… | |
… | |
43 | |
43 | |
44 | OpenCL::Event objects are used to signal when something is complete. |
44 | OpenCL::Event objects are used to signal when something is complete. |
45 | |
45 | |
46 | =head2 HELPFUL RESOURCES |
46 | =head2 HELPFUL RESOURCES |
47 | |
47 | |
48 | The OpenCL spec used to develop this module (1.2 spec was available, but |
48 | The OpenCL specs used to develop this module: |
49 | no implementation was available to me :). |
|
|
50 | |
49 | |
51 | http://www.khronos.org/registry/cl/specs/opencl-1.1.pdf |
50 | http://www.khronos.org/registry/cl/specs/opencl-1.1.pdf |
|
|
51 | http://www.khronos.org/registry/cl/specs/opencl-1.2.pdf |
|
|
52 | http://www.khronos.org/registry/cl/specs/opencl-1.2-extensions.pdf |
52 | |
53 | |
53 | OpenCL manpages: |
54 | OpenCL manpages: |
54 | |
55 | |
55 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/ |
56 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/ |
|
|
57 | http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/ |
56 | |
58 | |
57 | If you are into UML class diagrams, the following diagram might help - if |
59 | If you are into UML class diagrams, the following diagram might help - if |
58 | not, it will be mildly cobfusing: |
60 | not, it will be mildly confusing (also, the class hierarchy of this module |
|
|
61 | is much more fine-grained): |
59 | |
62 | |
60 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/classDiagram.html |
63 | http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/classDiagram.html |
61 | |
64 | |
62 | Here's a tutorial from AMD (very AMD-centric, too), not sure how useful it |
65 | Here's a tutorial from AMD (very AMD-centric, too), not sure how useful it |
63 | is, but at least it's free of charge: |
66 | is, but at least it's free of charge: |
64 | |
67 | |
65 | http://developer.amd.com/zones/OpenCLZone/courses/Documents/Introduction_to_OpenCL_Programming%20Training_Guide%20%28201005%29.pdf |
68 | http://developer.amd.com/zones/OpenCLZone/courses/Documents/Introduction_to_OpenCL_Programming%20Training_Guide%20%28201005%29.pdf |
… | |
… | |
138 | =head2 Create a buffer with some predefined data, read it back synchronously, |
141 | =head2 Create a buffer with some predefined data, read it back synchronously, |
139 | then asynchronously. |
142 | then asynchronously. |
140 | |
143 | |
141 | my $buf = $ctx->buffer_sv (OpenCL::MEM_COPY_HOST_PTR, "helmut"); |
144 | my $buf = $ctx->buffer_sv (OpenCL::MEM_COPY_HOST_PTR, "helmut"); |
142 | |
145 | |
143 | $queue->enqueue_read_buffer ($buf, 1, 1, 3, my $data); |
146 | $queue->read_buffer ($buf, 1, 1, 3, my $data); |
144 | print "$data\n"; |
147 | print "$data\n"; |
145 | |
148 | |
146 | my $ev = $queue->enqueue_read_buffer ($buf, 0, 1, 3, my $data); |
149 | my $ev = $queue->read_buffer ($buf, 0, 1, 3, my $data); |
147 | $ev->wait; |
150 | $ev->wait; |
148 | print "$data\n"; # prints "elm" |
151 | print "$data\n"; # prints "elm" |
149 | |
152 | |
150 | =head2 Create and build a program, then create a kernel out of one of its |
153 | =head2 Create and build a program, then create a kernel out of one of its |
151 | functions. |
154 | functions. |
… | |
… | |
157 | $id = get_global_id (0); |
160 | $id = get_global_id (0); |
158 | output [id] = input [id] * input [id]; |
161 | output [id] = input [id] * input [id]; |
159 | } |
162 | } |
160 | '; |
163 | '; |
161 | |
164 | |
162 | my $prog = $ctx->program_with_source ($src); |
165 | my $prog = $ctx->build_program ($src); |
163 | |
|
|
164 | # build croaks on compile errors, so catch it and print the compile errors |
|
|
165 | eval { $prog->build ($dev); 1 } |
|
|
166 | or die $prog->build_log; |
|
|
167 | |
|
|
168 | my $kernel = $prog->kernel ("squareit"); |
166 | my $kernel = $prog->kernel ("squareit"); |
169 | |
167 | |
170 | =head2 Create some input and output float buffers, then call the |
168 | =head2 Create some input and output float buffers, then call the |
171 | 'squareit' kernel on them. |
169 | 'squareit' kernel on them. |
172 | |
170 | |
… | |
… | |
176 | # set buffer |
174 | # set buffer |
177 | $kernel->set_buffer (0, $input); |
175 | $kernel->set_buffer (0, $input); |
178 | $kernel->set_buffer (1, $output); |
176 | $kernel->set_buffer (1, $output); |
179 | |
177 | |
180 | # execute it for all 4 numbers |
178 | # execute it for all 4 numbers |
181 | $queue->enqueue_nd_range_kernel ($kernel, undef, [4], undef); |
179 | $queue->nd_range_kernel ($kernel, undef, [4], undef); |
182 | |
180 | |
183 | # enqueue a synchronous read |
181 | # enqueue a synchronous read |
184 | $queue->enqueue_read_buffer ($output, 1, 0, OpenCL::SIZEOF_FLOAT * 4, my $data); |
182 | $queue->read_buffer ($output, 1, 0, OpenCL::SIZEOF_FLOAT * 4, my $data); |
185 | |
183 | |
186 | # print the results: |
184 | # print the results: |
187 | printf "%s\n", join ", ", unpack "f*", $data; |
185 | printf "%s\n", join ", ", unpack "f*", $data; |
188 | |
186 | |
189 | =head2 The same enqueue operations as before, but assuming an out-of-order queue, |
187 | =head2 The same enqueue operations as before, but assuming an out-of-order queue, |
190 | showing off barriers. |
188 | showing off barriers. |
191 | |
189 | |
192 | # execute it for all 4 numbers |
190 | # execute it for all 4 numbers |
193 | $queue->enqueue_nd_range_kernel ($kernel, undef, [4], undef); |
191 | $queue->nd_range_kernel ($kernel, undef, [4], undef); |
194 | |
192 | |
195 | # enqueue a barrier to ensure in-order execution |
193 | # enqueue a barrier to ensure in-order execution |
196 | $queue->enqueue_barrier; |
194 | $queue->barrier; |
197 | |
195 | |
198 | # enqueue an async read |
196 | # enqueue an async read |
199 | $queue->enqueue_read_buffer ($output, 0, 0, OpenCL::SIZEOF_FLOAT * 4, my $data); |
197 | $queue->read_buffer ($output, 0, 0, OpenCL::SIZEOF_FLOAT * 4, my $data); |
200 | |
198 | |
201 | # wait for all requests to finish |
199 | # wait for all requests to finish |
202 | $queue->finish; |
200 | $queue->finish; |
203 | |
201 | |
204 | =head2 The same enqueue operations as before, but assuming an out-of-order queue, |
202 | =head2 The same enqueue operations as before, but assuming an out-of-order queue, |
205 | showing off event objects and wait lists. |
203 | showing off event objects and wait lists. |
206 | |
204 | |
207 | # execute it for all 4 numbers |
205 | # execute it for all 4 numbers |
208 | my $ev = $queue->enqueue_nd_range_kernel ($kernel, undef, [4], undef); |
206 | my $ev = $queue->nd_range_kernel ($kernel, undef, [4], undef); |
209 | |
207 | |
210 | # enqueue an async read |
208 | # enqueue an async read |
211 | $ev = $queue->enqueue_read_buffer ($output, 0, 0, OpenCL::SIZEOF_FLOAT * 4, my $data, $ev); |
209 | $ev = $queue->read_buffer ($output, 0, 0, OpenCL::SIZEOF_FLOAT * 4, my $data, $ev); |
212 | |
210 | |
213 | # wait for the last event to complete |
211 | # wait for the last event to complete |
214 | $ev->wait; |
212 | $ev->wait; |
|
|
213 | |
|
|
214 | =head2 Use the OpenGL module to share a texture between OpenCL and OpenGL and draw some julia |
|
|
215 | set tunnel effect. |
|
|
216 | |
|
|
217 | This is quite a long example to get you going - you can download it from |
|
|
218 | L<http://cvs.schmorp.de/OpenCL/examples/juliaflight>. |
|
|
219 | |
|
|
220 | use OpenGL ":all"; |
|
|
221 | use OpenCL; |
|
|
222 | |
|
|
223 | my $S = $ARGV[0] || 256; # window/texture size, smaller is faster |
|
|
224 | |
|
|
225 | # open a window and create a gl texture |
|
|
226 | OpenGL::glpOpenWindow width => $S, height => $S; |
|
|
227 | my $texid = glGenTextures_p 1; |
|
|
228 | glBindTexture GL_TEXTURE_2D, $texid; |
|
|
229 | glTexImage2D_c GL_TEXTURE_2D, 0, GL_RGBA8, $S, $S, 0, GL_RGBA, GL_UNSIGNED_BYTE, 0; |
|
|
230 | |
|
|
231 | # find and use the first opencl device that let's us get a shared opengl context |
|
|
232 | my $platform; |
|
|
233 | my $dev; |
|
|
234 | my $ctx; |
|
|
235 | |
|
|
236 | for (OpenCL::platforms) { |
|
|
237 | $platform = $_; |
|
|
238 | for ($platform->devices) { |
|
|
239 | $dev = $_; |
|
|
240 | $ctx = $platform->context ([OpenCL::GLX_DISPLAY_KHR, undef, OpenCL::GL_CONTEXT_KHR, undef], [$dev]) |
|
|
241 | and last; |
|
|
242 | } |
|
|
243 | } |
|
|
244 | |
|
|
245 | $ctx |
|
|
246 | or die "cannot find suitable OpenCL device\n"; |
|
|
247 | |
|
|
248 | my $queue = $ctx->queue ($dev); |
|
|
249 | |
|
|
250 | # now attach an opencl image2d object to the opengl texture |
|
|
251 | my $tex = $ctx->gl_texture2d (OpenCL::MEM_WRITE_ONLY, GL_TEXTURE_2D, 0, $texid); |
|
|
252 | |
|
|
253 | # now the boring opencl code |
|
|
254 | my $src = <<EOF; |
|
|
255 | kernel void |
|
|
256 | juliatunnel (write_only image2d_t img, float time) |
|
|
257 | { |
|
|
258 | int2 xy = (int2)(get_global_id (0), get_global_id (1)); |
|
|
259 | float2 p = convert_float2 (xy) / $S.f * 2.f - 1.f; |
|
|
260 | |
|
|
261 | float2 m = (float2)(1.f, p.y) / fabs (p.x); // tunnel |
|
|
262 | m.x = fabs (fmod (m.x + time * 0.05f, 4.f) - 2.f); |
|
|
263 | |
|
|
264 | float2 z = m; |
|
|
265 | float2 c = (float2)(sin (time * 0.01133f), cos (time * 0.02521f)); |
|
|
266 | |
|
|
267 | for (int i = 0; i < 25 && dot (z, z) < 4.f; ++i) // standard julia |
|
|
268 | z = (float2)(z.x * z.x - z.y * z.y, 2.f * z.x * z.y) + c; |
|
|
269 | |
|
|
270 | float3 colour = (float3)(z.x, z.y, atan2 (z.y, z.x)); |
|
|
271 | write_imagef (img, xy, (float4)(colour * p.x * p.x, 1.)); |
|
|
272 | } |
|
|
273 | EOF |
|
|
274 | |
|
|
275 | my $prog = $ctx->build_program ($src); |
|
|
276 | my $kernel = $prog->kernel ("juliatunnel"); |
|
|
277 | |
|
|
278 | # program compiled, kernel ready, now draw and loop |
|
|
279 | |
|
|
280 | for (my $time; ; ++$time) { |
|
|
281 | # acquire objects from opengl |
|
|
282 | $queue->acquire_gl_objects ([$tex]); |
|
|
283 | |
|
|
284 | # configure and run our kernel |
|
|
285 | $kernel->setf ("mf", $tex, $time*2); # mf = memory object, float |
|
|
286 | $queue->nd_range_kernel ($kernel, undef, [$S, $S], undef); |
|
|
287 | |
|
|
288 | # release objects to opengl again |
|
|
289 | $queue->release_gl_objects ([$tex]); |
|
|
290 | |
|
|
291 | # wait |
|
|
292 | $queue->finish; |
|
|
293 | |
|
|
294 | # now draw the texture, the defaults should be all right |
|
|
295 | glTexParameterf GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST; |
|
|
296 | |
|
|
297 | glEnable GL_TEXTURE_2D; |
|
|
298 | glBegin GL_QUADS; |
|
|
299 | glTexCoord2f 0, 1; glVertex3i -1, -1, -1; |
|
|
300 | glTexCoord2f 0, 0; glVertex3i 1, -1, -1; |
|
|
301 | glTexCoord2f 1, 0; glVertex3i 1, 1, -1; |
|
|
302 | glTexCoord2f 1, 1; glVertex3i -1, 1, -1; |
|
|
303 | glEnd; |
|
|
304 | |
|
|
305 | glXSwapBuffers; |
|
|
306 | |
|
|
307 | select undef, undef, undef, 1/60; |
|
|
308 | } |
|
|
309 | |
|
|
310 | =head2 How to modify the previous example to not rely on GL sharing. |
|
|
311 | |
|
|
312 | For those poor souls with only a sucky CPU OpenCL implementation, you |
|
|
313 | currently have to read the image into some perl scalar, and then modify a |
|
|
314 | texture or use glDrawPixels or so). |
|
|
315 | |
|
|
316 | First, when you don't need gl sharing, you can create the context much simpler: |
|
|
317 | |
|
|
318 | $ctx = $platform->context (undef, [$dev]) |
|
|
319 | |
|
|
320 | To use a texture, you would modify the above example by creating an |
|
|
321 | OpenCL::Image manually instead of deriving it from a texture: |
|
|
322 | |
|
|
323 | my $tex = $ctx->image2d (OpenCL::MEM_WRITE_ONLY, OpenCL::RGBA, OpenCL::UNORM_INT8, $S, $S); |
|
|
324 | |
|
|
325 | And in the darw loop, intead of acquire_gl_objects/release_gl_objects, you |
|
|
326 | would read the image2d after the kernel has written it: |
|
|
327 | |
|
|
328 | $queue->read_image ($tex, 0, 0, 0, 0, $S, $S, 1, 0, 0, my $data); |
|
|
329 | |
|
|
330 | And then you would upload the pixel data to the texture (or use glDrawPixels): |
|
|
331 | |
|
|
332 | glTexSubImage2D_s GL_TEXTURE_2D, 0, 0, 0, $S, $S, GL_RGBA, GL_UNSIGNED_BYTE, $data; |
|
|
333 | |
|
|
334 | The fully modified example can be found at |
|
|
335 | L<http://cvs.schmorp.de/OpenCL/examples/juliaflight-nosharing>. |
215 | |
336 | |
216 | =head1 DOCUMENTATION |
337 | =head1 DOCUMENTATION |
217 | |
338 | |
218 | =head2 BASIC CONVENTIONS |
339 | =head2 BASIC CONVENTIONS |
219 | |
340 | |
… | |
… | |
241 | =item * Structures are often specified by flattening out their components |
362 | =item * Structures are often specified by flattening out their components |
242 | as with short vectors, and returned as arrayrefs. |
363 | as with short vectors, and returned as arrayrefs. |
243 | |
364 | |
244 | =item * When enqueuing commands, the wait list is specified by adding |
365 | =item * When enqueuing commands, the wait list is specified by adding |
245 | extra arguments to the function - anywhere a C<$wait_events...> argument |
366 | extra arguments to the function - anywhere a C<$wait_events...> argument |
246 | is documented this can be any number of event objects. |
367 | is documented this can be any number of event objects. As an extsnion |
|
|
368 | implemented by this module, C<undef> values will be ignored in the event |
|
|
369 | list. |
247 | |
370 | |
248 | =item * When enqueuing commands, if the enqueue method is called in void |
371 | =item * When enqueuing commands, if the enqueue method is called in void |
249 | context, no event is created. In all other contexts an event is returned |
372 | context, no event is created. In all other contexts an event is returned |
250 | by the method. |
373 | by the method. |
251 | |
374 | |
… | |
… | |
271 | ulong IV - Q |
394 | ulong IV - Q |
272 | float NV float f |
395 | float NV float f |
273 | half IV ushort S |
396 | half IV ushort S |
274 | double NV double d |
397 | double NV double d |
275 | |
398 | |
|
|
399 | =head2 GLX SUPPORT |
|
|
400 | |
|
|
401 | Due to the sad state that OpenGL support is in in Perl (mostly the OpenGL |
|
|
402 | module, which has little to no documentation and has little to no support |
|
|
403 | for glX), this module, as a special extension, treats context creation |
|
|
404 | properties C<OpenCL::GLX_DISPLAY_KHR> and C<OpenCL::GL_CONTEXT_KHR> |
|
|
405 | specially: If either or both of these are C<undef>, then the OpenCL |
|
|
406 | module tries to dynamically resolve C<glXGetCurrentDisplay> and |
|
|
407 | C<glXGetCurrentContext>, call these functions and use their return values |
|
|
408 | instead. |
|
|
409 | |
|
|
410 | For this to work, the OpenGL library must be loaded, a GLX context must |
|
|
411 | have been created and be made current, and C<dlsym> must be available and |
|
|
412 | capable of finding the function via C<RTLD_DEFAULT>. |
|
|
413 | |
|
|
414 | =head2 EVENT SYSTEM |
|
|
415 | |
|
|
416 | OpenCL can generate a number of (potentially) asynchronous events, for |
|
|
417 | example, after compiling a program, to signal a context-related error or, |
|
|
418 | perhaps most important, to signal completion of queued jobs (by setting |
|
|
419 | callbacks on OpenCL::Event objects). |
|
|
420 | |
|
|
421 | To facilitate this, this module maintains an event queue - each |
|
|
422 | time an asynchronous event happens, it is queued, and perl will be |
|
|
423 | interrupted. This is implemented via the L<Async::Interrupt> module. In |
|
|
424 | addition, this module has L<AnyEvent> support, so it can seamlessly |
|
|
425 | integrate itself into many event loops. |
|
|
426 | |
|
|
427 | Since this module is a bit hard to understand, here are some case examples: |
|
|
428 | |
|
|
429 | =head3 Don't use callbacks. |
|
|
430 | |
|
|
431 | When your program never uses any callbacks, then there will never be any |
|
|
432 | notifications you need to take care of, and therefore no need to worry |
|
|
433 | about all this. |
|
|
434 | |
|
|
435 | You can achieve a great deal by explicitly waiting for events, or using |
|
|
436 | barriers and flush calls. In many programs, there is no need at all to |
|
|
437 | tinker with asynchronous events. |
|
|
438 | |
|
|
439 | =head3 Use AnyEvent |
|
|
440 | |
|
|
441 | This module automatically registers a watcher that invokes all outstanding |
|
|
442 | event callbacks when AnyEvent is initialised (and block asynchronous |
|
|
443 | interruptions). Using this mode of operations is the safest and most |
|
|
444 | recommended one. |
|
|
445 | |
|
|
446 | To use this, simply use AnyEvent and this module normally, make sure you |
|
|
447 | have an event loop running: |
|
|
448 | |
|
|
449 | use Gtk2 -init; |
|
|
450 | use AnyEvent; |
|
|
451 | |
|
|
452 | # initialise AnyEvent, by creating a watcher, or: |
|
|
453 | AnyEvent::detect; |
|
|
454 | |
|
|
455 | my $e = $queue->marker; |
|
|
456 | $e->cb (sub { |
|
|
457 | warn "opencl is finished\n"; |
|
|
458 | }) |
|
|
459 | |
|
|
460 | main Gtk2; |
|
|
461 | |
|
|
462 | Note that this module will not initialise AnyEvent for you. Before |
|
|
463 | AnyEvent is initialised, the module will asynchronously interrupt perl |
|
|
464 | instead. To avoid any surprises, it's best to explicitly initialise |
|
|
465 | AnyEvent. |
|
|
466 | |
|
|
467 | You can temporarily enable asynchronous interruptions (see next paragraph) |
|
|
468 | by calling C<$OpenCL::INTERRUPT->unblock> and disable them again by |
|
|
469 | calling C<$OpenCL::INTERRUPT->block>. |
|
|
470 | |
|
|
471 | =head3 Let yourself be interrupted at any time |
|
|
472 | |
|
|
473 | This mode is the default unless AnyEvent is loaded and initialised. In |
|
|
474 | this mode, OpenCL asynchronously interrupts a running perl program. The |
|
|
475 | emphasis is on both I<asynchronously> and I<running> here. |
|
|
476 | |
|
|
477 | Asynchronously means that perl might execute your callbacks at any |
|
|
478 | time. For example, in the following code (I<THAT YOU SHOULD NOT COPY>), |
|
|
479 | the C<until> loop following the marker call will be interrupted by the |
|
|
480 | callback: |
|
|
481 | |
|
|
482 | my $e = $queue->marker; |
|
|
483 | my $flag; |
|
|
484 | $e->cb (sub { $flag = 1 }); |
|
|
485 | 1 until $flag; |
|
|
486 | # $flag is now 1 |
|
|
487 | |
|
|
488 | The reason why you shouldn't blindly copy the above code is that |
|
|
489 | busy waiting is a really really bad thing, and really really bad for |
|
|
490 | performance. |
|
|
491 | |
|
|
492 | While at first this asynchronous business might look exciting, it can be |
|
|
493 | really hard, because you need to be prepared for the callback code to be |
|
|
494 | executed at any time, which limits the amount of things the callback code |
|
|
495 | can do safely. |
|
|
496 | |
|
|
497 | This can be mitigated somewhat by using C<< |
|
|
498 | $OpenCL::INTERRUPT->scope_block >> (see the L<Async::Interrupt> |
|
|
499 | documentation for details). |
|
|
500 | |
|
|
501 | The other problem is that your program must be actively I<running> to be |
|
|
502 | interrupted. When you calculate stuff, your program is running. When you |
|
|
503 | hang in some C functions or other block execution (by calling C<sleep>, |
|
|
504 | C<select>, running an event loop and so on), your program is waiting, not |
|
|
505 | running. |
|
|
506 | |
|
|
507 | One way around that would be to attach a read watcher to your event loop, |
|
|
508 | listening for events on C<< $OpenCL::INTERRUPT->pipe_fileno >>, using a |
|
|
509 | dummy callback (C<sub { }>) to temporarily execute some perl code. |
|
|
510 | |
|
|
511 | That is then awfully close to using the built-in AnyEvent support above, |
|
|
512 | though, so consider that one instead. |
|
|
513 | |
|
|
514 | =head3 Be creative |
|
|
515 | |
|
|
516 | OpenCL exports the L<Async::Interrupt> object it uses in the global |
|
|
517 | variable C<$OpenCL::INTERRUPT>. You can configure it in any way you like. |
|
|
518 | |
|
|
519 | So if you want to feel like a real pro, err, wait, if you feel no risk |
|
|
520 | menas no fun, you can experiment by implementing your own mode of |
|
|
521 | operations. |
|
|
522 | |
|
|
523 | =cut |
|
|
524 | |
|
|
525 | package OpenCL; |
|
|
526 | |
|
|
527 | use common::sense; |
|
|
528 | use Carp (); |
|
|
529 | use Async::Interrupt (); |
|
|
530 | |
|
|
531 | our $POLL_FUNC; # set by XS |
|
|
532 | |
|
|
533 | BEGIN { |
|
|
534 | our $VERSION = '0.98'; |
|
|
535 | |
|
|
536 | require XSLoader; |
|
|
537 | XSLoader::load (__PACKAGE__, $VERSION); |
|
|
538 | |
|
|
539 | @OpenCL::Platform::ISA = |
|
|
540 | @OpenCL::Device::ISA = |
|
|
541 | @OpenCL::Context::ISA = |
|
|
542 | @OpenCL::Queue::ISA = |
|
|
543 | @OpenCL::Memory::ISA = |
|
|
544 | @OpenCL::Sampler::ISA = |
|
|
545 | @OpenCL::Program::ISA = |
|
|
546 | @OpenCL::Kernel::ISA = |
|
|
547 | @OpenCL::Event::ISA = OpenCL::Object::; |
|
|
548 | |
|
|
549 | @OpenCL::Buffer::ISA = |
|
|
550 | @OpenCL::Image::ISA = OpenCL::Memory::; |
|
|
551 | |
|
|
552 | @OpenCL::BufferObj::ISA = OpenCL::Buffer::; |
|
|
553 | |
|
|
554 | @OpenCL::Image2D::ISA = |
|
|
555 | @OpenCL::Image3D::ISA = |
|
|
556 | @OpenCL::Image2DArray::ISA = |
|
|
557 | @OpenCL::Image1D::ISA = |
|
|
558 | @OpenCL::Image1DArray::ISA = |
|
|
559 | @OpenCL::Image1DBuffer::ISA = OpenCL::Image::; |
|
|
560 | |
|
|
561 | @OpenCL::UserEvent::ISA = OpenCL::Event::; |
|
|
562 | |
|
|
563 | @OpenCL::MappedBuffer::ISA = |
|
|
564 | @OpenCL::MappedImage::ISA = OpenCL::Mapped::; |
|
|
565 | } |
|
|
566 | |
276 | =head2 THE OpenCL PACKAGE |
567 | =head2 THE OpenCL PACKAGE |
277 | |
568 | |
278 | =over 4 |
569 | =over 4 |
279 | |
570 | |
280 | =item $int = OpenCL::errno |
571 | =item $int = OpenCL::errno |
281 | |
572 | |
282 | The last error returned by a function - it's only valid after an error occured |
573 | The last error returned by a function - it's only valid after an error occured |
283 | and before calling another OpenCL function. |
574 | and before calling another OpenCL function. |
284 | |
575 | |
285 | =item $str = OpenCL::err2str $errval |
576 | =item $str = OpenCL::err2str [$errval] |
286 | |
577 | |
287 | Comverts an error value into a human readable string. |
578 | Converts an error value into a human readable string. IF no error value is |
|
|
579 | given, then the last error will be used (as returned by OpenCL::errno). |
288 | |
580 | |
289 | =item $str = OpenCL::enum2str $enum |
581 | =item $str = OpenCL::enum2str $enum |
290 | |
582 | |
291 | Converts most enum values (of parameter names, image format constants, |
583 | Converts most enum values (of parameter names, image format constants, |
292 | object types, addressing and filter modes, command types etc.) into a |
584 | object types, addressing and filter modes, command types etc.) into a |
… | |
… | |
298 | |
590 | |
299 | Returns all available OpenCL::Platform objects. |
591 | Returns all available OpenCL::Platform objects. |
300 | |
592 | |
301 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetPlatformIDs.html> |
593 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetPlatformIDs.html> |
302 | |
594 | |
303 | =item $ctx = OpenCL::context_from_type $properties, $type = OpenCL::DEVICE_TYPE_DEFAULT, $notify = undef |
595 | =item $ctx = OpenCL::context_from_type $properties, $type = OpenCL::DEVICE_TYPE_DEFAULT, $callback->($err, $pvt) = $print_stderr |
304 | |
596 | |
305 | Tries to create a context from a default device and platform - never worked for me. |
597 | Tries to create a context from a default device and platform type - never worked for me. |
306 | |
598 | |
307 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContextFromType.html> |
599 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContextFromType.html> |
308 | |
600 | |
|
|
601 | =item $ctx = OpenCL::context $properties, \@devices, $callback->($err, $pvt) = $print_stderr) |
|
|
602 | |
|
|
603 | Create a new OpenCL::Context object using the given device object(s). This |
|
|
604 | function isn't implemented yet, use C<< $platform->context >> instead. |
|
|
605 | |
|
|
606 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContext.html> |
|
|
607 | |
309 | =item OpenCL::wait_for_events $wait_events... |
608 | =item OpenCL::wait_for_events $wait_events... |
310 | |
609 | |
311 | Waits for all events to complete. |
610 | Waits for all events to complete. |
312 | |
611 | |
313 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clWaitForEvents.html> |
612 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clWaitForEvents.html> |
314 | |
613 | |
|
|
614 | =item OpenCL::poll |
|
|
615 | |
|
|
616 | Checks if there are any outstanding events (see L<EVENT SYSTEM>) and |
|
|
617 | invokes their callbacks. |
|
|
618 | |
|
|
619 | =item $OpenCL::INTERRUPT |
|
|
620 | |
|
|
621 | The L<Async::Interrupt> object used to signal asynchronous events (see |
|
|
622 | L<EVENT SYSTEM>). |
|
|
623 | |
|
|
624 | =cut |
|
|
625 | |
|
|
626 | our $INTERRUPT = new Async::Interrupt c_cb => [$POLL_FUNC, 0]; |
|
|
627 | |
|
|
628 | &_eq_initialise ($INTERRUPT->signal_func); |
|
|
629 | |
|
|
630 | =item $OpenCL::WATCHER |
|
|
631 | |
|
|
632 | The L<AnyEvent> watcher object used to watch for asynchronous events (see |
|
|
633 | L<EVENT SYSTEM>). This variable is C<undef> until L<AnyEvent> has been |
|
|
634 | loaded I<and> initialised (e.g. by calling C<AnyEvent::detect>). |
|
|
635 | |
|
|
636 | =cut |
|
|
637 | |
|
|
638 | our $WATCHER; |
|
|
639 | |
|
|
640 | sub _init_anyevent { |
|
|
641 | $INTERRUPT->block; |
|
|
642 | $WATCHER = AE::io ($INTERRUPT->pipe_fileno, 0, sub { $INTERRUPT->handle }); |
|
|
643 | } |
|
|
644 | |
|
|
645 | if (defined $AnyEvent::MODEL) { |
|
|
646 | _init_anyevent; |
|
|
647 | } else { |
|
|
648 | push @AnyEvent::post_detect, \&_init_anyevent; |
|
|
649 | } |
|
|
650 | |
315 | =back |
651 | =back |
316 | |
652 | |
|
|
653 | =head2 THE OpenCL::Object CLASS |
|
|
654 | |
|
|
655 | This is the base class for all objects in the OpenCL module. The only |
|
|
656 | method it implements is the C<id> method, which is only useful if you want |
|
|
657 | to interface to OpenCL on the C level. |
|
|
658 | |
|
|
659 | =over 4 |
|
|
660 | |
|
|
661 | =item $iv = $obj->id |
|
|
662 | |
|
|
663 | OpenCL objects are represented by pointers or integers on the C level. If |
|
|
664 | you want to interface to an OpenCL object directly on the C level, then |
|
|
665 | you need this value, which is returned by this method. You should use an |
|
|
666 | C<IV> type in your code and cast that to the correct type. |
|
|
667 | |
|
|
668 | =cut |
|
|
669 | |
|
|
670 | sub OpenCL::Object::id { |
|
|
671 | ref $_[0] eq "SCALAR" |
|
|
672 | ? ${ $_[0] } |
|
|
673 | : $_[0][0] |
|
|
674 | } |
|
|
675 | |
|
|
676 | =back |
|
|
677 | |
317 | =head2 THE OpenCL::Platform CLASS |
678 | =head2 THE OpenCL::Platform CLASS |
318 | |
679 | |
319 | =over 4 |
680 | =over 4 |
320 | |
681 | |
321 | =item @devices = $platform->devices ($type = OpenCL::DEVICE_TYPE_ALL) |
682 | =item @devices = $platform->devices ($type = OpenCL::DEVICE_TYPE_ALL) |
322 | |
683 | |
323 | Returns a list of matching OpenCL::Device objects. |
684 | Returns a list of matching OpenCL::Device objects. |
324 | |
685 | |
325 | =item $ctx = $platform->context_from_type ($properties, $type = OpenCL::DEVICE_TYPE_DEFAULT, $notify = undef) |
686 | =item $ctx = $platform->context_from_type ($properties, $type = OpenCL::DEVICE_TYPE_DEFAULT, $callback->($err, $pvt) = $print_stderr) |
326 | |
687 | |
327 | Tries to create a context. Never worked for me, and you need devices explicitly anyway. |
688 | Tries to create a context. Never worked for me, and you need devices explicitly anyway. |
328 | |
689 | |
329 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContextFromType.html> |
690 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContextFromType.html> |
330 | |
691 | |
331 | =item $ctx = $platform->context ($properties = undef, @$devices, $notify = undef) |
692 | =item $ctx = $platform->context ($properties, \@devices, $callback->($err, $pvt) = $print_stderr) |
332 | |
693 | |
333 | Create a new OpenCL::Context object using the given device object(s)- a |
694 | Create a new OpenCL::Context object using the given device object(s)- a |
334 | CL_CONTEXT_PLATFORM property is supplied automatically. |
695 | CL_CONTEXT_PLATFORM property is supplied automatically. |
335 | |
696 | |
336 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContext.html> |
697 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateContext.html> |
… | |
… | |
344 | It's best to avoid this method and use one of the following convenience |
705 | It's best to avoid this method and use one of the following convenience |
345 | wrappers. |
706 | wrappers. |
346 | |
707 | |
347 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetPlatformInfo.html> |
708 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetPlatformInfo.html> |
348 | |
709 | |
|
|
710 | =item $platform->unload_compiler |
|
|
711 | |
|
|
712 | Attempts to unload the compiler for this platform, for endless |
|
|
713 | profit. Does nothing on OpenCL 1.1. |
|
|
714 | |
|
|
715 | L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clUnloadPlatformCompiler.html> |
|
|
716 | |
349 | =for gengetinfo begin platform |
717 | =for gengetinfo begin platform |
350 | |
718 | |
351 | =item $string = $platform->profile |
719 | =item $string = $platform->profile |
352 | |
720 | |
353 | Calls C<clGetPlatformInfo> with C<CL_PLATFORM_PROFILE> and returns the result. |
721 | Calls C<clGetPlatformInfo> with C<CL_PLATFORM_PROFILE> and returns the result. |
… | |
… | |
638 | |
1006 | |
639 | =item @device_partition_property_exts = $device->affinity_domains_ext |
1007 | =item @device_partition_property_exts = $device->affinity_domains_ext |
640 | |
1008 | |
641 | Calls C<clGetDeviceInfo> with C<CL_DEVICE_AFFINITY_DOMAINS_EXT> and returns the result. |
1009 | Calls C<clGetDeviceInfo> with C<CL_DEVICE_AFFINITY_DOMAINS_EXT> and returns the result. |
642 | |
1010 | |
643 | =item $uint = $device->reference_count_ext |
1011 | =item $uint = $device->reference_count_ext |
644 | |
1012 | |
645 | Calls C<clGetDeviceInfo> with C<CL_DEVICE_REFERENCE_COUNT_EXT > and returns the result. |
1013 | Calls C<clGetDeviceInfo> with C<CL_DEVICE_REFERENCE_COUNT_EXT> and returns the result. |
646 | |
1014 | |
647 | =item @device_partition_property_exts = $device->partition_style_ext |
1015 | =item @device_partition_property_exts = $device->partition_style_ext |
648 | |
1016 | |
649 | Calls C<clGetDeviceInfo> with C<CL_DEVICE_PARTITION_STYLE_EXT> and returns the result. |
1017 | Calls C<clGetDeviceInfo> with C<CL_DEVICE_PARTITION_STYLE_EXT> and returns the result. |
650 | |
1018 | |
… | |
… | |
654 | |
1022 | |
655 | =head2 THE OpenCL::Context CLASS |
1023 | =head2 THE OpenCL::Context CLASS |
656 | |
1024 | |
657 | =over 4 |
1025 | =over 4 |
658 | |
1026 | |
|
|
1027 | =item $prog = $ctx->build_program ($program, $options = "") |
|
|
1028 | |
|
|
1029 | This convenience function tries to build the program on all devices in |
|
|
1030 | the context. If the build fails, then the function will C<croak> with the |
|
|
1031 | build log. Otherwise ti returns the program object. |
|
|
1032 | |
|
|
1033 | The C<$program> can either be a C<OpenCL::Program> object or a string |
|
|
1034 | containing the program. In the latter case, a program objetc will be |
|
|
1035 | created automatically. |
|
|
1036 | |
|
|
1037 | =cut |
|
|
1038 | |
|
|
1039 | sub OpenCL::Context::build_program { |
|
|
1040 | my ($self, $prog, $options) = @_; |
|
|
1041 | |
|
|
1042 | $prog = $self->program_with_source ($prog) |
|
|
1043 | unless ref $prog; |
|
|
1044 | |
|
|
1045 | eval { $prog->build (undef, $options); 1 } |
|
|
1046 | or errno == BUILD_PROGRAM_FAILURE |
|
|
1047 | or errno == INVALID_BINARY # workaround nvidia bug |
|
|
1048 | or Carp::croak "OpenCL::Context->build_program: " . err2str; |
|
|
1049 | |
|
|
1050 | # we check status for all devices |
|
|
1051 | for my $dev ($self->devices) { |
|
|
1052 | $prog->build_status ($dev) == BUILD_SUCCESS |
|
|
1053 | or Carp::croak "Building OpenCL program for device '" . $dev->name . "' failed:\n" |
|
|
1054 | . $prog->build_log ($dev); |
|
|
1055 | } |
|
|
1056 | |
|
|
1057 | $prog |
|
|
1058 | } |
|
|
1059 | |
659 | =item $queue = $ctx->queue ($device, $properties) |
1060 | =item $queue = $ctx->queue ($device, $properties) |
660 | |
1061 | |
661 | Create a new OpenCL::Queue object from the context and the given device. |
1062 | Create a new OpenCL::Queue object from the context and the given device. |
662 | |
1063 | |
663 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateCommandQueue.html> |
1064 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateCommandQueue.html> |
|
|
1065 | |
|
|
1066 | Example: create an out-of-order queue. |
|
|
1067 | |
|
|
1068 | $queue = $ctx->queue ($device, OpenCL::QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE); |
664 | |
1069 | |
665 | =item $ev = $ctx->user_event |
1070 | =item $ev = $ctx->user_event |
666 | |
1071 | |
667 | Creates a new OpenCL::UserEvent object. |
1072 | Creates a new OpenCL::UserEvent object. |
668 | |
1073 | |
… | |
… | |
678 | =item $buf = $ctx->buffer_sv ($flags, $data) |
1083 | =item $buf = $ctx->buffer_sv ($flags, $data) |
679 | |
1084 | |
680 | Creates a new OpenCL::Buffer (actually OpenCL::BufferObj) object and |
1085 | Creates a new OpenCL::Buffer (actually OpenCL::BufferObj) object and |
681 | initialise it with the given data values. |
1086 | initialise it with the given data values. |
682 | |
1087 | |
|
|
1088 | =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) |
|
|
1089 | |
|
|
1090 | Creates a new OpenCL::Image object and optionally initialises it with |
|
|
1091 | the given data values. |
|
|
1092 | |
|
|
1093 | L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clCreateImage.html> |
|
|
1094 | |
683 | =item $img = $ctx->image2d ($flags, $channel_order, $channel_type, $width, $height, $row_pitch = 0, $data = undef) |
1095 | =item $img = $ctx->image2d ($flags, $channel_order, $channel_type, $width, $height, $row_pitch = 0, $data = undef) |
684 | |
1096 | |
685 | Creates a new OpenCL::Image2D object and optionally initialises it with |
1097 | Creates a new OpenCL::Image2D object and optionally initialises it with |
686 | the given data values. |
1098 | the given data values. |
687 | |
1099 | |
… | |
… | |
699 | Creates a new OpenCL::Buffer (actually OpenCL::BufferObj) object that refers to the given |
1111 | Creates a new OpenCL::Buffer (actually OpenCL::BufferObj) object that refers to the given |
700 | OpenGL buffer object. |
1112 | OpenGL buffer object. |
701 | |
1113 | |
702 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateFromGLBuffer.html |
1114 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateFromGLBuffer.html |
703 | |
1115 | |
|
|
1116 | =item $img = $ctx->gl_texture ($flags, $target, $miplevel, $texture) |
|
|
1117 | |
|
|
1118 | Creates a new OpenCL::Image object that refers to the given OpenGL |
|
|
1119 | texture object or buffer. |
|
|
1120 | |
|
|
1121 | http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clCreateFromGLTexture.html |
|
|
1122 | |
704 | =item $ctx->gl_texture2d ($flags, $target, $miplevel, $texture) |
1123 | =item $img = $ctx->gl_texture2d ($flags, $target, $miplevel, $texture) |
705 | |
1124 | |
706 | Creates a new OpenCL::Image2D object that refers to the given OpenGL |
1125 | Creates a new OpenCL::Image2D object that refers to the given OpenGL |
707 | 2D texture object. |
1126 | 2D texture object. |
708 | |
1127 | |
709 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateFromGLTexture2D.html |
1128 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateFromGLTexture2D.html |
710 | |
1129 | |
711 | =item $ctx->gl_texture3d ($flags, $target, $miplevel, $texture) |
1130 | =item $img = $ctx->gl_texture3d ($flags, $target, $miplevel, $texture) |
712 | |
1131 | |
713 | Creates a new OpenCL::Image3D object that refers to the given OpenGL |
1132 | Creates a new OpenCL::Image3D object that refers to the given OpenGL |
714 | 3D texture object. |
1133 | 3D texture object. |
715 | |
1134 | |
716 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateFromGLTexture3D.html |
1135 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateFromGLTexture3D.html |
… | |
… | |
739 | |
1158 | |
740 | Creates a new OpenCL::Program object from the given source code. |
1159 | Creates a new OpenCL::Program object from the given source code. |
741 | |
1160 | |
742 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateProgramWithSource.html> |
1161 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateProgramWithSource.html> |
743 | |
1162 | |
|
|
1163 | =item ($program, \@status) = $ctx->program_with_binary (\@devices, \@binaries) |
|
|
1164 | |
|
|
1165 | Creates a new OpenCL::Program object from the given binaries. |
|
|
1166 | |
|
|
1167 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateProgramWithBinary.html> |
|
|
1168 | |
|
|
1169 | Example: clone an existing program object that contains a successfully |
|
|
1170 | compiled program, no matter how useless this is. |
|
|
1171 | |
|
|
1172 | my $clone = $ctx->program_with_binary ([$prog->devices], [$prog->binaries]); |
|
|
1173 | |
744 | =item $packed_value = $ctx->info ($name) |
1174 | =item $packed_value = $ctx->info ($name) |
745 | |
1175 | |
746 | See C<< $platform->info >> for details. |
1176 | See C<< $platform->info >> for details. |
747 | |
1177 | |
748 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetContextInfo.html> |
1178 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetContextInfo.html> |
… | |
… | |
770 | =back |
1200 | =back |
771 | |
1201 | |
772 | =head2 THE OpenCL::Queue CLASS |
1202 | =head2 THE OpenCL::Queue CLASS |
773 | |
1203 | |
774 | An OpenCL::Queue represents an execution queue for OpenCL. You execute |
1204 | An OpenCL::Queue represents an execution queue for OpenCL. You execute |
775 | requests by calling their respective C<enqueue_xxx> method and waitinf for |
1205 | requests by calling their respective method and waiting for it to complete |
776 | it to complete in some way. |
1206 | in some way. |
777 | |
1207 | |
778 | All the enqueue methods return an event object that can be used to wait |
1208 | Most methods that enqueue some request return an event object that can |
779 | for completion, unless the method is called in void context, in which case |
1209 | be used to wait for completion (optionally using a callback), unless |
780 | no event object is created. |
1210 | the method is called in void context, in which case no event object is |
|
|
1211 | created. |
781 | |
1212 | |
782 | They also allow you to specify any number of other event objects that this |
1213 | They also allow you to specify any number of other event objects that this |
783 | request has to wait for before it starts executing, by simply passing the |
1214 | request has to wait for before it starts executing, by simply passing the |
784 | event objects as extra parameters to the enqueue methods. |
1215 | event objects as extra parameters to the enqueue methods. To simplify |
|
|
1216 | program design, this module ignores any C<undef> values in the list of |
|
|
1217 | events. This makes it possible to code operations such as this, without |
|
|
1218 | having to put a valid event object into C<$event> first: |
|
|
1219 | |
|
|
1220 | $event = $queue->xxx (..., $event); |
785 | |
1221 | |
786 | Queues execute in-order by default, without any parallelism, so in most |
1222 | Queues execute in-order by default, without any parallelism, so in most |
787 | cases (i.e. you use only one queue) it's not necessary to wait for or |
1223 | cases (i.e. you use only one queue) it's not necessary to wait for or |
788 | create event objects. |
1224 | create event objects, althoguh an our of order queue is often a bit |
|
|
1225 | faster. |
789 | |
1226 | |
790 | =over 4 |
1227 | =over 4 |
791 | |
1228 | |
792 | =item $ev = $queue->enqueue_read_buffer ($buffer, $blocking, $offset, $len, $data, $wait_events...) |
1229 | =item $ev = $queue->read_buffer ($buffer, $blocking, $offset, $len, $data, $wait_events...) |
793 | |
1230 | |
794 | Reads data from buffer into the given string. |
1231 | Reads data from buffer into the given string. |
795 | |
1232 | |
796 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReadBuffer.html> |
1233 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReadBuffer.html> |
797 | |
1234 | |
798 | =item $ev = $queue->enqueue_write_buffer ($buffer, $blocking, $offset, $data, $wait_events...) |
1235 | =item $ev = $queue->write_buffer ($buffer, $blocking, $offset, $data, $wait_events...) |
799 | |
1236 | |
800 | Writes data to buffer from the given string. |
1237 | Writes data to buffer from the given string. |
801 | |
1238 | |
802 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWriteBuffer.html> |
1239 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWriteBuffer.html> |
803 | |
1240 | |
804 | =item $ev = $queue->enqueue_copy_buffer ($src, $dst, $src_offset, $dst_offset, $len, $wait_events...) |
1241 | =item $ev = $queue->copy_buffer ($src, $dst, $src_offset, $dst_offset, $len, $wait_events...) |
805 | |
1242 | |
806 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyBuffer.html> |
1243 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyBuffer.html> |
807 | |
1244 | |
808 | =item $ev = $queue->enqueue_read_buffer_rect (OpenCL::Memory 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...) |
1245 | =item $ev = $queue->read_buffer_rect (OpenCL::Memory 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...) |
809 | |
1246 | |
810 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReadBufferRect.html |
1247 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReadBufferRect.html |
811 | |
1248 | |
812 | =item $ev = $queue->enqueue_write_buffer_rect (OpenCL::Memory 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...) |
1249 | =item $ev = $queue->write_buffer_rect (OpenCL::Memory 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...) |
813 | |
1250 | |
814 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWriteBufferRect.html |
1251 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWriteBufferRect.html |
815 | |
1252 | |
816 | =item $ev = $queue->enqueue_read_image ($src, $blocking, $x, $y, $z, $width, $height, $depth, $row_pitch, $slice_pitch, $data, $wait_events...) |
|
|
817 | |
|
|
818 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyBufferRect.html> |
|
|
819 | |
|
|
820 | =item $ev = $queue->enqueue_copy_buffer_to_image ($src_buffer, $dst_image, $src_offset, $dst_x, $dst_y, $dst_z, $width, $height, $depth, $wait_events...) |
1253 | =item $ev = $queue->copy_buffer_to_image ($src_buffer, $dst_image, $src_offset, $dst_x, $dst_y, $dst_z, $width, $height, $depth, $wait_events...) |
|
|
1254 | |
|
|
1255 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyBufferToImage.html> |
|
|
1256 | |
|
|
1257 | =item $ev = $queue->read_image ($src, $blocking, $x, $y, $z, $width, $height, $depth, $row_pitch, $slice_pitch, $data, $wait_events...) |
|
|
1258 | |
|
|
1259 | C<$row_pitch> (and C<$slice_pitch>) can be C<0>, in which case the OpenCL |
|
|
1260 | module uses the image width (and height) to supply default values. |
821 | |
1261 | |
822 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReadImage.html> |
1262 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReadImage.html> |
823 | |
1263 | |
824 | =item $ev = $queue->enqueue_write_image ($src, $blocking, $x, $y, $z, $width, $height, $depth, $row_pitch, $slice_pitch, $data, $wait_events...) |
1264 | =item $ev = $queue->write_image ($src, $blocking, $x, $y, $z, $width, $height, $depth, $row_pitch, $slice_pitch, $data, $wait_events...) |
825 | |
1265 | |
|
|
1266 | C<$row_pitch> (and C<$slice_pitch>) can be C<0>, in which case the OpenCL |
|
|
1267 | module uses the image width (and height) to supply default values. |
826 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWriteImage.html> |
1268 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWriteImage.html> |
827 | |
1269 | |
828 | =item $ev = $queue->enqueue_copy_image ($src_image, $dst_image, $src_x, $src_y, $src_z, $dst_x, $dst_y, $dst_z, $width, $height, $depth, $wait_events...) |
1270 | =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...) |
829 | |
1271 | |
830 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyImage.html> |
1272 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyImage.html> |
831 | |
1273 | |
832 | =item $ev = $queue->enqueue_copy_image_to_buffer ($src_image, $dst_image, $src_x, $src_y, $src_z, $width, $height, $depth, $dst_offset, $wait_events...) |
1274 | =item $ev = $queue->copy_image_to_buffer ($src_image, $dst_image, $src_x, $src_y, $src_z, $width, $height, $depth, $dst_offset, $wait_events...) |
833 | |
1275 | |
834 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyImageToBuffer.html> |
1276 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyImageToBuffer.html> |
835 | |
1277 | |
836 | =item $ev = $queue->enqueue_copy_buffer_rect ($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...) |
1278 | =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...) |
837 | |
1279 | |
838 | Yeah. |
1280 | Yeah. |
839 | |
1281 | |
840 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyBufferToImage.html>. |
1282 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyBufferToImage.html>. |
841 | |
1283 | |
|
|
1284 | =item $ev = $queue->fill_buffer ($mem, $pattern, $offset, $size, ...) |
|
|
1285 | |
|
|
1286 | Fills the given buffer object with repeated applications of C<$pattern>, |
|
|
1287 | starting at C<$offset> for C<$size> octets. |
|
|
1288 | |
|
|
1289 | L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clEnqueueFillBuffer.html> |
|
|
1290 | |
|
|
1291 | =item $ev = $queue->fill_image ($img, $r, $g, $b, $a, $x, $y, $z, $width, $height, $depth, ...) |
|
|
1292 | |
|
|
1293 | Fills the given image area with the given rgba colour components. The |
|
|
1294 | components are normally floating point values between C<0> and C<1>, |
|
|
1295 | except when the image channel data type is a signe dor unsigned |
|
|
1296 | unnormalised format, in which case the range is determined by the format. |
|
|
1297 | |
|
|
1298 | L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clEnqueueFillImage.html> |
|
|
1299 | |
842 | =item $ev = $queue->enqueue_task ($kernel, $wait_events...) |
1300 | =item $ev = $queue->task ($kernel, $wait_events...) |
843 | |
1301 | |
844 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueTask.html> |
1302 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueTask.html> |
845 | |
1303 | |
846 | =item $ev = $queue->enqueue_nd_range_kernel ($kernel, @$global_work_offset, @$global_work_size, @$local_work_size, $wait_events...) |
1304 | =item $ev = $queue->nd_range_kernel ($kernel, \@global_work_offset, \@global_work_size, \@local_work_size, $wait_events...) |
847 | |
1305 | |
848 | Enqueues a kernel execution. |
1306 | Enqueues a kernel execution. |
849 | |
1307 | |
850 | @$global_work_size must be specified as a reference to an array of |
1308 | \@global_work_size must be specified as a reference to an array of |
851 | integers specifying the work sizes (element counts). |
1309 | integers specifying the work sizes (element counts). |
852 | |
1310 | |
853 | @$global_work_offset must be either C<undef> (in which case all offsets |
1311 | \@global_work_offset must be either C<undef> (in which case all offsets |
854 | are C<0>), or a reference to an array of work offsets, with the same number |
1312 | are C<0>), or a reference to an array of work offsets, with the same number |
855 | of elements as @$global_work_size. |
1313 | of elements as \@global_work_size. |
856 | |
1314 | |
857 | @$local_work_size must be either C<undef> (in which case the |
1315 | \@local_work_size must be either C<undef> (in which case the |
858 | implementation is supposed to choose good local work sizes), or a |
1316 | implementation is supposed to choose good local work sizes), or a |
859 | reference to an array of local work sizes, with the same number of |
1317 | reference to an array of local work sizes, with the same number of |
860 | elements as @$global_work_size. |
1318 | elements as \@global_work_size. |
861 | |
1319 | |
862 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueNDRangeKernel.html> |
1320 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueNDRangeKernel.html> |
863 | |
1321 | |
864 | =item $ev = $queue->enqueue_marker |
1322 | =item $ev = $queue->acquire_gl_objects ([object, ...], $wait_events...) |
865 | |
1323 | |
|
|
1324 | Enqueues a list (an array-ref of OpenCL::Memory objects) to be acquired |
|
|
1325 | for subsequent OpenCL usage. |
|
|
1326 | |
866 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueMarker.html> |
1327 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueAcquireGLObjects.html> |
867 | |
1328 | |
|
|
1329 | =item $ev = $queue->release_gl_objects ([object, ...], $wait_events...) |
|
|
1330 | |
|
|
1331 | Enqueues a list (an array-ref of OpenCL::Memory objects) to be released |
|
|
1332 | for subsequent OpenGL usage. |
|
|
1333 | |
|
|
1334 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueReleaseGLObjects.html> |
|
|
1335 | |
868 | =item $ev = $queue->enqueue_wait_for_events ($wait_events...) |
1336 | =item $ev = $queue->wait_for_events ($wait_events...) |
869 | |
1337 | |
870 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWaitForEvents.html> |
1338 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueWaitForEvents.html> |
871 | |
1339 | |
872 | =item $queue->enqueue_barrier |
1340 | =item $ev = $queue->marker ($wait_events...) |
873 | |
1341 | |
|
|
1342 | L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clEnqueueMarkerWithWaitList.html> |
|
|
1343 | |
|
|
1344 | =item $ev = $queue->barrier ($wait_events...) |
|
|
1345 | |
874 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueBarrier.html> |
1346 | L<http://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clEnqueueBarrierWithWaitList.html> |
875 | |
1347 | |
876 | =item $queue->flush |
1348 | =item $queue->flush |
877 | |
1349 | |
878 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clFlush.html> |
1350 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clFlush.html> |
879 | |
1351 | |
… | |
… | |
904 | =item $command_queue_properties = $command_queue->properties |
1376 | =item $command_queue_properties = $command_queue->properties |
905 | |
1377 | |
906 | Calls C<clGetCommandQueueInfo> with C<CL_QUEUE_PROPERTIES> and returns the result. |
1378 | Calls C<clGetCommandQueueInfo> with C<CL_QUEUE_PROPERTIES> and returns the result. |
907 | |
1379 | |
908 | =for gengetinfo end command_queue |
1380 | =for gengetinfo end command_queue |
|
|
1381 | |
|
|
1382 | =back |
|
|
1383 | |
|
|
1384 | =head3 MEMORY MAPPED BUFFERS |
|
|
1385 | |
|
|
1386 | OpenCL allows you to map buffers and images to host memory (read: perl |
|
|
1387 | scalars). This is done much like reading or copying a buffer, by enqueuing |
|
|
1388 | a map or unmap operation on the command queue. |
|
|
1389 | |
|
|
1390 | The map operations return an C<OpenCL::Mapped> object - see L<THE |
|
|
1391 | OpenCL::Mapped CLASS> section for details on what to do with these |
|
|
1392 | objects. |
|
|
1393 | |
|
|
1394 | The object will be unmapped automatically when the mapped object is |
|
|
1395 | destroyed (you can use a barrier to make sure the unmap has finished, |
|
|
1396 | before using the buffer in a kernel), but you can also enqueue an unmap |
|
|
1397 | operation manually. |
|
|
1398 | |
|
|
1399 | =over 4 |
|
|
1400 | |
|
|
1401 | =item $mapped_buffer = $queue->map_buffer ($buf, $blocking=1, $map_flags=OpenCL::MAP_READ|OpenCL::MAP_WRITE, $offset=0, $size=undef, $wait_events...) |
|
|
1402 | |
|
|
1403 | Maps the given buffer into host memory and returns an |
|
|
1404 | C<OpenCL::MappedBuffer> object. If C<$size> is specified as undef, then |
|
|
1405 | the map will extend to the end of the buffer. |
|
|
1406 | |
|
|
1407 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueMapBuffer.html> |
|
|
1408 | |
|
|
1409 | Example: map the buffer $buf fully and replace the first 4 bytes by "abcd", then unmap. |
|
|
1410 | |
|
|
1411 | { |
|
|
1412 | my $mapped = $queue->map_buffer ($buf, 1, OpenCL::MAP_WRITE); |
|
|
1413 | substr $$mapped, 0, 4, "abcd"; |
|
|
1414 | } # asynchronously unmap because $mapped is destroyed |
|
|
1415 | |
|
|
1416 | =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...) |
|
|
1417 | |
|
|
1418 | Maps the given image area into host memory and return an |
|
|
1419 | C<OpenCL::MappedImage> object. |
|
|
1420 | |
|
|
1421 | If any of C<$width>, C<$height> and/or C<$depth> are C<undef> then they |
|
|
1422 | will be replaced by the maximum possible value. |
|
|
1423 | |
|
|
1424 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueMapImage.html> |
|
|
1425 | |
|
|
1426 | Example: map an image (with OpenCL::UNSIGNED_INT8 channel type) and set |
|
|
1427 | the first channel of the leftmost column to 5, then explicitly unmap |
|
|
1428 | it. You are not necessarily meant to do it this way, this example just |
|
|
1429 | shows you the accessors to use :) |
|
|
1430 | |
|
|
1431 | my $mapped = $queue->map_image ($image, 1, OpenCL::MAP_WRITE); |
|
|
1432 | |
|
|
1433 | $mapped->set ($_ * $mapped->row_pitch, pack "C", 5) |
|
|
1434 | for 0..$image->height; |
|
|
1435 | |
|
|
1436 | $mapped->unmap;. |
|
|
1437 | $mapped->wait; # only needed for out of order queues normally |
|
|
1438 | |
|
|
1439 | =item $ev = $queue->unmap ($mapped, $wait_events...) |
|
|
1440 | |
|
|
1441 | Unmaps the data from host memory. You must not call any methods that |
|
|
1442 | modify the data, or modify the data scalar directly, after calling this |
|
|
1443 | method. |
|
|
1444 | |
|
|
1445 | The mapped event object will always be passed as part of the |
|
|
1446 | $wait_events. The mapped event object will be replaced by the new event |
|
|
1447 | object that this request creates. |
909 | |
1448 | |
910 | =back |
1449 | =back |
911 | |
1450 | |
912 | =head2 THE OpenCL::Memory CLASS |
1451 | =head2 THE OpenCL::Memory CLASS |
913 | |
1452 | |
… | |
… | |
959 | =item $int = $mem->offset |
1498 | =item $int = $mem->offset |
960 | |
1499 | |
961 | Calls C<clGetMemObjectInfo> with C<CL_MEM_OFFSET> and returns the result. |
1500 | Calls C<clGetMemObjectInfo> with C<CL_MEM_OFFSET> and returns the result. |
962 | |
1501 | |
963 | =for gengetinfo end mem |
1502 | =for gengetinfo end mem |
|
|
1503 | |
|
|
1504 | =item ($type, $name) = $mem->gl_object_info |
|
|
1505 | |
|
|
1506 | Returns the OpenGL object type (e.g. OpenCL::GL_OBJECT_TEXTURE2D) and the |
|
|
1507 | object "name" (e.g. the texture name) used to create this memory object. |
|
|
1508 | |
|
|
1509 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetGLObjectInfo.html> |
964 | |
1510 | |
965 | =back |
1511 | =back |
966 | |
1512 | |
967 | =head2 THE OpenCL::Buffer CLASS |
1513 | =head2 THE OpenCL::Buffer CLASS |
968 | |
1514 | |
… | |
… | |
987 | |
1533 | |
988 | =back |
1534 | =back |
989 | |
1535 | |
990 | =head2 THE OpenCL::Image CLASS |
1536 | =head2 THE OpenCL::Image CLASS |
991 | |
1537 | |
992 | This is the superclass of all image objects - OpenCL::Image2D and OpenCL::Image3D. |
1538 | This is the superclass of all image objects - OpenCL::Image1D, |
|
|
1539 | OpenCL::Image1DArray, OpenCL::Image1DBuffer, OpenCL::Image2D, |
|
|
1540 | OpenCL::Image2DArray and OpenCL::Image3D. |
993 | |
1541 | |
994 | =over 4 |
1542 | =over 4 |
995 | |
1543 | |
996 | =item $packed_value = $ev->image_info ($name) |
1544 | =item $packed_value = $image->image_info ($name) |
997 | |
1545 | |
998 | See C<< $platform->info >> for details. |
1546 | See C<< $platform->info >> for details. |
999 | |
1547 | |
1000 | The reason this method is not called C<info> is that there already is an |
1548 | The reason this method is not called C<info> is that there already is an |
1001 | C<< ->info >> method inherited from C<OpenCL::Memory>. |
1549 | C<< ->info >> method inherited from C<OpenCL::Memory>. |
1002 | |
1550 | |
1003 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetImageInfo.html> |
1551 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetImageInfo.html> |
1004 | |
1552 | |
|
|
1553 | =item ($channel_order, $channel_data_type) = $image->format |
|
|
1554 | |
|
|
1555 | Returns the channel order and type used to create the image by calling |
|
|
1556 | C<clGetImageInfo> with C<CL_IMAGE_FORMAT>. |
|
|
1557 | |
1005 | =for gengetinfo begin image |
1558 | =for gengetinfo begin image |
1006 | |
1559 | |
1007 | =item $int = $image->element_size |
1560 | =item $int = $image->element_size |
1008 | |
1561 | |
1009 | Calls C<clGetImageInfo> with C<CL_IMAGE_ELEMENT_SIZE> and returns the result. |
1562 | Calls C<clGetImageInfo> with C<CL_IMAGE_ELEMENT_SIZE> and returns the result. |
… | |
… | |
1028 | |
1581 | |
1029 | Calls C<clGetImageInfo> with C<CL_IMAGE_DEPTH> and returns the result. |
1582 | Calls C<clGetImageInfo> with C<CL_IMAGE_DEPTH> and returns the result. |
1030 | |
1583 | |
1031 | =for gengetinfo end image |
1584 | =for gengetinfo end image |
1032 | |
1585 | |
|
|
1586 | =for gengetinfo begin gl_texture |
|
|
1587 | |
|
|
1588 | =item $GLenum = $gl_texture->target |
|
|
1589 | |
|
|
1590 | Calls C<clGetGLTextureInfo> with C<CL_GL_TEXTURE_TARGET> and returns the result. |
|
|
1591 | |
|
|
1592 | =item $GLint = $gl_texture->gl_mipmap_level |
|
|
1593 | |
|
|
1594 | Calls C<clGetGLTextureInfo> with C<CL_GL_MIPMAP_LEVEL> and returns the result. |
|
|
1595 | |
|
|
1596 | =for gengetinfo end gl_texture |
|
|
1597 | |
1033 | =back |
1598 | =back |
1034 | |
1599 | |
1035 | =head2 THE OpenCL::Sampler CLASS |
1600 | =head2 THE OpenCL::Sampler CLASS |
1036 | |
1601 | |
1037 | =over 4 |
1602 | =over 4 |
… | |
… | |
1070 | |
1635 | |
1071 | =head2 THE OpenCL::Program CLASS |
1636 | =head2 THE OpenCL::Program CLASS |
1072 | |
1637 | |
1073 | =over 4 |
1638 | =over 4 |
1074 | |
1639 | |
1075 | =item $program->build ($device, $options = "") |
1640 | =item $program->build (\@devices = undef, $options = "", $cb->($program) = undef) |
1076 | |
1641 | |
1077 | Tries to build the program with the givne options. |
1642 | Tries to build the program with the given options. See also the |
|
|
1643 | C<$ctx->build> convenience function. |
|
|
1644 | |
|
|
1645 | If a callback is specified, then it will be called when compilation is |
|
|
1646 | finished. Note that many OpenCL implementations block your program while |
|
|
1647 | compiling whether you use a callback or not. See C<build_async> if you |
|
|
1648 | want to make sure the build is done in the background. |
|
|
1649 | |
|
|
1650 | Note that some OpenCL implementations act up badly, and don't call the |
|
|
1651 | callback in some error cases (but call it in others). This implementation |
|
|
1652 | assumes the callback will always be called, and leaks memory if this is |
|
|
1653 | not so. So best make sure you don't pass in invalid values. |
|
|
1654 | |
|
|
1655 | Some implementations fail with C<OpenCL::INVALID_BINARY> when the |
|
|
1656 | compilation state is successful but some later stage fails. |
1078 | |
1657 | |
1079 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clBuildProgram.html> |
1658 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clBuildProgram.html> |
|
|
1659 | |
|
|
1660 | =item $program->build_async (\@devices = undef, $options = "", $cb->($program) = undef) |
|
|
1661 | |
|
|
1662 | Similar to C<< ->build >>, except it starts a thread, and never fails (you |
|
|
1663 | need to check the compilation status form the callback, or by polling). |
1080 | |
1664 | |
1081 | =item $packed_value = $program->build_info ($device, $name) |
1665 | =item $packed_value = $program->build_info ($device, $name) |
1082 | |
1666 | |
1083 | Similar to C<< $platform->info >>, but returns build info for a previous |
1667 | Similar to C<< $platform->info >>, but returns build info for a previous |
1084 | build attempt for the given device. |
1668 | build attempt for the given device. |
… | |
… | |
1089 | |
1673 | |
1090 | Creates an OpenCL::Kernel object out of the named C<__kernel> function in |
1674 | Creates an OpenCL::Kernel object out of the named C<__kernel> function in |
1091 | the program. |
1675 | the program. |
1092 | |
1676 | |
1093 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateKernel.html> |
1677 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateKernel.html> |
|
|
1678 | |
|
|
1679 | =item @kernels = $program->kernels_in_program |
|
|
1680 | |
|
|
1681 | Returns all kernels successfully compiled for all devices in program. |
|
|
1682 | |
|
|
1683 | http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clCreateKernelsInProgram.html |
1094 | |
1684 | |
1095 | =for gengetinfo begin program_build |
1685 | =for gengetinfo begin program_build |
1096 | |
1686 | |
1097 | =item $build_status = $program->build_status ($device) |
1687 | =item $build_status = $program->build_status ($device) |
1098 | |
1688 | |
… | |
… | |
1220 | |
1810 | |
1221 | Calls C<clGetKernelWorkGroupInfo> with C<CL_KERNEL_PRIVATE_MEM_SIZE> and returns the result. |
1811 | Calls C<clGetKernelWorkGroupInfo> with C<CL_KERNEL_PRIVATE_MEM_SIZE> and returns the result. |
1222 | |
1812 | |
1223 | =for gengetinfo end kernel_work_group |
1813 | =for gengetinfo end kernel_work_group |
1224 | |
1814 | |
|
|
1815 | =item $kernel->setf ($format, ...) |
|
|
1816 | |
|
|
1817 | Sets the arguments of a kernel. Since OpenCL 1.1 doesn't have a generic |
|
|
1818 | way to set arguments (and with OpenCL 1.2 it might be rather slow), you |
|
|
1819 | need to specify a format argument, much as with C<printf>, to tell OpenCL |
|
|
1820 | what type of argument it is. |
|
|
1821 | |
|
|
1822 | The format arguments are single letters: |
|
|
1823 | |
|
|
1824 | c char |
|
|
1825 | C unsigned char |
|
|
1826 | s short |
|
|
1827 | S unsigned short |
|
|
1828 | i int |
|
|
1829 | I unsigned int |
|
|
1830 | l long |
|
|
1831 | L unsigned long |
|
|
1832 | |
|
|
1833 | h half float (0..65535) |
|
|
1834 | f float |
|
|
1835 | d double |
|
|
1836 | |
|
|
1837 | z local (octet size) |
|
|
1838 | |
|
|
1839 | m memory object (buffer or image) |
|
|
1840 | a sampler |
|
|
1841 | e event |
|
|
1842 | |
|
|
1843 | Space characters in the format string are ignored. |
|
|
1844 | |
|
|
1845 | Example: set the arguments for a kernel that expects an int, two floats, a buffer and an image. |
|
|
1846 | |
|
|
1847 | $kernel->setf ("i ff mm", 5, 0.5, 3, $buffer, $image); |
|
|
1848 | |
1225 | =item $kernel->set_TYPE ($index, $value) |
1849 | =item $kernel->set_TYPE ($index, $value) |
1226 | |
1850 | |
|
|
1851 | =item $kernel->set_char ($index, $value) |
|
|
1852 | |
|
|
1853 | =item $kernel->set_uchar ($index, $value) |
|
|
1854 | |
|
|
1855 | =item $kernel->set_short ($index, $value) |
|
|
1856 | |
|
|
1857 | =item $kernel->set_ushort ($index, $value) |
|
|
1858 | |
|
|
1859 | =item $kernel->set_int ($index, $value) |
|
|
1860 | |
|
|
1861 | =item $kernel->set_uint ($index, $value) |
|
|
1862 | |
|
|
1863 | =item $kernel->set_long ($index, $value) |
|
|
1864 | |
|
|
1865 | =item $kernel->set_ulong ($index, $value) |
|
|
1866 | |
|
|
1867 | =item $kernel->set_half ($index, $value) |
|
|
1868 | |
|
|
1869 | =item $kernel->set_float ($index, $value) |
|
|
1870 | |
|
|
1871 | =item $kernel->set_double ($index, $value) |
|
|
1872 | |
|
|
1873 | =item $kernel->set_memory ($index, $value) |
|
|
1874 | |
|
|
1875 | =item $kernel->set_buffer ($index, $value) |
|
|
1876 | |
|
|
1877 | =item $kernel->set_image ($index, $value) |
|
|
1878 | |
|
|
1879 | =item $kernel->set_sampler ($index, $value) |
|
|
1880 | |
|
|
1881 | =item $kernel->set_local ($index, $value) |
|
|
1882 | |
|
|
1883 | =item $kernel->set_event ($index, $value) |
|
|
1884 | |
1227 | This is a family of methods to set the kernel argument with the number C<$index> to the give C<$value>. |
1885 | This is a family of methods to set the kernel argument with the number |
1228 | |
1886 | C<$index> to the give C<$value>. |
1229 | TYPE is one of C<char>, C<uchar>, C<short>, C<ushort>, C<int>, C<uint>, |
|
|
1230 | C<long>, C<ulong>, C<half>, C<float>, C<double>, C<memory>, C<buffer>, |
|
|
1231 | C<image2d>, C<image3d>, C<sampler> or C<event>. |
|
|
1232 | |
1887 | |
1233 | Chars and integers (including the half type) are specified as integers, |
1888 | Chars and integers (including the half type) are specified as integers, |
1234 | float and double as floating point values, memory/buffer/image2d/image3d |
1889 | float and double as floating point values, memory/buffer/image must be |
1235 | must be an object of that type or C<undef>, and sampler and event must be |
1890 | an object of that type or C<undef>, local-memory arguments are set by |
1236 | objects of that type. |
1891 | specifying the size, and sampler and event must be objects of that type. |
|
|
1892 | |
|
|
1893 | Note that C<set_memory> works for all memory objects (all types of buffers |
|
|
1894 | and images) - the main purpose of the more specific C<set_TYPE> functions |
|
|
1895 | is type checking. |
|
|
1896 | |
|
|
1897 | Setting an argument for a kernel does NOT keep a reference to the object - |
|
|
1898 | for example, if you set an argument to some image object, free the image, |
|
|
1899 | and call the kernel, you will run into undefined behaviour. |
1237 | |
1900 | |
1238 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clSetKernelArg.html> |
1901 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clSetKernelArg.html> |
1239 | |
1902 | |
1240 | =back |
1903 | =back |
1241 | |
1904 | |
… | |
… | |
1250 | |
1913 | |
1251 | Waits for the event to complete. |
1914 | Waits for the event to complete. |
1252 | |
1915 | |
1253 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clWaitForEvents.html> |
1916 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clWaitForEvents.html> |
1254 | |
1917 | |
|
|
1918 | =item $ev->cb ($exec_callback_type, $callback->($event, $event_command_exec_status)) |
|
|
1919 | |
|
|
1920 | Adds a callback to the callback stack for the given event type. There is |
|
|
1921 | no way to remove a callback again. |
|
|
1922 | |
|
|
1923 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clSetEventCallback.html> |
|
|
1924 | |
1255 | =item $packed_value = $ev->info ($name) |
1925 | =item $packed_value = $ev->info ($name) |
1256 | |
1926 | |
1257 | See C<< $platform->info >> for details. |
1927 | See C<< $platform->info >> for details. |
1258 | |
1928 | |
1259 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetEventInfo.html> |
1929 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clGetEventInfo.html> |
… | |
… | |
1319 | |
1989 | |
1320 | =over 4 |
1990 | =over 4 |
1321 | |
1991 | |
1322 | =item $ev->set_status ($execution_status) |
1992 | =item $ev->set_status ($execution_status) |
1323 | |
1993 | |
|
|
1994 | Sets the execution status of the user event. Can only be called once, |
|
|
1995 | either with OpenCL::COMPLETE or a negative number as status. |
|
|
1996 | |
1324 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clSetUserEventStatus.html> |
1997 | L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clSetUserEventStatus.html> |
1325 | |
1998 | |
1326 | =back |
1999 | =back |
1327 | |
2000 | |
|
|
2001 | =head2 THE OpenCL::Mapped CLASS |
|
|
2002 | |
|
|
2003 | This class represents objects mapped into host memory. They are |
|
|
2004 | represented by a blessed string scalar. The string data is the mapped |
|
|
2005 | memory area, that is, if you read or write it, then the mapped object is |
|
|
2006 | accessed directly. |
|
|
2007 | |
|
|
2008 | You must only ever use operations that modify the string in-place - for |
|
|
2009 | example, a C<substr> that doesn't change the length, or maybe a regex that |
|
|
2010 | doesn't change the length. Any other operation might cause the data to be |
|
|
2011 | copied. |
|
|
2012 | |
|
|
2013 | When the object is destroyed it will enqueue an implicit unmap operation |
|
|
2014 | on the queue that was used to create it. |
|
|
2015 | |
|
|
2016 | Keep in mind that you I<need> to unmap (or destroy) mapped objects before |
|
|
2017 | OpenCL sees the changes, even if some implementations don't need this |
|
|
2018 | sometimes. |
|
|
2019 | |
|
|
2020 | Example, replace the first two floats in the mapped buffer by 1 and 2. |
|
|
2021 | |
|
|
2022 | my $mapped = $queue->map_buffer ($buf, ... |
|
|
2023 | $mapped->event->wait; # make sure it's there |
|
|
2024 | |
|
|
2025 | # now replace first 8 bytes by new data, which is exactly 8 bytes long |
|
|
2026 | # we blindly assume device endianness to equal host endianness |
|
|
2027 | # (and of course, we assume iee 754 single precision floats :) |
|
|
2028 | substr $$mapped, 0, 8, pack "f*", 1, 2; |
|
|
2029 | |
|
|
2030 | =over 4 |
|
|
2031 | |
|
|
2032 | =item $ev = $mapped->unmap ($wait_events...) |
|
|
2033 | |
|
|
2034 | Unmaps the mapped memory object, using the queue originally used to create |
|
|
2035 | it, quite similarly to C<< $queue->unmap ($mapped, ...) >>. |
|
|
2036 | |
|
|
2037 | =item $bool = $mapped->mapped |
|
|
2038 | |
|
|
2039 | Returns whether the object is still mapped - true before an C<unmap> is |
|
|
2040 | enqueued, false afterwards. |
|
|
2041 | |
|
|
2042 | =item $ev = $mapped->event |
|
|
2043 | |
|
|
2044 | Return the event object associated with the mapped object. Initially, this |
|
|
2045 | will be the event object created when mapping the object, and after an |
|
|
2046 | unmap, this will be the event object that the unmap operation created. |
|
|
2047 | |
|
|
2048 | =item $mapped->wait |
|
|
2049 | |
|
|
2050 | Same as C<< $mapped->event->wait >> - makes sure no operations on this |
|
|
2051 | mapped object are outstanding. |
|
|
2052 | |
|
|
2053 | =item $bytes = $mapped->size |
|
|
2054 | |
|
|
2055 | Returns the size of the mapped area, in bytes. Same as C<length $$mapped>. |
|
|
2056 | |
|
|
2057 | =item $ptr = $mapped->ptr |
|
|
2058 | |
|
|
2059 | Returns the raw memory address of the mapped area. |
|
|
2060 | |
|
|
2061 | =item $mapped->set ($offset, $data) |
|
|
2062 | |
|
|
2063 | Replaces the data at the given C<$offset> in the memory area by the new |
|
|
2064 | C<$data>. This method is safer than direct manipulation of C<$mapped> |
|
|
2065 | because it does bounds-checking, but also slower. |
|
|
2066 | |
|
|
2067 | =item $data = $mapped->get ($offset, $length) |
|
|
2068 | |
|
|
2069 | Returns (without copying) a scalar representing the data at the given |
|
|
2070 | C<$offset> and C<$length> in the mapped memory area. This is the same as |
|
|
2071 | the following substr, except much slower; |
|
|
2072 | |
|
|
2073 | $data = substr $$mapped, $offset, $length |
|
|
2074 | |
1328 | =cut |
2075 | =cut |
1329 | |
2076 | |
1330 | package OpenCL; |
2077 | sub OpenCL::Mapped::get { |
1331 | |
2078 | substr ${$_[0]}, $_[1], $_[2] |
1332 | use common::sense; |
|
|
1333 | |
|
|
1334 | BEGIN { |
|
|
1335 | our $VERSION = '0.92'; |
|
|
1336 | |
|
|
1337 | require XSLoader; |
|
|
1338 | XSLoader::load (__PACKAGE__, $VERSION); |
|
|
1339 | |
|
|
1340 | @OpenCL::Buffer::ISA = |
|
|
1341 | @OpenCL::Image::ISA = OpenCL::Memory::; |
|
|
1342 | |
|
|
1343 | @OpenCL::BufferObj::ISA = OpenCL::Buffer::; |
|
|
1344 | |
|
|
1345 | @OpenCL::Image2D::ISA = |
|
|
1346 | @OpenCL::Image3D::ISA = OpenCL::Image::; |
|
|
1347 | |
|
|
1348 | @OpenCL::UserEvent::ISA = OpenCL::Event::; |
|
|
1349 | } |
2079 | } |
|
|
2080 | |
|
|
2081 | =back |
|
|
2082 | |
|
|
2083 | =head2 THE OpenCL::MappedBuffer CLASS |
|
|
2084 | |
|
|
2085 | This is a subclass of OpenCL::Mapped, representing mapped buffers. |
|
|
2086 | |
|
|
2087 | =head2 THE OpenCL::MappedImage CLASS |
|
|
2088 | |
|
|
2089 | This is a subclass of OpenCL::Mapped, representing mapped images. |
|
|
2090 | |
|
|
2091 | =over 4 |
|
|
2092 | |
|
|
2093 | =item $bytes = $mapped->row_pitch |
|
|
2094 | |
|
|
2095 | =item $bytes = $mapped->slice_pitch |
|
|
2096 | |
|
|
2097 | Return the row or slice pitch of the image that has been mapped. |
|
|
2098 | |
|
|
2099 | =back |
|
|
2100 | |
|
|
2101 | |
|
|
2102 | =cut |
1350 | |
2103 | |
1351 | 1; |
2104 | 1; |
1352 | |
2105 | |
1353 | =head1 AUTHOR |
2106 | =head1 AUTHOR |
1354 | |
2107 | |