We're trying to move to more of the new style intrinsics with include
the correct target name, and map directly to ISA instructions.
v2:
- Only do this with LLVM 3.8 and newer.
---
src/gallium/drivers/radeonsi/si_shader.c | 17 -
1 file changed, 16 insertions(+), 1 deletion(-)
The ds_bpermute instruction allows threads to transfer data directly
to or from the vgprs of other threads. These instructions use the lds
hardware to transfer data, but do not read or write lds memory.
DDX BEFORE:| DDX AFTER:
|
v_mbcnt_
The range metadata tells LLVM the range of expected values for this intrinsic,
so it can do some additional optimizations on the result.
---
src/gallium/drivers/radeonsi/si_shader.c | 29 ++---
1 file changed, 26 insertions(+), 3 deletions(-)
diff --git a/src/gallium/drive
On Tue, Apr 19, 2016 at 07:39:13PM +0100, Emil Velikov wrote:
> Hi Chuck,
>
> Thanks for chipping in.
>
> On 19 April 2016 at 15:47, Chuck Atkins wrote:
> > This still doesn't quite give what you want. One can also have an llvm with
> > component shared libs. So there's three different options
On Tue, Apr 19, 2016 at 08:12:08PM +0200, Michael Schellenberger Costa wrote:
> Hi Tom,
>
> Am 19.04.2016 um 19:52 schrieb Tom Stellard:
> > The range metadata tells LLVM the range of expected values for this
> > intrinsic,
> > so it can do some additional
On Thu, Apr 21, 2016 at 06:28:15PM +0200, Bas Nieuwenhuizen wrote:
> shader->config is not updated for compute kernels.
>
> Signed-off-by: Bas Nieuwenhuizen
This fixes compute shaders that use scratch. Thanks.
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers/rad
On Fri, Apr 24, 2015 at 12:59:53PM +0200, EdB wrote:
> Since clover should compile use -std=c++11,
> compat classes are no longer neccessary
>
> EdB (4):
> clover: remove compat class that matche std one
> clover: remove compat::string
> clover: make module::symbol::name a string
> clover:
I've pushed this patch, thanks!
-Tom
On Tue, Apr 21, 2015 at 03:49:09PM +0200, EdB wrote:
> Later we can remove the compat code
>
> According to Francisco Jerez, it happen that some time llvm requiring exotic
> compilation flags and it would be preferable not to to contaminate the rest of
> the
I've pushed patches 1-3, thanks!
-Tom
On Fri, Apr 24, 2015 at 12:59:53PM +0200, EdB wrote:
> Since clover should compile use -std=c++11,
> compat classes are no longer neccessary
>
> EdB (4):
> clover: remove compat class that matche std one
> clover: remove compat::string
> clover: make m
I've pushed this patch, thanks!
-Tom
On Wed, Apr 29, 2015 at 08:22:02AM +0200, EdB wrote:
>
> ---
> src/gallium/state_trackers/clover/Makefile.sources | 1 -
> src/gallium/state_trackers/clover/api/program.cpp | 14 +-
> .../state_trackers/clover/core/compiler.hpp| 4 +-
> src/gal
On Sat, May 02, 2015 at 01:31:41PM -0400, Ilia Mirkin wrote:
> On Sat, May 2, 2015 at 1:19 PM, EdB wrote:
> > The standard ICD file path is /etc/OpenCL/vendor/.
> > However it doesn't fit well with custom build.
> > This option allow ICD vendor file installation path override
> > ---
> > configur
On Mon, May 04, 2015 at 10:13:19AM -0400, Ilia Mirkin wrote:
> On Mon, May 4, 2015 at 10:04 AM, Tom Stellard wrote:
> > On Sat, May 02, 2015 at 01:31:41PM -0400, Ilia Mirkin wrote:
> >> On Sat, May 2, 2015 at 1:19 PM, EdB wrote:
> >> > The standard ICD f
On Tue, May 05, 2015 at 03:48:29PM +0100, Emil Velikov wrote:
> On 4 May 2015 at 18:11, Ilia Mirkin wrote:
> > On Mon, May 4, 2015 at 12:47 PM, Tom Stellard wrote:
> >> On Mon, May 04, 2015 at 10:13:19AM -0400, Ilia Mirkin wrote:
> >>> On Mon, May 4, 2015 at
:3: error: non-void function 'r600_begin_query' should return
> a value [-Wreturn-type]
> return;
> ^
>
> Signed-off-by: Vinson Lee
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers/radeon/r600_query.c | 2 +-
> 1 file change
On Thu, May 07, 2015 at 04:59:41PM +0900, Michel Dänzer wrote:
> On 05.05.2015 01:47, Tom Stellard wrote:
> > On Mon, May 04, 2015 at 10:13:19AM -0400, Ilia Mirkin wrote:
> >> On Mon, May 4, 2015 at 10:04 AM, Tom Stellard wrote:
> >>> On Sat, May 02, 2015 at 01:31
ONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE
+ * SOFTWARE.
+ *
+ * Authors: Tom Stellard
+ *
+ */
+
+#include
+
+/**
+ * \file
+ *
+ * threadsafe_context is a wrapper around a pipe_context to make it thread
+ * safe.
+ */
+
+#include "os/os_thread.h"
+#include
This consolidates signalled checks into the same place.
---
src/gallium/state_trackers/clover/core/event.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/gallium/state_trackers/clover/core/event.cpp
b/src/gallium/state_trackers/clover/core/event.cpp
index 58de888..3c933
It was possible for some events never to get triggered if one thread
was creating events and another threads was waiting for them.
This patch consolidates soft_event::wait() and hard_event::wait()
into event::wait() so that hard_event objects will now wait for
all their dependencies to be submitte
This fixes a potential crash where on a sequence like this:
Thread 0: Check if queue is not empty.
Thread 1: Remove item from queue, making it empty.
Thread 0: Do something assuming queue is not empty.
---
src/gallium/state_trackers/clover/core/queue.cpp | 2 ++
src/gallium/state_trackers/clover/
This mutex effectively prevents an event's chain or wait_count from
being updated while it is in the process of triggering. Otherwise it
may be possible to add to an event's chain after it has been triggered,
which causes the chained event to never be triggered.
---
src/gallium/state_trackers/clo
ome deadlock situations when locking is implemented.
I've pushed the two bug-fixes that you reviewed, so this series should apply
cleanly to master now.
For the series:
Tested-by: Tom Stellard
CC: stable should also be added to these.
Thanks,
Tom
> ---
> src/gallium/state_trackers/clo
Events can be added to an OpenCL command queue concurrently from
multiple threads, but pipe_context and pipe_screen objects
are not threadsafe. The threadsafe wrappers protect all pipe_screen
and pipe_context function calls with a mutex, so we can
safely use them with multiple threads.
CC: 10.5 d
CTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE
+ * SOFTWARE.
+ *
+ * Authors: Tom Stellard
+ *
+ */
+
+#include
+
+/**
+ * \file
+ *
+ * threadsafe_context is a wrapper around a pipe_context to make it thread
+ * safe.
+ *
+ * TODO: Implement all gallium functions. Only the functi
On Tue, May 26, 2015 at 10:27:34PM -0700, Vinson Lee wrote:
> TargetOptions::NoFramePointerElim was removed in llvm-3.7.0svn r238244
> "Remove NoFramePointerElim and NoFramePointerElimOverride from
> TargetOptions and remove ExecutionEngine's dependence on CodeGen. NFC.&qu
u are
attending please add your name as early as possible.
I am looking forward to seeing you there, if you have any
inquiries/questions, please send them to me (please also CC: board at
foundation.x.org).
Tom Stellard
___
mesa-dev mailin
The only use of lp_profile() is wrapped in #if defined(PROFILE),
so there is no reason to build it unless this macro is defined.
---
src/gallium/auxiliary/gallivm/lp_bld_debug.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_debug.cpp
b/
cc Francisco
On Thu, May 28, 2015 at 10:10:19AM +0200, Grigori Goronzy wrote:
> Wrap MapBuffer and MapImage as hard_event actions, like other
> operations. This enables correct profiling. Also make sure to wait
> for events to finish when blocking is requested by the caller.
> ---
> src/gallium/s
The spec says that mem objects should maintain a stack of callbacks
not just one.
v2:
- Remove stray printf.
CC: "10.3"
---
src/gallium/state_trackers/clover/core/memory.cpp | 10 ++
src/gallium/state_trackers/clover/core/memory.hpp | 3 ++-
2 files changed, 8 insertions(+), 5 deleti
The spec says that mem objects should maintain a stack of callbacks
not just one.
CC: "10.3"
---
src/gallium/state_trackers/clover/core/memory.cpp | 11 +++
src/gallium/state_trackers/clover/core/memory.hpp | 3 ++-
2 files changed, 9 insertions(+), 5 deletions(-)
diff --git a/src/gall
LLVM commit r218316 removes the JITMemoryManager class, which is
the parent for a seemingly important class in gallivm. In order to
fix the build, I've wrapped most of lp_bld_misc.cpp in
if HAVE_LLVM < 0x0306 and modifyed the
lp_build_create_jit_compiler_for_module() function to return false
for 3
On Sat, Sep 20, 2014 at 12:11:02PM +0200, Marek Olšák wrote:
> From: Marek Olšák
Patches 1-3 are:
Reviewed-by: Tom Stellard
>
> ---
> src/gallium/drivers/radeonsi/si_compute.c | 6 +-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/src/gal
On Mon, Sep 22, 2014 at 09:48:43PM +0200, Marek Olšák wrote:
> No, we cannot detect compute-only contexts yet. We need to add a new
> parameter to pipe_context::context_create which says that a context is
> compute-only. That should be OpenCL but not OpenGL.
>
> Also, some code paths like resource
This is the only guaranteed way get the patch level for llvm,
since the define cannot always be found in config.h depending
on the version of llvm or the build system used.
CC: mesa-sta...@lists.freedesktop.org
---
configure.ac | 9 -
1 file changed, 4 insertions(+), 5 deletions(-)
diff
On Thu, Sep 25, 2014 at 12:24:14PM +0900, Michel Dänzer wrote:
> From: Michel Dänzer
>
> The draw module would still try to use gallivm, causing many piglit tests
> to fail with an assertion failure. llvmpipe might have been similarly
> affected.
>
Reviewed-by: Tom Stellar
On Tue, Sep 30, 2014 at 12:29:52PM -0400, Ilia Mirkin wrote:
> Perhaps do the same thing as util_bitcount, i.e.
>
> #if defined(PIPE_CC_GCC) && (PIPE_CC_GCC_VERSION >= 304)
> return __builtin_popcountll(n);
> #else
> ...
> #endif
>
> Perhaps the gcc version check is no longer necessary, unlikel
---
.../state_trackers/clover/llvm/invocation.cpp | 30 ++
1 file changed, 30 insertions(+)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp
b/src/gallium/state_trackers/clover/llvm/invocation.cpp
index 088039a..a1a54e0 100644
--- a/src/gallium/state_tr
---
.../state_trackers/clover/llvm/invocation.cpp | 142 +++--
1 file changed, 75 insertions(+), 67 deletions(-)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp
b/src/gallium/state_trackers/clover/llvm/invocation.cpp
index 7bca0d6..3137591 100644
--- a/src
---
.../state_trackers/clover/llvm/invocation.cpp | 173 -
src/gallium/targets/opencl/Makefile.am | 1 +
2 files changed, 167 insertions(+), 7 deletions(-)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp
b/src/gallium/state_trackers/clove
Hi,
The attached patches add support to clover for compiling kernels to
native object code and it updates r600g and radeonsi to accept these
binaries from clover.
This change significantly reduces the number of compiles that the
drivers need to do. Since the OpenCL programs can now be stored in
LLVM >= 3.5.1 will be requried in order to enable PIPE_SHADER_IR_NATIVE
for compute shaders.
---
configure.ac | 4
src/gallium/drivers/radeonsi/si_compute.c | 4
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/configure.ac b/configure.ac
index
---
src/gallium/drivers/radeonsi/si_compute.c | 51 +--
src/gallium/drivers/radeonsi/si_pipe.c| 2 +-
src/gallium/drivers/radeonsi/si_shader.c | 104 ++
src/gallium/drivers/radeonsi/si_shader.h | 7 ++
4 files changed, 88 insertions(+), 76 deletio
---
src/gallium/drivers/r600/evergreen_compute.c | 95 ++
.../drivers/r600/evergreen_compute_internal.h | 15 +---
src/gallium/drivers/r600/r600_llvm.c | 71 ++--
src/gallium/drivers/r600/r600_llvm.h | 10 +++
src/gallium/drive
---
.../state_trackers/clover/llvm/invocation.cpp | 74 ++
1 file changed, 63 insertions(+), 11 deletions(-)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp
b/src/gallium/state_trackers/clover/llvm/invocation.cpp
index a1a54e0..3e6a186 100644
--- a/src
Drivers can return this value for PIPE_COMPUTE_CAP_IR_TARGET
if they want clover to give them native object code.
---
src/gallium/docs/source/screen.rst | 4 ++--
src/gallium/include/pipe/p_defines.h | 3 ++-
2 files changed, 4 insertions(+), 3 deletions(-)
diff --git a/src/gallium/docs/source/
This adds a query which allows drivers to access the config
information of a specific function within the LLVM generated ELF
binary. This makes it possible for the driver to handle ELF
binaries with multiple kernels / global functions.
---
src/gallium/drivers/radeon/r600_pipe_common.h | 8
WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
FROM,
- * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE
- * SOFTWARE.
- *
- * Authors: Tom Stellard
- *
- */
-
-#include "radeon_llvm_util.h"
-#include "util/u_memory.h"
-
-#include
-#
On Sun, Sep 28, 2014 at 09:48:11PM +0200, Niels Ole Salscheider wrote:
> On Sunday 28 September 2014, 17:44:53, Bruno Jimenez wrote:
> > Hi,
> >
> > Sorry for not answering until now, but I have had some personal issues
> > (changing university, moving to another city...)
> >
> > As you said, thi
v2:
- Split build_module_native() into three separate functions.
- Code cleanups.
---
.../state_trackers/clover/llvm/invocation.cpp | 200 -
src/gallium/targets/opencl/Makefile.am | 1 +
2 files changed, 192 insertions(+), 9 deletions(-)
diff --git a/src
This will allow us to handle internal compiler errors.
v2:
- Code cleanups.
---
.../state_trackers/clover/llvm/invocation.cpp | 27 ++
1 file changed, 27 insertions(+)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp
b/src/gallium/state_trackers/clov
On Sun, Sep 28, 2014 at 12:57:22PM +0200, EdB wrote:
> ---
> src/gallium/state_trackers/clover/api/dispatch.cpp | 2 +-
> src/gallium/state_trackers/clover/api/program.cpp | 39
> +++---
> .../state_trackers/clover/core/compiler.hpp| 12 ---
> src/gallium/state_track
On Wed, Oct 08, 2014 at 09:15:39AM -0700, Matt Arsenault wrote:
>
> On Oct 6, 2014, at 12:44 PM, Tom Stellard wrote:
>
> > ---
> > .../state_trackers/clover/llvm/invocation.cpp | 74
> > ++
> > 1 file changed, 63 insertions(+), 11 delet
There are two debug variables:
CLOVER_DEBUG which you can set to any combination of llvm,clc,asm
(separated by commas) to dump llvm IR, OpenCL C, and native assembly.
CLOVER_DEBUG_FILE which you can set to a file name for dumping output
instead of stderr. If you set this variable, the output wil
v2:
- Split build_module_native() into three separate functions.
- Code cleanups.
v3:
- More cleanups.
---
.../state_trackers/clover/llvm/invocation.cpp | 208 -
src/gallium/targets/opencl/Makefile.am | 1 +
2 files changed, 200 insertions(+), 9 deleti
v2:
- Code cleanups.
---
.../state_trackers/clover/llvm/invocation.cpp | 141 +++--
1 file changed, 74 insertions(+), 67 deletions(-)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp
b/src/gallium/state_trackers/clover/llvm/invocation.cpp
index 7bca0d6..b
Drivers can return this value for PIPE_COMPUTE_CAP_IR_TARGET
if they want clover to give them native object code.
Reviewed-by: Francisco Jerez
---
src/gallium/docs/source/screen.rst | 4 ++--
src/gallium/include/pipe/p_defines.h | 3 ++-
2 files changed, 4 insertions(+), 3 deletions(-)
diff -
This will allow us to handle internal compiler errors.
v2:
- Code cleanups.
v3:
- More cleanups.
---
.../state_trackers/clover/llvm/invocation.cpp | 25 ++
1 file changed, 25 insertions(+)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp
b/src/gal
On Thu, Oct 09, 2014 at 09:22:46PM +0200, EdB wrote:
> On Thursday, October 09, 2014 06:29:40 AM Tom Stellard wrote:
> > On Sun, Sep 28, 2014 at 12:57:22PM +0200, EdB wrote:
> > > ---
> > >
> > > src/gallium/state_trackers/clover/api/dispatch.cpp | 2 +-
>
On Fri, Oct 10, 2014 at 07:51:40PM +0200, EdB wrote:
> On Friday 10 October 2014 10:16:08 Tom Stellard wrote:
> > On Thu, Oct 09, 2014 at 09:22:46PM +0200, EdB wrote:
> > > On Thursday, October 09, 2014 06:29:40 AM Tom Stellard wrote:
> > > > On Sun, Sep 28, 2014
On Tue, Oct 14, 2014 at 03:27:58AM +0200, David Heidelberger wrote:
>
> This code is already in if (!variable->C->is_r500) so no need check
> twice.
>
I've pushed this patch, thanks!
-Tom
> Signed-off-by: David Heidelberger
> ---
> src/gallium/drivers/r300/compiler/radeon_pair_regalloc.c |
v2:
- Drop dependency on LLVM >= 3.5.1
---
src/gallium/drivers/r600/evergreen_compute.c | 167 +
.../drivers/r600/evergreen_compute_internal.h | 12 +-
src/gallium/drivers/r600/r600_llvm.c | 71 ++---
src/gallium/drivers/r600/r600_llvm.h
v2:
- Drop dependency on LLVM >= 3.5.1
- Rename si_create_shader() to si_shader_binary_read()
---
src/gallium/drivers/radeonsi/si_compute.c | 79 ++-
src/gallium/drivers/radeonsi/si_pipe.c| 4 ++
src/gallium/drivers/radeonsi/si_shader.c | 104 ++-
This adds a query which allows drivers to access the config
information of a specific function within the LLVM generated ELF
binary. This makes it possible for the driver to handle ELF
binaries with multiple kernels / global functions.
---
src/gallium/drivers/radeon/r600_pipe_common.h | 8
On Fri, Oct 17, 2014 at 05:34:40PM +0200, Vincent DEDIEU wrote:
> Hi
>
>
>
> I compiled the very last mesa package (V10.4.0-devel) on Ubuntu-14.04 with
> AMD RADEON.
>
>
>
> On a test program when I used clCreateProgramWithSource() and
> clBuildProgram(), my kernel works perfectly.
>
>
>
On Mon, Oct 20, 2014 at 05:44:53PM -0400, Jan Vesely wrote:
> From: Jan Vesely
>
> CC: Tom Stellard
> CC: Emil Velikov
> CC: Francisco Jerez
> Signed-off-by: Jan Vesely
> ---
>
> This fixes clover build failures when building without radeon
>
> configu
t; CXX llvm/libclllvm_la-invocation.lo
> llvm/invocation.cpp: In function 'void {anonymous}::optimize(llvm::Module*,
> unsigned int, const std::vector&)':
> llvm/invocation.cpp:324:18: error: expected type-specifier
>PM.add(new llvm::DataLayoutPas
On Mon, Oct 20, 2014 at 01:06:34PM +0900, Michel Dänzer wrote:
> On 17.10.2014 08:41, Tom Stellard wrote:
> >v2:
> > - Drop dependency on LLVM >= 3.5.1
>
> Have you tested this patch with LLVM < 3.6?
Yes, I tested them with 3.5 on both radeonsi and r600.
>
&
On Mon, Oct 20, 2014 at 10:53:21PM +0200, Christopher Egert wrote:
> sqrt is not supported by the LLVM backend.
>
What version of LLVM are you using, and what error message are you seeing?
-Tom
> Fixes rendering errors in Passing by Still.
> http://demozoo.org/productions/156/
>
> Signed-off-b
This was a regression introduced by
611d66fe4513e53bde052dd2bab95d448c909a2a
Passing a binary program to clBuildProgram() is legal, but passing one
to clCompileProgram() is not.
---
src/gallium/state_trackers/clover/api/program.cpp | 21 -
1 file changed, 16 insertions(+), 5 d
This factors out the validation that is common with clBuildProgram().
---
src/gallium/state_trackers/clover/api/program.cpp | 36 ---
1 file changed, 26 insertions(+), 10 deletions(-)
diff --git a/src/gallium/state_trackers/clover/api/program.cpp
b/src/gallium/state_trackers/
On Tue, Oct 21, 2014 at 12:16:12PM -0400, Jan Vesely wrote:
> On Thu, 2014-10-09 at 11:07 -0400, Tom Stellard wrote:
> > v2:
> > - Split build_module_native() into three separate functions.
> > - Code cleanups.
> >
> > v3:
> > - More cleanups.
>
On Wed, Oct 08, 2014 at 12:06:20PM +0300, Francisco Jerez wrote:
> Tom Stellard writes:
>
> > v2:
> > - Split build_module_native() into three separate functions.
> > - Code cleanups.
> > ---
> > .../state_trackers/clover/llvm/invocation.cpp | 200
On Wed, Oct 15, 2014 at 05:32:11PM -0700, Kenneth Graunke wrote:
> Consider GLSL code such as:
>
>const ivec2 offsets[] =
> ivec2[](ivec2(-1, -1), ivec2(-1, 0), ivec2(-1, 1),
> ivec2(0, -1), ivec2(0, 0), ivec2(0, 1),
> ivec2(1, -1), ivec2(1, 0), ivec2(1, 1
On Sun, Nov 02, 2014 at 08:03:31PM +0200, Francisco Jerez wrote:
> EdB writes:
>
> > should trigger CL_INVALID_VALUE
> > if device_list is NULL and num_devices is greater than zero.
> >
> > introduced by e5468dfa523be2a7a0d04bb9efcf8ae780957563
>
> Tom, can you just drop the the vector of device
Should trigger CL_INVALID_VALUE if device_list is NULL and num_devices
is greater than zero.
Introduced by e5468dfa523be2a7a0d04bb9efcf8ae780957563
Reported by: EdB
---
Hi Francisco,
I understand what you are saying now about why we don't need to pass the
vector of devices. It's because the de
On Thu, Nov 06, 2014 at 11:46:41AM -0500, Jan Vesely wrote:
> Signed-off-by: Jan Vesely
I've pushed this, thanks!
-Tom
> ---
> src/gallium/state_trackers/clover/llvm/invocation.cpp | 4
> 1 file changed, 4 insertions(+)
>
> diff --git a/src/gallium/state_trackers/clover/llvm/invocation.c
On Wed, Nov 12, 2014 at 07:36:04PM +0100, Laurent Carlier wrote:
> Le jeudi 6 novembre 2014, 09:45:40 Tom Stellard a écrit :
> > On Thu, Nov 06, 2014 at 11:46:41AM -0500, Jan Vesely wrote:
> > > Signed-off-by: Jan Vesely
> >
> > I've pushed this, thanks!
> &
On Fri, Nov 07, 2014 at 04:52:25PM +, jfons...@vmware.com wrote:
> From: José Fonseca
>
Hi Jose,
This patch is causing random segfaults with OpenCL programs on radeonsi.
I haven't been able to figure out exactly what is happening, so I was
hoping you could help.
I think the problem has som
;t think that the Windows deadlock ever happens on Linux.
>
This solution works for me. Feel free to commit.
I wonder if the problem may be the pipe-loader is unloading pipe_swrast.so
before all the threads have finished.
-Tom
> Jose
>
>
>
On Thu, Nov 20, 2014 at 10:21:07PM +0100, Marek Olšák wrote:
> From: Marek Olšák
>
Reviewed-by: Tom Stellard
> So far it has been compiled into pretty ugly code (8 instructions or so
> for either opcode).
> ---
> src/gallium/drivers/radeonsi/si_shader.c | 7 +++
&g
On Sun, Mar 15, 2015 at 07:59:52PM +0100, Marek Olšák wrote:
> From: Marek Olšák
>
> v2: Don't use the intrinsics, the shader backend can recognize these
> patterns and generates optimal code automatically.
Reviewed-by: Tom Stellard
> ---
> .../drivers/radeon/r
On Sun, Mar 15, 2015 at 07:59:51PM +0100, Marek Olšák wrote:
> From: Marek Olšák
>
> This will be used a lot (especially by tessellation).
>
> v2: don't use the bfe intrinsic
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers
On Wed, Mar 11, 2015 at 03:27:18PM +0100, Giuseppe Bilotta wrote:
> The only hackish ones are llvmpipe and softpipe, which currently return
> the same string as for get_vendor(), while ideally they should return
> the CPU vendor.
>
> Signed-off-by: Giuseppe Bilotta
Reviewed-b
The storage size for local kernel args can be queried before the
arguments are set by using the CL_KERNEL_LOCAL_MEM_SIZE param
of clGetKernelWorkGroupInfo().
The spec says that if local kernel arguments have not been specified,
then we should assume their size is 0.
---
src/gallium/state_trackers
---
src/gallium/state_trackers/clover/api/device.cpp | 8 +++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/src/gallium/state_trackers/clover/api/device.cpp
b/src/gallium/state_trackers/clover/api/device.cpp
index 04f293d..43e7475 100644
--- a/src/gallium/state_trackers/clover
---
src/gallium/drivers/radeonsi/si_pipe.c | 6 +-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/src/gallium/drivers/radeonsi/si_pipe.c
b/src/gallium/drivers/radeonsi/si_pipe.c
index f1a5388..545c156 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.c
+++ b/src/gallium/drivers/
This is required by the spec.
---
src/gallium/state_trackers/clover/api/device.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/gallium/state_trackers/clover/api/device.cpp
b/src/gallium/state_trackers/clover/api/device.cpp
index 5d1f4ab..04f293d 100644
--- a/src/galliu
---
src/gallium/state_trackers/clover/api/device.cpp | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/gallium/state_trackers/clover/api/device.cpp
b/src/gallium/state_trackers/clover/api/device.cpp
index 43e7475..bc93f91 100644
--- a/src/gallium/state_trackers/clover/api/device.cpp
+++
On Sun, Mar 22, 2015 at 07:20:58AM +0100, Giuseppe Bilotta wrote:
> OpenCL (as opposed to OpenGL) has separate vendor strings for the
> implementation/driver/platform and the device. CL_PLATFORM_VENDOR
> is akin to the GL_VENDOR string, while CL_DEVICE_VENDOR is supposed to
> return the actual devi
Hi,
I just did a fetch from mesa git and I received a force update from mesa/master.
The current head commit form that remote is:
commit 1e02f2badfd925739099bcfaa55ca99a484470d8
Author: Tiziano Bacocco
Date: Fri Feb 6 19:03:41 2015 +0100
nv50,nvc0: Mark PIPE_QUERY_TIMESTAMP_DISJOINT as re
On Mon, Mar 23, 2015 at 07:53:06AM -0700, Tom Stellard wrote:
> Hi,
>
> I just did a fetch from mesa git and I received a force update from
> mesa/master.
> The current head commit form that remote is:
>
> commit 1e02f2badfd925739099bcfaa55ca99a484470d8
> Author: Tizian
The storage size for local kernel args can be queried before the
arguments are set by using the CL_KERNEL_LOCAL_MEM_SIZE param
of clGetKernelWorkGroupInfo().
The spec says that if local kernel arguments have not been specified,
then we should assume their size is 0.
v2:
- Implement using c++11
v2:
- Fix typo
---
src/gallium/drivers/radeonsi/si_pipe.c | 6 +-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/src/gallium/drivers/radeonsi/si_pipe.c
b/src/gallium/drivers/radeonsi/si_pipe.c
index d335bda..c7773b7 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.c
+++ b/src
This patch enables clover to return the correct value for
CL_KERNEL_COMPILE_WORK_GROUP_SIZE and also verify that the correct
local_work_size is used when enqueuing kernels with this attribute.
---
src/gallium/state_trackers/clover/api/kernel.cpp | 9 +++-
src/gallium/state_trackers/clover/core/
---
.../state_trackers/clover/llvm/invocation.cpp | 71 +-
1 file changed, 41 insertions(+), 30 deletions(-)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp
b/src/gallium/state_trackers/clover/llvm/invocation.cpp
index 4da62b9..28198a5 100644
--- a/src
On Tue, Mar 24, 2015 at 06:51:31PM +0200, Francisco Jerez wrote:
> Tom Stellard writes:
>
> > ---
> > .../state_trackers/clover/llvm/invocation.cpp | 71
> > +-
> > 1 file changed, 41 insertions(+), 30 deletions(-)
> >
>
> T
Cc: 10.5 10.4
---
src/gallium/state_trackers/clover/core/program.cpp | 6 +-
src/gallium/state_trackers/clover/core/program.hpp | 1 +
2 files changed, 6 insertions(+), 1 deletion(-)
diff --git a/src/gallium/state_trackers/clover/core/program.cpp
b/src/gallium/state_trackers/clover/core/pro
On Tue, Mar 24, 2015 at 10:36:19PM +0200, Francisco Jerez wrote:
> Jan Vesely writes:
>
> > On Tue, 2015-03-24 at 22:24 +0200, Francisco Jerez wrote:
> >> Tom Stellard writes:
> >>
> >> > Cc: 10.5 10.4
> >> > ---
> >
Cc: 10.5 10.4
---
src/gallium/state_trackers/clover/api/program.cpp | 4
1 file changed, 4 insertions(+)
diff --git a/src/gallium/state_trackers/clover/api/program.cpp
b/src/gallium/state_trackers/clover/api/program.cpp
index 60184ed..fcec1d7 100644
--- a/src/gallium/state_trackers/clover/
Cc: 10.5 10.4
---
src/gallium/state_trackers/clover/core/event.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/gallium/state_trackers/clover/core/event.cpp
b/src/gallium/state_trackers/clover/core/event.cpp
index 58de888..9d78b48 100644
--- a/src/gallium/state_tracker
On Wed, Mar 25, 2015 at 04:35:03PM -0400, Ilia Mirkin wrote:
> On Wed, Mar 25, 2015 at 4:27 PM, Dave Airlie wrote:
> > On 26 March 2015 at 06:07, Ilia Mirkin wrote:
> >> So what do you do when someone goes to shadertoy.com which on
> >> average uses 1000 temps?
> >
> > Fall over in a heap, li
901 - 1000 of 1703 matches
Mail list logo
