---
configure.ac | 13 -
src/gallium/drivers/radeon/radeon_llvm_emit.c | 8
2 files changed, 16 insertions(+), 5 deletions(-)
diff --git a/configure.ac b/configure.ac
index d32aa24..eda8d23 100644
--- a/configure.ac
+++ b/configure.ac
@@ -204
On Sat, Jun 13, 2015 at 08:16:23PM +0200, Marek Olšák wrote:
> From: Marek Olšák
>
Reviewed-by: Tom Stellard
> ---
> configure.ac | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/configure.ac b/configure.ac
> index 34d1ac9..e6d947e 100644
>
---
si-report.py | 8 +++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/si-report.py b/si-report.py
index fbeed11..ec88112 100755
--- a/si-report.py
+++ b/si-report.py
@@ -145,7 +145,13 @@ class si_stats:
self.__dict__[name] += 1
def is_empty(self):
-
On Tue, Jun 16, 2015 at 08:07:57PM +0100, Emil Velikov wrote:
> On 13 June 2015 at 19:16, Marek Olšák wrote:
> > From: Marek Olšák
> >
> > ---
> > configure.ac | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/configure.ac b/configure.ac
> > index 34d1ac9..e6d947e 10
On Mon, Jun 22, 2015 at 12:23:54PM +0200, Marek Olšák wrote:
> On Mon, Jun 22, 2015 at 5:36 AM, Ilia Mirkin wrote:
> > On Sun, Jun 21, 2015 at 11:33 PM, Michel Dänzer wrote:
> >> On 22.06.2015 00:31, Ilia Mirkin wrote:
> >>> On Sun, Jun 21, 2015 at 12:22 PM, Emil Velikov
> >>> wrote:
> On
On Tue, Jun 23, 2015 at 08:18:45PM -0400, Rob Clark wrote:
> On Tue, Jun 23, 2015 at 7:27 PM, Dylan Baker wrote:
> > I have a couple of python pointers for you, feel free to take them or
> > leave them.
>
> cool, thanks..
>
> What do others think about including shadertoy in shader-db? If it is
On Thu, Jun 25, 2015 at 03:19:40PM +0100, Emil Velikov wrote:
> On 3 June 2015 at 14:27, Francisco Jerez wrote:
> > Emil Velikov writes:
> >
> >> Hi Tom,
> >>
> >> On 31 March 2015 at 15:29, Francisco Jerez wrote:
> >>> Tom Stellard writ
On Tue, Jul 07, 2015 at 05:43:19PM +0100, Emil Velikov wrote:
> On 30/06/15 16:09, Emil Velikov wrote:
> > Hello all,
> >
> > As mentioned over IRC a few weeks back, here is a series that removes
> > support for non-render node devices.
> >
> > The two main motivations being:
> > - Currently we
pipe_context::flush() can return a NULL fence if the queue is already
empty, so we should not assume that an event with a NULL fence
has the status of CL_QUEUED.
CC: 10.6
---
src/gallium/state_trackers/clover/core/event.cpp | 7 ---
src/gallium/state_trackers/clover/core/event.hpp | 1 +
2 f
VENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, 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
Events can be added to an OpenCL command queue concurrently from multiple
threads, but pipe_context bjects are not threadsafe. The threadsafe
wrappers protect all pipe_context function calls with a mutex, so we
can safely use them with multiple threads.
v2:
- Don't use wrapper for pipe_screen.
On Thu, Jul 16, 2015 at 06:55:26PM +0200, Marek Olšák wrote:
> From: Marek Olšák
>
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers/radeonsi/si_compute.c | 2 +-
> src/gallium/drivers/radeonsi/si_shader.c | 8 +++-
> src/gallium/drivers/radeonsi/si_shader.h
On Thu, Jul 16, 2015 at 06:55:27PM +0200, Marek Olšák wrote:
> From: Marek Olšák
>
Reviewed-by: Tom Stellard
> Cc: 10.5 10.6
> ---
> src/gallium/drivers/radeonsi/si_shader.c| 49
> ++---
> src/gallium/drivers/radeonsi/si_shader.h
On Sat, Jul 11, 2015 at 02:35:53PM +0300, Francisco Jerez wrote:
> Tom Stellard writes:
>
> > pipe_context::flush() can return a NULL fence if the queue is already
> > empty, so we should not assume that an event with a NULL fence
> > has the status of CL_QUEUED.
> &
All LLVM API calls that require an ostream object have been removed from
the disassemble() function, so we don't need to use this class to wrap
_debug_printf() we can just call this function directly.
---
src/gallium/auxiliary/gallivm/lp_bld_debug.cpp | 27 +-
1 file change
Its only use is to implement a custom version of LLVMDumpValue
on some Windows and embedded platforms.
---
src/gallium/auxiliary/gallivm/lp_bld_debug.cpp | 2 ++
1 file changed, 2 insertions(+)
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_debug.cpp
b/src/gallium/auxiliary/gallivm/lp_bld_deb
This fixes crashes in some piglit tests on radeonsi that use the draw
module, and llvmpipe is likely completely broken without this on LLVM
3.8.
This is just a temporary solution. The correct solution will require
creating a TargetMachine during gallivm initialization and pulling the
DataLayout f
On Sun, Oct 11, 2015 at 03:29:43AM +0200, Marek Olšák wrote:
> From: Marek Olšák
>
> This removes "v_mov v0, 0" which typically occurs before exports.
>
Reviewed-by: Tom Stellard
> Totals:
> SGPRS: 345216 -> 344552 (-0.19 %)
> VGPRS: 197684 -> 197132 (-0.
On Sun, Oct 11, 2015 at 03:29:44AM +0200, Marek Olšák wrote:
> From: Marek Olšák
>
Reviewed-by: Tom Stellard
> No difference according to shader-db.
> ---
> src/gallium/drivers/radeon/radeon_setup_tgsi_llvm.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
On Sun, Oct 11, 2015 at 03:29:45AM +0200, Marek Olšák wrote:
> From: Marek Olšák
>
Reviewed-by: Tom Stellard
> No difference according to shader-db. (with the new S_ABS_I32 pattern)
> ---
> .../drivers/radeon/radeon_setup_tgsi_llvm.c| 60
> ++
On Sun, Oct 11, 2015 at 03:29:47AM +0200, Marek Olšák wrote:
> From: Marek Olšák
>
Reviewed-by: Tom Stellard
> The increase in VGPRs in unfortunate, but the decrease in the scratch size
> is always welcome.
>
> Totals:
> SGPRS: 344552 -> 344368 (-0.05 %)
> VG
On Sun, Oct 11, 2015 at 03:29:48AM +0200, Marek Olšák wrote:
> From: Marek Olšák
>
I don't think we should globally enable this until we are sure it does not
introduce any illegal transforms.
> Required for 1/sqrt ==> rsq.
I think the arcp fast-math flag for instruction is supposed to allow th
On Fri, Nov 13, 2015 at 02:46:52PM +0100, Hans de Goede wrote:
> Hi All,
>
> So as discussed I've started working on a TGSI backend for
> llvm to use as a way to get compute going on nouveau (and other gpu-s).
>
> I'm still learning all the ins and outs of llvm so I do not have
> much to show yet
Reviewed-by: Tom Stellard
On Fri, Jan 23, 2015 at 10:01:56PM +0100, Marek Olšák wrote:
> Reviewed-by: Marek Olšák
>
> Marek
>
> On Thu, Jan 22, 2015 at 4:41 AM, Michel Dänzer wrote:
> > From: Michel Dänzer
> >
> > Fixes writing beyond the allocated buffer:
&
what I've done wrong?
Thanks,
Tom
>From ba673155672756fb0bf9873b2ae76c3f5ccd02e2 Mon Sep 17 00:00:00 2001
From: Tom Stellard
Date: Wed, 10 Dec 2014 09:13:59 -0500
Subject: [PATCH] radeonsi: Enable VGPR spilling for all shader types v5 (WIP)
v2:
- Only emit write SPI_TMPRING_SIZE once p
On Sun, Jan 25, 2015 at 04:11:40PM -0500, Jan Vesely wrote:
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=88783
> Signed-off-by: Jan Vesely
Pushed, thanks!
> ---
> src/gallium/state_trackers/clover/llvm/invocation.cpp | 4
> 1 file changed, 4 insertions(+)
>
> diff --git a/src/
From: Marek Olšák
---
src/gallium/drivers/radeonsi/si_pm4.c | 12
src/gallium/drivers/radeonsi/si_pm4.h | 1 +
src/gallium/drivers/radeonsi/si_state_shaders.c | 4
3 files changed, 13 insertions(+), 4 deletions(-)
diff --git a/src/gallium/drivers/radeonsi
---
src/gallium/drivers/radeonsi/si_compute.c | 11 +--
src/gallium/drivers/radeonsi/si_shader.h | 1 +
2 files changed, 6 insertions(+), 6 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_compute.c
b/src/gallium/drivers/radeonsi/si_compute.c
index ba63afd..840d21f 100644
---
This moves scratch buffer allocation from si_launch_grid() to
si_create_compute_state(). This helps to reduce the overhead of
launching a kernel and also fixes a bug in the code that would cause
the scratch buffer to be too small if a kernel with smaller scratch size
was launched before a kernel w
v2:
- Only emit write SPI_TMPRING_SIZE once per packet.
- Use context global scratch buffer.
v3:
- Patch shaders using WRITE_DATA packet instead of map/unmap.
- Emit ICACHE_FLUSH, CS_PARTIAL_FLUSH, PS_PARTIAL_FLUSH, and
VS_PARTIAL_FLUSH when patching shaders.
v4:
- Code cleanups.
---
src/gallium/drivers/radeonsi/si_compute.c | 10 +-
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_compute.c
b/src/gallium/drivers/radeonsi/si_compute.c
index 4427d3b..ba63afd 100644
--- a/src/gallium/drivers/radeonsi/si_compute.c
+++ b/sr
, we'll
> just disable it until we actually have a C++ user.
>
Tested-by: Tom Stellard
> Reported-by: Tom Stellard
Should use my AMD email here too ^ or you can drop this line all together if
you want.
-Tom
> ---
> src/util/macros.h | 14 --
> 1 file c
On Thu, Feb 12, 2015 at 03:36:37PM -0800, Matt Turner wrote:
> instructions in affected programs: 968 -> 942 (-2.69%)
> helped:4
> ---
> src/mesa/drivers/dri/i965/brw_fs.cpp | 9 +
> src/mesa/drivers/dri/i965/brw_vec4.cpp | 5 +
> 2 files changed,
On Fri, Feb 13, 2015 at 09:16:17PM -0500, Shawn Starr wrote:
> v3: Fix my typo, 3rd time's a charm, right?
>
Pushed, thanks!
-Tom
> From 5712d95cb12d8e82d34fafe66bf827ee6726220e Mon Sep 17 00:00:00 2001
> From: Shawn Starr
> Date: Fri, 13 Feb 2015 20:30:01 -0500
> Subject: [PATCH] clover: Use
On Tue, Feb 17, 2015 at 05:15:42PM +0900, Michel Dänzer wrote:
> From: Michel Dänzer
>
Reviewed-by: Tom Stellard
> This reverts commit 0e9cdedd2e3943bdb7f3543a3508b883b167e427.
>
> It caused the grass to disappear in The Talos Principle.
>
> Bugzilla: https:/
On Tue, Feb 17, 2015 at 03:23:05PM +0200, Francisco Jerez wrote:
> The round-robin allocation strategy is expected to decrease the amount
> of false dependencies created by the register allocator and give the
> post-RA scheduling pass more freedom to move instructions around. On
> the other hand i
On Tue, Feb 17, 2015 at 04:41:41PM +0200, Francisco Jerez wrote:
> Tom Stellard writes:
>
> > On Tue, Feb 17, 2015 at 03:23:05PM +0200, Francisco Jerez wrote:
> >> The round-robin allocation strategy is expected to decrease the amount
> >> of false dependencies cre
On Thu, Feb 26, 2015 at 12:34:56PM -0700, Brian Paul wrote:
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89343
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers/r300/compiler/tests/radeon_compiler_optimize_tests.c | 1
> +
> 1 file changed, 1 insertion(+)
>
This should be done by the frontend for devices that support this
extension.
---
src/gallium/state_trackers/clover/llvm/invocation.cpp | 1 -
1 file changed, 1 deletion(-)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp
b/src/gallium/state_trackers/clover/llvm/invocation.cpp
i
v2:
- Report correct values for CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE and
CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE.
- Only define cl_khr_fp64 if the extension is supported.
- Remove trailing space from extension string.
- Rename device query function from cl_khr_fp64() to has_doubles().
-
---
src/gallium/drivers/radeonsi/si_pipe.c | 7 +--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_pipe.c
b/src/gallium/drivers/radeonsi/si_pipe.c
index 26182c2..c7a7622 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.c
+++ b/src/gallium/driver
This adds SGPR count, VGPR count, shader size, LDS size, and scratch
usage to shader dumps.
---
src/gallium/drivers/radeonsi/si_shader.c | 18 --
1 file changed, 12 insertions(+), 6 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_shader.c
b/src/gallium/drivers/radeonsi/
Hi,
This patch breaks the build for me:
CFLAGS="-g" CXXFLAGS="$CFLAGS" CC="ccache gcc" CXX="ccache g++"
./autogen.sh \
--prefix=/usr/local \
--with-dri-drivers="no" \
--with-gallium-drivers="r600,radeonsi" \
--enable-glx-tls \
--enable-debug \
--enable-shared-glapi \
--with-egl-platforms=x11,drm
From: Daniel Stone
This fixes almost all piglit regressions when running with
PIGLIT_PLATFORM=gbm
Tom Stellard:
- Fix ARGB2101010 format
Cc: "10.4 10.5"
---
I'm not sure if the commit message makes sense, I'm open to suggestions.
src/egl/drivers/dri2
From: Daniel Stone
This fixes almost all piglit regressions when running with
PIGLIT_PLATFORM=gbm
Tom Stellard:
- Fix ARGB2101010 format
Cc: "10.4 10.5"
---
I'm not sure if the commit message makes sense, I'm open to suggestions.
src/egl/drivers/dri2
v2:
- Report correct values for CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE
and CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE.
- Only define cl_khr_fp64 if the extension is supported.
- Remove trailing space from extension string.
- Rename device query function from cl_khr_fp86() to
has_doubles(
---
src/gallium/state_trackers/clover/api/device.cpp | 3 +--
src/gallium/state_trackers/clover/core/device.cpp | 6 ++
src/gallium/state_trackers/clover/core/device.hpp | 1 +
3 files changed, 8 insertions(+), 2 deletions(-)
diff --git a/src/gallium/state_trackers/clover/api/device.cpp
b/s
On Mon, Mar 02, 2015 at 12:54:15PM +0100, Marek Olšák wrote:
> From: Marek Olšák
>
For some reason I thought doing this would require changes to LLVM,
but I guess I was wrong.
Reviewed-by: Tom Stellard
> None of the instructions use a pointer argument.
> (+ small cosm
On Mon, Mar 02, 2015 at 12:54:16PM +0100, Marek Olšák wrote:
> From: Marek Olšák
>
> ---
> src/gallium/drivers/radeonsi/si_shader.c | 22 --
> 1 file changed, 16 insertions(+), 6 deletions(-)
>
> diff --git a/src/gallium/drivers/radeonsi/si_shader.c
> b/src/gallium/drivers/
On Mon, Mar 02, 2015 at 10:14:00PM +0100, Marek Olšák wrote:
> On Mon, Mar 2, 2015 at 10:05 PM, Tom Stellard wrote:
> > On Mon, Mar 02, 2015 at 12:54:16PM +0100, Marek Olšák wrote:
> >> From: Marek Olšák
> >>
> >> ---
> >>
> Cc: 10.4, 10.5
> Bugzilla: https://bugs.gentoo.org/show_bug.cgi?id=540970
Both patches:
Reviewed-by: Tom Stellard
Thanks!
> ---
> src/gallium/drivers/r300/compiler/tests/rc_test_helpers.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/s
v2:
- Report correct values for CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE
and CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE.
- Only define cl_khr_fp64 if the extension is supported.
- Remove trailing space from extension string.
- Rename device query function from cl_khr_fp86() to
has_doubles(
On Mon, Mar 02, 2015 at 02:09:29PM -0800, Matt Arsenault wrote:
>
> > On Mar 2, 2015, at 1:19 PM, Tom Stellard wrote:
> >
> > On Mon, Mar 02, 2015 at 10:14:00PM +0100, Marek Olšák wrote:
> >> On Mon, Mar 2, 2015 at 10:05 PM, Tom Stellard wrote:
> >>>
On Mon, Mar 02, 2015 at 12:54:20PM +0100, Marek Olšák wrote:
> From: Marek Olšák
>
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers/radeon/radeon_setup_tgsi_llvm.c | 2 ++
> src/gallium/drivers/radeonsi/si_pipe.c | 2 +-
> 2 files changed, 3 inserti
On Mon, Mar 02, 2015 at 12:54:19PM +0100, Marek Olšák wrote:
> From: Marek Olšák
Reviewed-by: Tom Stellard
>
> ---
> src/gallium/drivers/radeon/radeon_setup_tgsi_llvm.c | 2 ++
> src/gallium/drivers/radeonsi/si_pipe.c | 3 ++-
> 2 files changed, 4 inserti
On Mon, Mar 02, 2015 at 12:54:17PM +0100, Marek Olšák wrote:
> From: Marek Olšák
Reviewed-by: Tom Stellard
>
> And use AND/OR in special cases.
>
> This universal helper will be used a lot (especially by tessellation).
> ---
> src/gallium/drivers/rade
On Mon, Mar 02, 2015 at 12:54:21PM +0100, Marek Olšák wrote:
> From: Marek Olšák
>
I'm still unsure whether it's better to use intrinsics or LLVM IR
to implement these. I will think about this some more.
-Tom
> ---
> src/gallium/drivers/radeon/radeon_setup_tgsi_llvm.c | 8
> 1 file
This means dropping CL_FP_DENORM from the current return value.
---
src/gallium/state_trackers/clover/api/device.cpp | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/src/gallium/state_trackers/clover/api/device.cpp
b/src/gallium/state_trackers/clover/api/device.cpp
index b1f
On Thu, Mar 05, 2015 at 05:14:09PM +0100, Marek Olšák wrote:
> Since you acked patch #3, which depends on this, I assume this has your Rb
> too?
>
No, I still want to resolved with Matt what to do about intrinsics vs
IR.
-Tom
> Marek
>
> On Mon, Mar 2, 2015 at 10:05 PM, T
On Mon, Mar 02, 2015 at 12:54:21PM +0100, Marek Olšák wrote:
> From: Marek Olšák
>
Hi Marek,
After discussing with Matt, I think we should use LLVM IR rather than
intrinsics for IBFE and UBFE and then add patterns for them either in
the TableGen Files or AMDGPUISelDAGToDAG.cpp.
Using intrinsic
On Thu, Mar 05, 2015 at 08:42:25PM +0200, Francisco Jerez wrote:
> Tom Stellard writes:
>
> > This means dropping CL_FP_DENORM from the current return value.
> > ---
> > src/gallium/state_trackers/clover/api/device.cpp | 4 +++-
> > 1 file changed, 3 insertions(
---
src/gallium/drivers/radeonsi/si_compute.c | 3 ++-
src/gallium/drivers/radeonsi/si_shader.c | 1 +
src/gallium/drivers/radeonsi/si_shader.h | 1 +
3 files changed, 4 insertions(+), 1 deletion(-)
diff --git a/src/gallium/drivers/radeonsi/si_compute.c
b/src/gallium/drivers/radeonsi/si_comput
This means dropping CL_FP_DENORM from the current return value.
v2:
- Add comments about minimum values for OpenCL 1.2.
---
src/gallium/state_trackers/clover/api/device.cpp | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/src/gallium/state_trackers/clover/api/device.cpp
On Tue, Mar 10, 2015 at 12:42:38PM +0100, Marek Olšák wrote:
> OK. What about patches 8 an 9?
>
I think the intrinsics in 9 are OK, but 8 should be using LLVM IR.
-Tom
> Marek
>
> On Thu, Mar 5, 2015 at 8:30 PM, Tom Stellard wrote:
> > On Mon, Mar 02, 2015 at 12:54:21
bt it would be easy to
> > recognize them and match them precisely. They are also very unlikely
> > to appear open-coded in real applications, therefore, expanding them
> > and then matching them again seems like a huge waste of time. Apps
> > will likely just use LSHR+AND,
On Wed, Mar 11, 2015 at 01:27:32AM +0100, Marek Olšák wrote:
> On Wed, Mar 11, 2015 at 12:09 AM, Tom Stellard wrote:
> > On Tue, Mar 10, 2015 at 11:01:21PM +0100, Marek Olšák wrote:
> >> I've looked into how to recognize BFM and BFI and discovered that if
> >> T
Advantages of using llvm.amdgcn.s.buffer.load
- We can use a real pointer type, which LLVM can better reason about and do
alias analysis on. This will also ease the transition to using fat pointers
and LLVM IR loads.
- llvm.amdgcn.s.buffer.load is defined in IntrinsicsAMDGPU.td so passes can
Also modify build_buffer_load to always pass soffset to an intrinsic
if it is set. This is required to avoid failing buffer range checks
in some cases.
---
src/gallium/drivers/radeonsi/si_shader.c | 67 ++--
1 file changed, 20 insertions(+), 47 deletions(-)
diff --git
This was broken by: e0cc0a614c96011958bc3a1b84da9168e0e1ccbb
---
src/gallium/drivers/radeonsi/si_shader_tgsi_setup.c | 14 ++
1 file changed, 10 insertions(+), 4 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_shader_tgsi_setup.c
b/src/gallium/drivers/radeonsi/si_shader_tgs
This was broken by: e0cc0a614c96011958bc3a1b84da9168e0e1ccbb
v2:
- Use preprocessor macro
---
src/gallium/drivers/radeonsi/si_shader_tgsi_setup.c | 6 --
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_shader_tgsi_setup.c
b/src/gallium/drivers/
On Fri, Feb 24, 2017 at 03:30:50PM -0800, Matt Arsenault wrote:
>
> > On Feb 24, 2017, at 14:39, Marek Olšák wrote:
> >
> > On Fri, Feb 24, 2017 at 7:20 PM, Matt Arsenault wrote:
> >>
> >> On Feb 24, 2017, at 01:45, Marek Olšák wrote:
> >>
> >> The main requirement is that if there is indire
ee.
>
Hi Marek,
This patch is: Reviewed-by: Tom Stellard
Setting this fast-math flags on individual floating-point instructions
may have an even bigger impact, because this function attribute is only
used in a few places.
Here is documentation for the fast-math flags:
http://llvm.org/docs/La
On Tue, Jul 28, 2015 at 12:05:40PM +0200, Marek Olšák wrote:
> From: Marek Olšák
>
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers/r600/r600_llvm.c | 152
> -
> .../drivers/radeon/radeon_setup_tgsi_llvm.c| 150 -
On Wed, Jul 29, 2015 at 09:54:05AM +0100, Jose Fonseca wrote:
> On 23/07/15 17:06, Jose Fonseca wrote:
> > On 20/07/15 21:39, Jose Fonseca wrote:
> >> On 20/07/15 18:35, Tom Stellard wrote:
> >>> All LLVM API calls that require an ostream object have been remo
Ref floor = lp_build_intrinsic(builder, "floor",
> emit_data->dst_type,
The intrinsics name should be "llvm.floor.f32" for float and "llvm.floor.f64"
for double.
With that fixed, this is:
Reviewed-by: Tom Stellard
> +
On Fri, Jul 31, 2015 at 04:59:19PM +0200, Marek Olšák wrote:
> On Fri, Jul 31, 2015 at 4:18 PM, Tom Stellard wrote:
> > On Tue, Jul 28, 2015 at 12:05:43PM +0200, Marek Olšák wrote:
> >> From: Marek Olšák
> >>
> >> There are 2 reasons for this:
> >> -
return;
> > +}
>
> The code you're adding here uses spaces for indentation, whereas the
> surrounding code seems to use tabs. With that fixed,
>
> Acked-by: Michel Dänzer
>
Reviewed-by: Tom Stellard
> but Tom should take a look as well.
>
>
> --
On Mon, Aug 03, 2015 at 12:53:32PM +0300, Francisco Jerez wrote:
> EdB writes:
>
> > As a side effect, this fix clRetain/ReleaseEvent Piglit test
> > ---
> > src/gallium/state_trackers/clover/api/transfer.cpp | 29
> > --
> > 1 file changed, 27 insertions(+), 2 deletions(-)
Calls to LLVMIntialize* fail when we are linking against individual
component libraries rather than one large shared object, because
we only include component libraries that are required by the drivers.
We need to make sure to only initialize the targets that we need.
CC: 10.6
---
configure.ac
---
src/gallium/drivers/radeonsi/si_compute.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/gallium/drivers/radeonsi/si_compute.c
b/src/gallium/drivers/radeonsi/si_compute.c
index 5041761..a79c224 100644
--- a/src/gallium/drivers/radeonsi/si_compute.c
+++ b/src/gallium/drivers/radeon
This patch switches non-TGSI compute shaders over to using the HSA
ABI described here:
https://github.com/RadeonOpenCompute/ROCm-Docs/blob/master/AMDGPU-ABI.md
The HSA ABI provides a much cleaner interface for compute shaders and allows
us to share more code in the compiler with the HSA stack.
T
The existing function only worked for integer types.
---
src/gallium/drivers/radeonsi/si_shader.c | 42
1 file changed, 32 insertions(+), 10 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_shader.c
b/src/gallium/drivers/radeonsi/si_shader.c
index 8254cb
---
src/gallium/drivers/radeonsi/si_shader.c | 59 +---
1 file changed, 46 insertions(+), 13 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_shader.c
b/src/gallium/drivers/radeonsi/si_shader.c
index 4e07317..1f1fdf2 100644
--- a/src/gallium/drivers/radeonsi/
---
src/gallium/drivers/radeonsi/si_shader.c | 29 ++---
1 file changed, 18 insertions(+), 11 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_shader.c
b/src/gallium/drivers/radeonsi/si_shader.c
index 49d4121..8254cb2 100644
--- a/src/gallium/drivers/radeonsi/si_
On Tue, Oct 11, 2016 at 03:21:24PM +0200, Nicolai Hähnle wrote:
> On 11.10.2016 07:36, Dave Airlie wrote:
> > On 11 October 2016 at 12:13, Dave Airlie wrote:
> >> On 11 October 2016 at 11:42, Dave Airlie wrote:
> >>> On 11 October 2016 at 05:50, Dave Airlie wrote:
> On 10 October 2016 at 21
---
src/amd/common/ac_nir_to_llvm.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
index e6ff7c8..9c764c7 100644
--- a/src/amd/common/ac_nir_to_llvm.c
+++ b/src/amd/common/ac_nir_to_llvm.c
@@ -2373,8 +2373,8
These were changed in LLVM r284024.
---
src/amd/common/ac_nir_to_llvm.c | 131
1 file changed, 107 insertions(+), 24 deletions(-)
diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
index 9c764c7..4fba7d3 100644
--- a/src/amd/com
---
src/amd/common/ac_nir_to_llvm.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
index e6ff7c8..9c764c7 100644
--- a/src/amd/common/ac_nir_to_llvm.c
+++ b/src/amd/common/ac_nir_to_llvm.c
@@ -2373,8 +2373,8
applies for me. I hope I didn't butcher anything inadvertently.
>
Hi,
I just sent rebased patches. Can you try those.
-Tom
> Cheers,
> Kai
>
>
> Tom Stellard wrote on 13.10.2016 17:21:
> > These were changed in LLVM r2840
These were changed in LLVM r284024.
v2:
- Only use float types for vdata of llvm.amdgcn.image.store. LLVM doesn't
support integer types for this intrinsic.
---
src/amd/common/ac_nir_to_llvm.c | 133
1 file changed, 108 insertions(+), 25 deletions(-)
On Tue, Aug 09, 2016 at 12:36:37PM +0200, Nicolai Hähnle wrote:
> From: Nicolai Hähnle
>
> In the alloca'd array case, no longer create redundant and unused allocas
> for the individual elements; create getelementptrs instead.
Reviewed-by: Tom Stellard
> ---
&
On Tue, Aug 09, 2016 at 12:36:38PM +0200, Nicolai Hähnle wrote:
> From: Nicolai Hähnle
>
> We can use the pointer stored in the temps array directly.
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers/radeon/radeon_setup_tgsi_llvm.c | 5 -
> 1 file changed, 5 dele
On Tue, Aug 09, 2016 at 12:36:39PM +0200, Nicolai Hähnle wrote:
> From: Nicolai Hähnle
>
> We can use the pointer stored in the temps array directly.
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers/radeon/radeon_setup_tgsi_llvm.c | 7 ---
> 1 file changed, 7 dele
On Tue, Aug 09, 2016 at 12:36:40PM +0200, Nicolai Hähnle wrote:
> From: Nicolai Hähnle
>
This is a great idea.
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers/radeon/radeon_setup_tgsi_llvm.c | 14 --
> 1 file changed, 12 insertions(+), 2 deletions(-)
>
On Fri, Aug 12, 2016 at 01:26:08AM +0200, Marek Olšák wrote:
> From: Marek Olšák
>
> we should do that allegedly
Reviewed-by: Tom Stellard
> ---
> src/gallium/drivers/radeon/radeon_setup_tgsi_llvm.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/sr
We were allocating global variables for the maximum LDS size
which made the compiler think we were using all of LDS, which
isn't the case.
---
src/gallium/drivers/radeonsi/si_shader.c | 15 ++-
1 file changed, 6 insertions(+), 9 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/s
On Wed, Nov 02, 2016 at 11:26:08AM +1000, Dave Airlie wrote:
> So it appears at least the LLVM 3.9 backend can get confused
> when it gets hit with if (cond) discard type constructs, and we
> have a GLSL optimisation to convert this to discard_if, so I've
> ported that to NIR, and enabled it for ra
---
Build tested only so far.
src/gallium/auxiliary/draw/draw_llvm.c| 6 +-
src/gallium/auxiliary/gallivm/lp_bld_intr.c | 48 +++-
src/gallium/auxiliary/gallivm/lp_bld_intr.h | 13 -
src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c | 4 +-
src/gallium/d
v2:
Fix adding parameter attributes with LLVM < 4.0.
---
src/gallium/auxiliary/draw/draw_llvm.c| 6 +-
src/gallium/auxiliary/gallivm/lp_bld_intr.c | 52 -
src/gallium/auxiliary/gallivm/lp_bld_intr.h | 13 -
src/gallium/auxiliary/gallivm/lp_bld_sample_
From: Aaron Watry
Applies on top of v3 of Tom's gallivm change.
v2:
- Tom Stellard: Use enums instread of strings.
Signed-off-by: Aaron Watry
CC: Tom Stellard
CC: Jan Vesely
---
src/gallium/drivers/llvmpipe/lp_state_fs.c| 2 +-
src/gallium/drivers/llvmpipe/lp_state_setup.c | 3 +
1 - 100 of 1703 matches
Mail list logo