[ViewVC] Diff of: cvs/OpenCL/OpenCL.pm

Comparing OpenCL/OpenCL.pm (file contents):
Revision 1.62 by root, Sat Apr 28 13:33:04 2012 UTC vs.
Revision 1.70 by root, Thu May 3 23:32:47 2012 UTC

…		…
212	$ev->wait;	212	$ev->wait;
213		213
214	=head2 Use the OpenGL module to share a texture between OpenCL and OpenGL and draw some julia	214	=head2 Use the OpenGL module to share a texture between OpenCL and OpenGL and draw some julia
215	set tunnel effect.	215	set tunnel effect.
216		216
217	This is quite a long example to get you going.	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>.
218		219
219	use OpenGL ":all";	220	use OpenGL ":all";
220	use OpenCL;	221	use OpenCL;
221		222
		223	my $S = $ARGV[0] \|\| 256; # window/texture size, smaller is faster
		224
222	# open a window and create a gl texture	225	# open a window and create a gl texture
223	OpenGL::glpOpenWindow width => 256, height => 256;	226	OpenGL::glpOpenWindow width => $S, height => $S;
224	my $texid = glGenTextures_p 1;	227	my $texid = glGenTextures_p 1;
225	glBindTexture GL_TEXTURE_2D, $texid;	228	glBindTexture GL_TEXTURE_2D, $texid;
226	glTexImage2D_c GL_TEXTURE_2D, 0, GL_RGBA8, 256, 256, 0, GL_RGBA, GL_UNSIGNED_BYTE, 0;	229	glTexImage2D_c GL_TEXTURE_2D, 0, GL_RGBA8, $S, $S, 0, GL_RGBA, GL_UNSIGNED_BYTE, 0;
227		230
228	# find and use the first opencl device that let's us get a shared opengl context	231	# find and use the first opencl device that let's us get a shared opengl context
229	my $platform;	232	my $platform;
230	my $dev;	233	my $dev;
231	my $ctx;	234	my $ctx;
…		…
250	# now the boring opencl code	253	# now the boring opencl code
251	my $src = <<EOF;	254	my $src = <<EOF;
252	kernel void	255	kernel void
253	juliatunnel (write_only image2d_t img, float time)	256	juliatunnel (write_only image2d_t img, float time)
254	{	257	{
255	float2 p = (float2)(get_global_id (0), get_global_id (1)) / 256.f * 2.f - 1.f;	258	int2 xy = (int2)(get_global_id (0), get_global_id (1));
		259	float2 p = convert_float2 (xy) / $S.f * 2.f - 1.f;
256		260
257	float2 m = (float2)(1.f, p.y) / fabs (p.x);	261	float2 m = (float2)(1.f, p.y) / fabs (p.x); // tunnel
258	m.x = fabs (fmod (m.x + time * 0.05f, 4.f)) - 2.f;	262	m.x = fabs (fmod (m.x + time * 0.05f, 4.f) - 2.f);
259		263
260	float2 z = m;	264	float2 z = m;
261	float2 c = (float2)(sin (time * 0.05005), cos (time * 0.06001));	265	float2 c = (float2)(sin (time * 0.01133f), cos (time * 0.02521f));
262		266
263	for (int i = 0; i < 25 && dot (z, z) < 4.f; ++i)	267	for (int i = 0; i < 25 && dot (z, z) < 4.f; ++i) // standard julia
264	z = (float2)(z.x * z.x - z.y * z.y, 2.f * z.x * z.y) + c;	268	z = (float2)(z.x * z.x - z.y * z.y, 2.f * z.x * z.y) + c;
265		269
266	float3 colour = (float3)(z.x, z.y, z.x * z.y);	270	float3 colour = (float3)(z.x, z.y, atan2 (z.y, z.x));
267	write_imagef (img, (int2)(get_global_id (0), get_global_id (1)), (float4)(colour * p.x * p.x, 1.));	271	write_imagef (img, xy, (float4)(colour * p.x * p.x, 1.));
268	}	272	}
269	EOF	273	EOF
270		274
271	my $prog = $ctx->build_program ($src);	275	my $prog = $ctx->build_program ($src);
272	my $kernel = $prog->kernel ("juliatunnel");	276	my $kernel = $prog->kernel ("juliatunnel");
…		…
276	for (my $time; ; ++$time) {	280	for (my $time; ; ++$time) {
277	# acquire objects from opengl	281	# acquire objects from opengl
278	$queue->acquire_gl_objects ([$tex]);	282	$queue->acquire_gl_objects ([$tex]);
279		283
280	# configure and run our kernel	284	# configure and run our kernel
281	$kernel->set_image2d (0, $tex);	285	$kernel->setf ("mf", $tex, $time*2); # mf = memory object, float
282	$kernel->set_float (1, $time);
283	$queue->nd_range_kernel ($kernel, undef, [256, 256], undef);	286	$queue->nd_range_kernel ($kernel, undef, [$S, $S], undef);
284		287
285	# release objects to opengl again	288	# release objects to opengl again
286	$queue->release_gl_objects ([$tex]);	289	$queue->release_gl_objects ([$tex]);
287		290
288	# wait	291	# wait
…		…
301		304
302	glXSwapBuffers;	305	glXSwapBuffers;
303		306
304	select undef, undef, undef, 1/60;	307	select undef, undef, undef, 1/60;
305	}	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>.
306		336
307	=head1 DOCUMENTATION	337	=head1 DOCUMENTATION
308		338
309	=head2 BASIC CONVENTIONS	339	=head2 BASIC CONVENTIONS
310		340
…		…
499	use Async::Interrupt ();	529	use Async::Interrupt ();
500		530
501	our $POLL_FUNC; # set by XS	531	our $POLL_FUNC; # set by XS
502		532
503	BEGIN {	533	BEGIN {
504	our $VERSION = '0.98';	534	our $VERSION = '0.99';
505		535
506	require XSLoader;	536	require XSLoader;
507	XSLoader::load (__PACKAGE__, $VERSION);	537	XSLoader::load (__PACKAGE__, $VERSION);
508		538
509	@OpenCL::Platform::ISA =	539	@OpenCL::Platform::ISA =
…		…
527	@OpenCL::Image1D::ISA =	557	@OpenCL::Image1D::ISA =
528	@OpenCL::Image1DArray::ISA =	558	@OpenCL::Image1DArray::ISA =
529	@OpenCL::Image1DBuffer::ISA = OpenCL::Image::;	559	@OpenCL::Image1DBuffer::ISA = OpenCL::Image::;
530		560
531	@OpenCL::UserEvent::ISA = OpenCL::Event::;	561	@OpenCL::UserEvent::ISA = OpenCL::Event::;
		562
		563	@OpenCL::MappedBuffer::ISA =
		564	@OpenCL::MappedImage::ISA = OpenCL::Mapped::;
532	}	565	}
533		566
534	=head2 THE OpenCL PACKAGE	567	=head2 THE OpenCL PACKAGE
535		568
536	=over 4	569	=over 4
…		…
1009	$prog = $self->program_with_source ($prog)	1042	$prog = $self->program_with_source ($prog)
1010	unless ref $prog;	1043	unless ref $prog;
1011		1044
1012	eval { $prog->build (undef, $options); 1 }	1045	eval { $prog->build (undef, $options); 1 }
1013	or errno == BUILD_PROGRAM_FAILURE	1046	or errno == BUILD_PROGRAM_FAILURE
		1047	or errno == INVALID_BINARY # workaround nvidia bug
1014	or Carp::croak "OpenCL::Context->build_program: " . err2str;	1048	or Carp::croak "OpenCL::Context->build_program: " . err2str;
1015		1049
1016	# we check status for all devices	1050	# we check status for all devices
1017	for my $dev ($self->devices) {	1051	for my $dev ($self->devices) {
1018	$prog->build_status ($dev) == BUILD_SUCCESS	1052	$prog->build_status ($dev) == BUILD_SUCCESS
…		…
1124		1158
1125	Creates a new OpenCL::Program object from the given source code.	1159	Creates a new OpenCL::Program object from the given source code.
1126		1160
1127	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>
1128		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
1129	=item $packed_value = $ctx->info ($name)	1174	=item $packed_value = $ctx->info ($name)
1130		1175
1131	See C<< $platform->info >> for details.	1176	See C<< $platform->info >> for details.
1132		1177
1133	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>
…		…
1203		1248
1204	=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...)	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...)
1205		1250
1206	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
1207		1252
		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
1208	=item $ev = $queue->read_image ($src, $blocking, $x, $y, $z, $width, $height, $depth, $row_pitch, $slice_pitch, $data, $wait_events...)	1257	=item $ev = $queue->read_image ($src, $blocking, $x, $y, $z, $width, $height, $depth, $row_pitch, $slice_pitch, $data, $wait_events...)
1209		1258
1210	L<http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/clEnqueueCopyBufferRect.html>	1259	C<$row_pitch> (and C<$slice_pitch>) can be C<0>, in which case the OpenCL
1211		1260	module uses the image width (and height) to supply default values.
1212	=item $ev = $queue->copy_buffer_to_image ($src_buffer, $dst_image, $src_offset, $dst_x, $dst_y, $dst_z, $width, $height, $depth, $wait_events...)
1213		1261
1214	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>
1215		1263
1216	=item $ev = $queue->write_image ($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, $wait_events...)
1217		1265
		1266	C<$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.
1218	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>
1219		1269
1220	=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...)	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...)
1221		1271
1222	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>
…		…
1329		1379
1330	=for gengetinfo end command_queue	1380	=for gengetinfo end command_queue
1331		1381
1332	=back	1382	=back
1333		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.
		1448
		1449	=back
		1450
1334	=head2 THE OpenCL::Memory CLASS	1451	=head2 THE OpenCL::Memory CLASS
1335		1452
1336	This the superclass of all memory objects - OpenCL::Buffer, OpenCL::Image,	1453	This the superclass of all memory objects - OpenCL::Buffer, OpenCL::Image,
1337	OpenCL::Image2D and OpenCL::Image3D.	1454	OpenCL::Image2D and OpenCL::Image3D.
1338		1455
…		…
1528	If a callback is specified, then it will be called when compilation is	1645	If a callback is specified, then it will be called when compilation is
1529	finished. Note that many OpenCL implementations block your program while	1646	finished. Note that many OpenCL implementations block your program while
1530	compiling whether you use a callback or not. See C<build_async> if you	1647	compiling whether you use a callback or not. See C<build_async> if you
1531	want to make sure the build is done in the background.	1648	want to make sure the build is done in the background.
1532		1649
1533	Note that some OpenCL implementations atc up badly, and don't call the	1650	Note that some OpenCL implementations act up badly, and don't call the
1534	callback in some error cases (but call it in others). This implementation	1651	callback in some error cases (but call it in others). This implementation
1535	assumes the callback will always be called, and leaks memory if this is	1652	assumes the callback will always be called, and leaks memory if this is
1536	not so. So best make sure you don't pass in invalid values.	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.
1537		1657
1538	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>
1539		1659
1540	=item $program->build_async (\@devices = undef, $options = "", $cb->($program) = undef)	1660	=item $program->build_async (\@devices = undef, $options = "", $cb->($program) = undef)
1541		1661
…		…
1876		1996
1877	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>
1878		1998
1879	=back	1999	=back
1880		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
		2075	=cut
		2076
		2077	sub OpenCL::Mapped::get {
		2078	substr ${$_[0]}, $_[1], $_[2]
		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
1881	=cut	2102	=cut
1882		2103
1883	1;	2104	1;
1884		2105
1885	=head1 AUTHOR	2106	=head1 AUTHOR

Diff Legend

-–
+Removed lines
-+
+Added lines
-<
+Changed lines
->
+Changed lines

Comparing OpenCL/OpenCL.pm (file contents): Revision 1.62 by root, Sat Apr 28 13:33:04 2012 UTC vs. Revision 1.70 by root, Thu May 3 23:32:47 2012 UTC

Diff Legend

Comparing OpenCL/OpenCL.pm (file contents):
Revision 1.62 by root, Sat Apr 28 13:33:04 2012 UTC vs.
Revision 1.70 by root, Thu May 3 23:32:47 2012 UTC