555=item $packed_value = $queue->info ($name)
1556
1557See C<< $platform->info >> for details.
1558
1559L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetCommandQueueInfo.html>
1560
1561=for gengetinfo begin command_queue
1562
1563=item $ctx = $command_queue->context
1564
1565Calls C<clGetCommandQueueInfo> with C<OpenCL::QUEUE_CONTEXT> and returns the result.
1566
1567=item $device = $command_queue->device
1568
1569Calls C<clGetCommandQueueInfo> with C<OpenCL::QUEUE_DEVICE> and returns the result.
1570
1571=item $uint = $command_queue->reference_count
1572
1573Calls C<clGetCommandQueueInfo> with C<OpenCL::QUEUE_REFERENCE_COUNT> and returns the result.
1574
1575=item $command_queue_properties = $command_queue->properties
1576
1577Calls C<clGetCommandQueueInfo> with C<OpenCL::QUEUE_PROPERTIES> and returns the result.
1578
1579=for gengetinfo end command_queue
1580
1581=back
1582
1583=head3 MEMORY MAPPED BUFFERS
1584
1585OpenCL allows you to map buffers and images to host memory (read: perl
1586scalars). This is done much like reading or copying a buffer, by enqueuing
1587a map or unmap operation on the command queue.
1588
1589The map operations return an C<OpenCL::Mapped> object - see L<THE
1590OpenCL::Mapped CLASS> section for details on what to do with these
1591objects.
1592
1593The object will be unmapped automatically when the mapped object is
1594destroyed (you can use a barrier to make sure the unmap has finished,
1595before using the buffer in a kernel), but you can also enqueue an unmap
1596operation manually.
1597
1598=over 4
1599
1600=item $mapped_buffer = $queue->map_buffer ($buf, $blocking=1, $map_flags=OpenCL::MAP_READ|OpenCL::MAP_WRITE, $offset=0, $size=undef, $wait_events...)
1601
1602Maps the given buffer into host memory and returns an
1603C<OpenCL::MappedBuffer> object. If C<$size> is specified as undef, then
1604the map will extend to the end of the buffer.
1605
1606map_flags: OpenCL::MAP_READ, OpenCL::MAP_WRITE, OpenCL::MAP_WRITE_INVALIDATE_REGION.
1607
1608L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueMapBuffer.html>
1609
1610Example: map the buffer $buf fully and replace the first 4 bytes by "abcd", then unmap.
1611
1612 {
1613 my $mapped = $queue->map_buffer ($buf, 1, OpenCL::MAP_WRITE);
1614 substr $$mapped, 0, 4, "abcd";
1615 } # asynchronously unmap because $mapped is destroyed
1616
1617=item $mapped_image = $queue->map_image ($img, $blocking=1, $map_flags=OpenCL::MAP_READ|OpenCL::MAP_WRITE, $x=0, $y=0, $z=0, $width=undef, $height=undef, $depth=undef, $wait_events...)
1618
1619Maps the given image area into host memory and return an
1620C<OpenCL::MappedImage> object.
1621
1622If any of C<$width>, C<$height> and/or C<$depth> are C<undef> then they
1623will be replaced by the maximum possible value.
1624
1625L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueMapImage.html>
1626
1627Example: map an image (with OpenCL::UNSIGNED_INT8 channel type) and set
1628the first channel of the leftmost column to 5, then explicitly unmap
1629it. You are not necessarily meant to do it this way, this example just
1630shows you the accessors to use :)
1631
1632 my $mapped = $queue->map_image ($image, 1, OpenCL::MAP_WRITE);
1633
1634 $mapped->write ($_ * $mapped->row_pitch, pack "C", 5)
1635 for 0 .. $mapped->height - 1;
1636
1637 $mapped->unmap;.
1638 $mapped->wait; # only needed for out of order queues normally
1639
1640=item $ev = $queue->unmap ($mapped, $wait_events...)
1641
1642Unmaps the data from host memory. You must not call any methods that
1643modify the data, or modify the data scalar directly, after calling this
1644method.
1645
1646The mapped event object will always be passed as part of the
1647$wait_events. The mapped event object will be replaced by the new event
1648object that this request creates.
1649
1650=back
1651
1652=head2 THE OpenCL::Memory CLASS
1653
1654This the superclass of all memory objects - OpenCL::Buffer, OpenCL::Image,
1655OpenCL::Image2D and OpenCL::Image3D.
1656
1657=over 4
1658
1659=item $packed_value = $memory->info ($name)
1660
1661See C<< $platform->info >> for details.
1662
1663L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetMemObjectInfo.html>
1664
1665=item $memory->destructor_callback ($cb->())
1666
1667Sets a callback that will be invoked after the memory object is destructed.
1668
1669L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clSetMemObjectDestructorCallback.html>
1670
1671=for gengetinfo begin mem
1672
1673=item $mem_object_type = $mem->type
1674
1675Calls C<clGetMemObjectInfo> with C<OpenCL::MEM_TYPE> and returns the result.
1676
1677=item $mem_flags = $mem->flags
1678
1679Calls C<clGetMemObjectInfo> with C<OpenCL::MEM_FLAGS> and returns the result.
1680
1681=item $int = $mem->size
1682
1683Calls C<clGetMemObjectInfo> with C<OpenCL::MEM_SIZE> and returns the result.
1684
1685=item $ptr_value = $mem->host_ptr
1686
1687Calls C<clGetMemObjectInfo> with C<OpenCL::MEM_HOST_PTR> and returns the result.
1688
1689=item $uint = $mem->map_count
1690
1691Calls C<clGetMemObjectInfo> with C<OpenCL::MEM_MAP_COUNT> and returns the result.
1692
1693=item $uint = $mem->reference_count
1694
1695Calls C<clGetMemObjectInfo> with C<OpenCL::MEM_REFERENCE_COUNT> and returns the result.
1696
1697=item $ctx = $mem->context
1698
1699Calls C<clGetMemObjectInfo> with C<OpenCL::MEM_CONTEXT> and returns the result.
1700
1701=item $mem = $mem->associated_memobject
1702
1703Calls C<clGetMemObjectInfo> with C<OpenCL::MEM_ASSOCIATED_MEMOBJECT> and returns the result.
1704
1705=item $int = $mem->offset
1706
1707Calls C<clGetMemObjectInfo> with C<OpenCL::MEM_OFFSET> and returns the result.
1708
1709=for gengetinfo end mem
1710
1711=item ($type, $name) = $mem->gl_object_info
1712
1713Returns the OpenGL object type (e.g. OpenCL::GL_OBJECT_TEXTURE2D) and the
1714object "name" (e.g. the texture name) used to create this memory object.
1715
1716L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetGLObjectInfo.html>
1717
1718=back
1719
1720=head2 THE OpenCL::Buffer CLASS
1721
1722This is a subclass of OpenCL::Memory, and the superclass of
1723OpenCL::BufferObj. Its purpose is simply to distinguish between buffers
1724and sub-buffers.
1725
1726=head2 THE OpenCL::BufferObj CLASS
1727
1728This is a subclass of OpenCL::Buffer and thus OpenCL::Memory. It exists
1729because one cna create sub buffers of OpenLC::BufferObj objects, but not
1730sub buffers from these sub buffers.
1731
1732=over 4
1733
1734=item $subbuf = $buf_obj->sub_buffer_region ($flags, $origin, $size)
1735
1736Creates an OpenCL::Buffer objects from this buffer and returns it. The
1737C<buffer_create_type> is assumed to be C<OpenCL::BUFFER_CREATE_TYPE_REGION>.
1738
1739L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateSubBuffer.html>
1740
1741=back
1742
1743=head2 THE OpenCL::Image CLASS
1744
1745This is the superclass of all image objects - OpenCL::Image1D,
1746OpenCL::Image1DArray, OpenCL::Image1DBuffer, OpenCL::Image2D,
1747OpenCL::Image2DArray and OpenCL::Image3D.
1748
1749=over 4
1750
1751=item $packed_value = $image->image_info ($name)
1752
1753See C<< $platform->info >> for details.
1754
1755The reason this method is not called C<info> is that there already is an
1756C<< ->info >> method inherited from C<OpenCL::Memory>.
1757
1758L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetImageInfo.html>
1759
1760=item ($channel_order, $channel_data_type) = $image->format
1761
1762Returns the channel order and type used to create the image by calling
1763C<clGetImageInfo> with C<OpenCL::IMAGE_FORMAT>.
1764
1765=for gengetinfo begin image
1766
1767=item $int = $image->element_size
1768
1769Calls C<clGetImageInfo> with C<OpenCL::IMAGE_ELEMENT_SIZE> and returns the result.
1770
1771=item $int = $image->row_pitch
1772
1773Calls C<clGetImageInfo> with C<OpenCL::IMAGE_ROW_PITCH> and returns the result.
1774
1775=item $int = $image->slice_pitch
1776
1777Calls C<clGetImageInfo> with C<OpenCL::IMAGE_SLICE_PITCH> and returns the result.
1778
1779=item $int = $image->width
1780
1781Calls C<clGetImageInfo> with C<OpenCL::IMAGE_WIDTH> and returns the result.
1782
1783=item $int = $image->height
1784
1785Calls C<clGetImageInfo> with C<OpenCL::IMAGE_HEIGHT> and returns the result.
1786
1787=item $int = $image->depth
1788
1789Calls C<clGetImageInfo> with C<OpenCL::IMAGE_DEPTH> and returns the result.
1790
1791=for gengetinfo end image
1792
1793=for gengetinfo begin gl_texture
1794
1795=item $GLenum = $gl_texture->target
1796
1797Calls C<clGetGLTextureInfo> with C<OpenCL::GL_TEXTURE_TARGET> and returns the result.
1798
1799=item $GLint = $gl_texture->gl_mipmap_level
1800
1801Calls C<clGetGLTextureInfo> with C<OpenCL::GL_MIPMAP_LEVEL> and returns the result.
1802
1803=for gengetinfo end gl_texture
1804
1805=back
1806
1807=head2 THE OpenCL::Sampler CLASS
1808
1809=over 4
1810
1811=item $packed_value = $sampler->info ($name)
1812
1813See C<< $platform->info >> for details.
1814
1815L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetSamplerInfo.html>
1816
1817=for gengetinfo begin sampler
1818
1819=item $uint = $sampler->reference_count
1820
1821Calls C<clGetSamplerInfo> with C<OpenCL::SAMPLER_REFERENCE_COUNT> and returns the result.
1822
1823=item $ctx = $sampler->context
1824
1825Calls C<clGetSamplerInfo> with C<OpenCL::SAMPLER_CONTEXT> and returns the result.
1826
1827=item $addressing_mode = $sampler->normalized_coords
1828
1829Calls C<clGetSamplerInfo> with C<OpenCL::SAMPLER_NORMALIZED_COORDS> and returns the result.
1830
1831=item $filter_mode = $sampler->addressing_mode
1832
1833Calls C<clGetSamplerInfo> with C<OpenCL::SAMPLER_ADDRESSING_MODE> and returns the result.
1834
1835=item $boolean = $sampler->filter_mode
1836
1837Calls C<clGetSamplerInfo> with C<OpenCL::SAMPLER_FILTER_MODE> and returns the result.
1838
1839=for gengetinfo end sampler
1840
1841=back
1842
1843=head2 THE OpenCL::Program CLASS
1844
1845=over 4
1846
1847=item $program->build (\@devices = undef, $options = "", $cb->($program) = undef)
1848
1849Tries to build the program with the given options. See also the
1850C<$ctx->build> convenience function.
1851
1852If a callback is specified, then it will be called when compilation is
1853finished. Note that many OpenCL implementations block your program while
1854compiling whether you use a callback or not. See C<build_async> if you
1855want to make sure the build is done in the background.
1856
1857Note that some OpenCL implementations act up badly, and don't call the
1858callback in some error cases (but call it in others). This implementation
1859assumes the callback will always be called, and leaks memory if this is
1860not so. So best make sure you don't pass in invalid values.
1861
1862Some implementations fail with C<OpenCL::INVALID_BINARY> when the
1863compilation state is successful but some later stage fails.
1864
1865options: C<-D name>, C<-D name=definition>, C<-I dir>,
1866C<-cl-single-precision-constant>, C<-cl-denorms-are-zero>,
1867C<-cl-fp32-correctly-rounded-divide-sqrt>, C<-cl-opt-disable>,
1868C<-cl-mad-enable>, C<-cl-no-signed-zeros>, C<-cl-unsafe-math-optimizations>,
1869C<-cl-finite-math-only>, C<-cl-fast-relaxed-math>,
1870C<-w>, C<-Werror>, C<-cl-std=CL1.1/CL1.2>, C<-cl-kernel-arg-info>,
1871C<-create-library>, C<-enable-link-options>.
1872
1873build_status: OpenCL::BUILD_SUCCESS, OpenCL::BUILD_NONE,
1874OpenCL::BUILD_ERROR, OpenCL::BUILD_IN_PROGRESS.
1875
1876L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clBuildProgram.html>
1877
1878=item $program->build_async (\@devices = undef, $options = "", $cb->($program) = undef)
1879
1880Similar to C<< ->build >>, except it starts a thread, and never fails (you
1881need to check the compilation status form the callback, or by polling).
1882
1883=item $program->compile (\@devices = undef, $options = "", \%headers = undef, $cb->($program) = undef)
1884
1885Compiles the given program for the given devices (or all devices if
1886undef). If C<$headers> is given, it must be a hashref with include name =>
1887OpenCL::Program pairs.
1888
1889L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clCompileProgram.html>
1890
1891=item $packed_value = $program->build_info ($device, $name)
1892
1893Similar to C<< $platform->info >>, but returns build info for a previous
1894build attempt for the given device.
1895
1896binary_type: OpenCL::PROGRAM_BINARY_TYPE_NONE,
1897OpenCL::PROGRAM_BINARY_TYPE_COMPILED_OBJECT,
1898OpenCL::PROGRAM_BINARY_TYPE_LIBRARY,
1899OpenCL::PROGRAM_BINARY_TYPE_EXECUTABLE.
1900
1901L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetBuildInfo.html>
1902
1903=item $kernel = $program->kernel ($function_name)
1904
1905Creates an OpenCL::Kernel object out of the named C<__kernel> function in
1906the program.
1907
1908L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateKernel.html>
1909
1910=item @kernels = $program->kernels_in_program
1911
1912Returns all kernels successfully compiled for all devices in program.
1913
1914http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateKernelsInProgram.html
1915
1916=for gengetinfo begin program_build
1917
1918=item $build_status = $program->build_status ($device)
1919
1920Calls C<clGetProgramBuildInfo> with C<OpenCL::PROGRAM_BUILD_STATUS> and returns the result.
1921
1922=item $string = $program->build_options ($device)
1923
1924Calls C<clGetProgramBuildInfo> with C<OpenCL::PROGRAM_BUILD_OPTIONS> and returns the result.
1925
1926=item $string = $program->build_log ($device)
1927
1928Calls C<clGetProgramBuildInfo> with C<OpenCL::PROGRAM_BUILD_LOG> and returns the result.
1929
1930=item $binary_type = $program->binary_type ($device)
1931
1932Calls C<clGetProgramBuildInfo> with C<OpenCL::PROGRAM_BINARY_TYPE> and returns the result.
1933
1934=for gengetinfo end program_build
1935
1936=item $packed_value = $program->info ($name)
1937
1938See C<< $platform->info >> for details.
1939
1940L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetProgramInfo.html>
1941
1942=for gengetinfo begin program
1943
1944=item $uint = $program->reference_count
1945
1946Calls C<clGetProgramInfo> with C<OpenCL::PROGRAM_REFERENCE_COUNT> and returns the result.
1947
1948=item $ctx = $program->context
1949
1950Calls C<clGetProgramInfo> with C<OpenCL::PROGRAM_CONTEXT> and returns the result.
1951
1952=item $uint = $program->num_devices
1953
1954Calls C<clGetProgramInfo> with C<OpenCL::PROGRAM_NUM_DEVICES> and returns the result.
1955
1956=item @devices = $program->devices
1957
1958Calls C<clGetProgramInfo> with C<OpenCL::PROGRAM_DEVICES> and returns the result.
1959
1960=item $string = $program->source
1961
1962Calls C<clGetProgramInfo> with C<OpenCL::PROGRAM_SOURCE> and returns the result.
1963
1964=item @ints = $program->binary_sizes
1965
1966Calls C<clGetProgramInfo> with C<OpenCL::PROGRAM_BINARY_SIZES> and returns the result.
1967
1968=for gengetinfo end program
1969
1970=item @blobs = $program->binaries
1971
1972Returns a string for the compiled binary for every device associated with
1973the program, empty strings indicate missing programs, and an empty result
1974means no program binaries are available.
1975
1976These "binaries" are often, in fact, informative low-level assembly
1977sources.
1978
1979L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetProgramInfo.html>
1980
1981=back
1982
1983=head2 THE OpenCL::Kernel CLASS
1984
1985=over 4
1986
1987=item $packed_value = $kernel->info ($name)
1988
1989See C<< $platform->info >> for details.
1990
1991L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetKernelInfo.html>
1992
1993=for gengetinfo begin kernel
1994
1995=item $string = $kernel->function_name
1996
1997Calls C<clGetKernelInfo> with C<OpenCL::KERNEL_FUNCTION_NAME> and returns the result.
1998
1999=item $uint = $kernel->num_args
2000
2001Calls C<clGetKernelInfo> with C<OpenCL::KERNEL_NUM_ARGS> and returns the result.
2002
2003=item $uint = $kernel->reference_count
2004
2005Calls C<clGetKernelInfo> with C<OpenCL::KERNEL_REFERENCE_COUNT> and returns the result.
2006
2007=item $ctx = $kernel->context
2008
2009Calls C<clGetKernelInfo> with C<OpenCL::KERNEL_CONTEXT> and returns the result.
2010
2011=item $program = $kernel->program
2012
2013Calls C<clGetKernelInfo> with C<OpenCL::KERNEL_PROGRAM> and returns the result.
2014
2015=for gengetinfo end kernel
2016
2017=item $packed_value = $kernel->work_group_info ($device, $name)
2018
2019See C<< $platform->info >> for details.
2020
2021L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetKernelWorkGroupInfo.html>
2022
2023=for gengetinfo begin kernel_work_group
2024
2025=item $int = $kernel->work_group_size ($device)
2026
2027Calls C<clGetKernelWorkGroupInfo> with C<OpenCL::KERNEL_WORK_GROUP_SIZE> and returns the result.
2028
2029=item @ints = $kernel->compile_work_group_size ($device)
2030
2031Calls C<clGetKernelWorkGroupInfo> with C<OpenCL::KERNEL_COMPILE_WORK_GROUP_SIZE> and returns the result.
2032
2033=item $ulong = $kernel->local_mem_size ($device)
2034
2035Calls C<clGetKernelWorkGroupInfo> with C<OpenCL::KERNEL_LOCAL_MEM_SIZE> and returns the result.
2036
2037=item $int = $kernel->preferred_work_group_size_multiple ($device)
2038
2039Calls C<clGetKernelWorkGroupInfo> with C<OpenCL::KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE> and returns the result.
2040
2041=item $ulong = $kernel->private_mem_size ($device)
2042
2043Calls C<clGetKernelWorkGroupInfo> with C<OpenCL::KERNEL_PRIVATE_MEM_SIZE> and returns the result.
2044
2045=for gengetinfo end kernel_work_group
2046
2047=item $packed_value = $kernel->arg_info ($idx, $name)
2048
2049See C<< $platform->info >> for details.
2050
2051L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clGetKernelArgInfo.html>
2052
2053=for gengetinfo begin kernel_arg
2054
2055=item $kernel_arg_address_qualifier = $kernel->arg_address_qualifier ($idx)
2056
2057Calls C<clGetKernelArgInfo> with C<OpenCL::KERNEL_ARG_ADDRESS_QUALIFIER> and returns the result.
2058
2059=item $kernel_arg_access_qualifier = $kernel->arg_access_qualifier ($idx)
2060
2061Calls C<clGetKernelArgInfo> with C<OpenCL::KERNEL_ARG_ACCESS_QUALIFIER> and returns the result.
2062
2063=item $string = $kernel->arg_type_name ($idx)
2064
2065Calls C<clGetKernelArgInfo> with C<OpenCL::KERNEL_ARG_TYPE_NAME> and returns the result.
2066
2067=item $kernel_arg_type_qualifier = $kernel->arg_type_qualifier ($idx)
2068
2069Calls C<clGetKernelArgInfo> with C<OpenCL::KERNEL_ARG_TYPE_QUALIFIER> and returns the result.
2070
2071=item $string = $kernel->arg_name ($idx)
2072
2073Calls C<clGetKernelArgInfo> with C<OpenCL::KERNEL_ARG_NAME> and returns the result.
2074
2075=for gengetinfo end kernel_arg
2076
2077=item $kernel->setf ($format, ...)
2078
2079Sets the arguments of a kernel. Since OpenCL 1.1 doesn't have a generic
2080way to set arguments (and with OpenCL 1.2 it might be rather slow), you
2081need to specify a format argument, much as with C<printf>, to tell OpenCL
2082what type of argument it is.
2083
2084The format arguments are single letters:
2085
2086 c char
2087 C unsigned char
2088 s short
2089 S unsigned short
2090 i int
2091 I unsigned int
2092 l long
2093 L unsigned long
2094
2095 h half float (0..65535)
2096 f float
2097 d double
2098
2099 z local (octet size)
2100
2101 m memory object (buffer or image)
2102 a sampler
2103 e event
2104
2105Space characters in the format string are ignored.
2106
2107Example: set the arguments for a kernel that expects an int, two floats, a buffer and an image.
2108
2109 $kernel->setf ("i ff mm", 5, 0.5, 3, $buffer, $image);
2110
2111=item $kernel->set_TYPE ($index, $value)
2112
2113=item $kernel->set_char ($index, $value)
2114
2115=item $kernel->set_uchar ($index, $value)
2116
2117=item $kernel->set_short ($index, $value)
2118
2119=item $kernel->set_ushort ($index, $value)
2120
2121=item $kernel->set_int ($index, $value)
2122
2123=item $kernel->set_uint ($index, $value)
2124
2125=item $kernel->set_long ($index, $value)
2126
2127=item $kernel->set_ulong ($index, $value)
2128
2129=item $kernel->set_half ($index, $value)
2130
2131=item $kernel->set_float ($index, $value)
2132
2133=item $kernel->set_double ($index, $value)
2134
2135=item $kernel->set_memory ($index, $value)
2136
2137=item $kernel->set_buffer ($index, $value)
2138
2139=item $kernel->set_image ($index, $value)
2140
2141=item $kernel->set_sampler ($index, $value)
2142
2143=item $kernel->set_local ($index, $value)
2144
2145=item $kernel->set_event ($index, $value)
2146
2147This is a family of methods to set the kernel argument with the number
2148C<$index> to the give C<$value>.
2149
2150Chars and integers (including the half type) are specified as integers,
2151float and double as floating point values, memory/buffer/image must be
2152an object of that type or C<undef>, local-memory arguments are set by
2153specifying the size, and sampler and event must be objects of that type.
2154
2155Note that C<set_memory> works for all memory objects (all types of buffers
2156and images) - the main purpose of the more specific C<set_TYPE> functions
2157is type checking.
2158
2159Setting an argument for a kernel does NOT keep a reference to the object -
2160for example, if you set an argument to some image object, free the image,
2161and call the kernel, you will run into undefined behaviour.
2162
2163L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clSetKernelArg.html>
2164
2165=back
2166
2167=head2 THE OpenCL::Event CLASS
2168
2169This is the superclass for all event objects (including OpenCL::UserEvent
2170objects).
2171
2172=over 4
2173
2174=item $ev->wait
2175
2176Waits for the event to complete.
2177
2178L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clWaitForEvents.html>
2179
2180=item $ev->cb ($exec_callback_type, $callback->($event, $event_command_exec_status))
2181
2182Adds a callback to the callback stack for the given event type. There is
2183no way to remove a callback again.
2184
2185L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clSetEventCallback.html>
2186
2187=item $packed_value = $ev->info ($name)
2188
2189See C<< $platform->info >> for details.
2190
2191L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetEventInfo.html>
2192
2193=for gengetinfo begin event
2194
2195=item $queue = $event->command_queue
2196
2197Calls C<clGetEventInfo> with C<OpenCL::EVENT_COMMAND_QUEUE> and returns the result.
2198
2199=item $command_type = $event->command_type
2200
2201Calls C<clGetEventInfo> with C<OpenCL::EVENT_COMMAND_TYPE> and returns the result.
2202
2203=item $uint = $event->reference_count
2204
2205Calls C<clGetEventInfo> with C<OpenCL::EVENT_REFERENCE_COUNT> and returns the result.
2206
2207=item $uint = $event->command_execution_status
2208
2209Calls C<clGetEventInfo> with C<OpenCL::EVENT_COMMAND_EXECUTION_STATUS> and returns the result.
2210
2211=item $ctx = $event->context
2212
2213Calls C<clGetEventInfo> with C<OpenCL::EVENT_CONTEXT> and returns the result.
2214
2215=for gengetinfo end event
2216
2217=item $packed_value = $ev->profiling_info ($name)
2218
2219See C<< $platform->info >> for details.
2220
2221The reason this method is not called C<info> is that there already is an
2222C<< ->info >> method.
2223
2224L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetProfilingInfo.html>
2225
2226=for gengetinfo begin profiling
2227
2228=item $ulong = $event->profiling_command_queued
2229
2230Calls C<clGetEventProfilingInfo> with C<OpenCL::PROFILING_COMMAND_QUEUED> and returns the result.
2231
2232=item $ulong = $event->profiling_command_submit
2233
2234Calls C<clGetEventProfilingInfo> with C<OpenCL::PROFILING_COMMAND_SUBMIT> and returns the result.
2235
2236=item $ulong = $event->profiling_command_start
2237
2238Calls C<clGetEventProfilingInfo> with C<OpenCL::PROFILING_COMMAND_START> and returns the result.
2239
2240=item $ulong = $event->profiling_command_end
2241
2242Calls C<clGetEventProfilingInfo> with C<OpenCL::PROFILING_COMMAND_END> and returns the result.
2243
2244=for gengetinfo end profiling
2245
2246=back
2247
2248=head2 THE OpenCL::UserEvent CLASS
2249
2250This is a subclass of OpenCL::Event.
2251
2252=over 4
2253
2254=item $ev->set_status ($execution_status)
2255
2256Sets the execution status of the user event. Can only be called once,
2257either with OpenCL::COMPLETE or a negative number as status.
2258
2259execution_status: OpenCL::COMPLETE or a negative integer.
2260
2261L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clSetUserEventStatus.html>
2262
2263=back
2264
2265=head2 THE OpenCL::Mapped CLASS
2266
2267This class represents objects mapped into host memory. They are
2268represented by a blessed string scalar. The string data is the mapped
2269memory area, that is, if you read or write it, then the mapped object is
2270accessed directly.
2271
2272You must only ever use operations that modify the string in-place - for
2273example, a C<substr> that doesn't change the length, or maybe a regex that
2274doesn't change the length. Any other operation might cause the data to be
2275copied.
2276
2277When the object is destroyed it will enqueue an implicit unmap operation
2278on the queue that was used to create it.
2279
2280Keep in mind that you I<need> to unmap (or destroy) mapped objects before
2281OpenCL sees the changes, even if some implementations don't need this
2282sometimes.
2283
2284Example, replace the first two floats in the mapped buffer by 1 and 2.
2285
2286 my $mapped = $queue->map_buffer ($buf, ...
2287 $mapped->event->wait; # make sure it's there
2288
2289 # now replace first 8 bytes by new data, which is exactly 8 bytes long
2290 # we blindly assume device endianness to equal host endianness
2291 # (and of course, we assume iee 754 single precision floats :)
2292 substr $$mapped, 0, 8, pack "f*", 1, 2;
2293
2294=over 4
2295
2296=item $ev = $mapped->unmap ($wait_events...)
2297
2298Unmaps the mapped memory object, using the queue originally used to create
2299it, quite similarly to C<< $queue->unmap ($mapped, ...) >>.
2300
2301=item $bool = $mapped->mapped
2302
2303Returns whether the object is still mapped - true before an C<unmap> is
2304enqueued, false afterwards.
2305
2306=item $ev = $mapped->event
2307
2308Return the event object associated with the mapped object. Initially, this
2309will be the event object created when mapping the object, and after an
2310unmap, this will be the event object that the unmap operation created.
2311
2312=item $mapped->wait
2313
2314Same as C<< $mapped->event->wait >> - makes sure no operations on this
2315mapped object are outstanding.
2316
2317=item $bytes = $mapped->size
2318
2319Returns the size of the mapped area, in bytes. Same as C<length $$mapped>.
2320
2321=item $ptr = $mapped->ptr
2322
2323Returns the raw memory address of the mapped area.
2324
2325=item $mapped->set ($offset, $data)
2326
2327Replaces the data at the given C<$offset> in the memory area by the new
2328C<$data>. This method is safer than direct manipulation of C<$mapped>
2329because it does bounds-checking, but also slower.
2330
2331=item $data = $mapped->get ($offset, $length)
2332
2333Returns (without copying) a scalar representing the data at the given
2334C<$offset> and C<$length> in the mapped memory area. This is the same as
2335the following substr, except much slower;
2336
2337 $data = substr $$mapped, $offset, $length
2338
2339=cut
2340
2341sub OpenCL::Mapped::get {
2342 substr ${$_[0]}, $_[1], $_[2]
2343}
2344
2345=back
2346
2347=head2 THE OpenCL::MappedBuffer CLASS
2348
2349This is a subclass of OpenCL::Mapped, representing mapped buffers.
2350
2351=head2 THE OpenCL::MappedImage CLASS
2352
2353This is a subclass of OpenCL::Mapped, representing mapped images.
2354
2355=over 4
2356
2357=item $pixels = $mapped->width
2358
2359=item $pixels = $mapped->height
2360
2361=item $pixels = $mapped->depth
2362
2363Return the width/height/depth of the mapped image region, in pixels.
2364
2365=item $bytes = $mapped->row_pitch
2366
2367=item $bytes = $mapped->slice_pitch
2368
2369Return the row or slice pitch of the image that has been mapped.
2370
2371=item $bytes = $mapped->element_size
2372
2373Return the size of a single pixel.
2374
2375=item $data = $mapped->get_row ($count, $x=0, $y=0, $z=0)
2376
2377Return C<$count> pixels from the given coordinates. The pixel data must
2378be completely contained within a single row.
2379
2380If C<$count> is C<undef>, then all the remaining pixels in that row are
2381returned.
2382
2383=item $mapped->set_row ($data, $x=0, $y=0, $z=0)
2384
2385Write the given pixel data at the given coordinate. The pixel data must
2386be completely contained within a single row.
2387
2388=back
2389
2390=cut
2391
461; 23921;
47
48=back
49 2393
50=head1 AUTHOR 2394=head1 AUTHOR
51 2395
52 Marc Lehmann <schmorp@schmorp.de> 2396 Marc Lehmann <schmorp@schmorp.de>
53 http://home.schmorp.de/ 2397 http://home.schmorp.de/

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines