On 06/12/2013 05:42 PM, Tom Stellard wrote:
Hi,
The attached patches add support for local address space on
Evergreen / Northern Islands GPUs.
Please Review.
-Tom
> + def int_AMDGPU_barrier_local : Intrinsic<[], [], []>;
You probably want to mark this as IntrReadMem to try to avoid reorderi
I spent today trying to get GLSL 1.50 GSes up and running, so I can port
the basic UBO VS/FS tests to 1.50 GSes. I made significant progress on
parsing and error checking, and the results are in the "gs" branch of my
tree (recently rebased on Paul's). It's stuck currently at:
"error: size of arr
From: Tom Stellard
And allocate the correct amount before dispatching the kernel.
---
src/gallium/drivers/r600/evergreen_compute.c | 53 +++---
.../drivers/r600/evergreen_compute_internal.h | 1 +
src/gallium/drivers/r600/evergreen_state.c | 6 +--
src/galliu
From: Tom Stellard
---
src/gallium/drivers/r600/evergreen_compute.c | 23 +++-
src/gallium/drivers/r600/r600_shader.c | 32
2 files changed, 22 insertions(+), 33 deletions(-)
diff --git a/src/gallium/drivers/r600/evergreen_compute.c
b/src/gall
From: Tom Stellard
---
r600/lib/SOURCES | 2 ++
r600/lib/synchronization/barrier.cl | 15 +++
r600/lib/synchronization/barrier_impl.ll | 12
3 files changed, 29 insertions(+)
create mode 100644 r600/lib/synchronization/barrier.cl
create mo
On Thu, Jun 13, 2013 at 3:33 AM, Eric Anholt wrote:
> Frank Henigman writes:
>
>> On Tue, Jun 11, 2013 at 1:10 PM, Eric Anholt wrote:
>>
>>> Frank Henigman writes:
>>>
>>> > Replace the one texture lock with a lock per texture. This allows
>>> > uploading textures from one thread concurrently
On Wed, Jun 12, 2013 at 11:41 AM, Jakob Bornecrantz
wrote:
> The changes looks good at a first glance.
>
> There are some thoughts tho,
> maybe we should try and make libxatracker.so autodetect
> which driver to use so we don't have to create a libxatracker.so
> for each driver, at least for linux
https://bugs.freedesktop.org/show_bug.cgi?id=61919
Darxus changed:
What|Removed |Added
Status|NEW |RESOLVED
Resolution|---
- Original Message -
[...]
> On an unrelated note ... packing clip distances into vec4s sucks, I want
> scalar arrays.
OMG, then we must fix it!
Now seriously, we had the discussion about packed vs unpacked clip distances --
I don't feel strongly either way and I don't recall the ration
On 12.06.2013 15:57, Jose Fonseca wrote:
>
>
> - Original Message -
>> Am 11.06.2013 05:39, schrieb Zack Rusin:
>>> There are strict limits on those registers. Define the maximums
>>> and use them instead of magic numbers. Also allows us to add
>>> some extra sanity checks.
>>> Suggested
On 10 June 2013 16:51, Eric Anholt wrote:
> Fabian Bieler writes:
>
> > Hello!
> >
> > I'm obviously no expert for this code but as far as I understand it
> > this was supposed to resize the zero length built-in arrays when they
> > are not redeclared by the user (and only indexed with constant
Frank Henigman writes:
> On Tue, Jun 11, 2013 at 1:10 PM, Eric Anholt wrote:
>
>> Frank Henigman writes:
>>
>> > Replace the one texture lock with a lock per texture. This allows
>> > uploading textures from one thread concurrently with drawing in another
>> > thread. _mesa_lock_context_textu
On 06/12/2013 01:50 AM, Chris Forbes wrote:
I've just completed a piglit run of kwg/g45-math16 picked to master on
my ironlake. No regressions vs master.
Helps cs:s video stress test by about 2%.
-- Chris
That's a surprising amount!
As is, the patch doesn't actually enable 16-wide math on G4
https://bugs.freedesktop.org/show_bug.cgi?id=64386
Emil Velikov changed:
What|Removed |Added
Attachment #79044|text/plain |image/png
mime type|
The changes looks good at a first glance.
There are some thoughts tho,
maybe we should try and make libxatracker.so autodetect
which driver to use so we don't have to create a libxatracker.so
for each driver, at least for linux that should be do-able.
Thoughts?
Cheers, Jakob.
Bind the transform feedback buffer before drawing into it und unbind it
afterwards.
---
src/glsl/gsraytrace.cpp | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/glsl/gsraytrace.cpp b/src/glsl/gsraytrace.cpp
index 62a584d..f34a3da 100644
--- a/src/glsl/gsraytrace.cpp
+++ b/
NV_transform_feedback is currently not supported by mesa (and I'm
somewhat doubtful it will be implemented in the foreseeable future).
Note that with this patch the demo doesn't work anymore with the nvidia
binary driver.
---
src/glsl/gsraytrace.cpp | 71 +-
On Die, 2013-06-11 at 16:26 +0100, Richard Sandiford wrote:
> Signed-off-by: Richard Sandiford
> ---
> src/gallium/auxiliary/util/u_format.h | 42
> ++-
> 1 file changed, 41 insertions(+), 1 deletion(-)
>
> diff --git a/src/gallium/auxiliary/util/u_format.h
> b/
On 06/07/2013 05:20 PM, Stéphane Marchesin wrote:
Ping, does anyone else want to review this patch?
Stéphane
On Fri, May 10, 2013 at 3:56 PM, Manfred Ernst wrote:
Problem: The IEEE float optimized version of UNCLAMPED_FLOAT_TO_UBYTE in
macros.h
computed incorrect results for inputs in the r
On 06/10/2013 11:31 AM, Rob Clark wrote:
From: Rob Clark
The newer snapdragon devices (at least, the one I have in nexus4) no
longer has a dedicated 2D core. For these devices, DDX needs to use
the 3D core. I've implemented support to use XA in the DDX, but it
requires some changes which brea
With the big-endian changes, there can be two swizzle orders for each format.
This patch turns Format.inv_swizzle() into a global function that takes the
swizzle list as a parameter.
There is no change to the generated u_format_table.c.
Signed-off-by: Richard Sandiford
---
src/gallium/auxiliary
Signed-off-by: Richard Sandiford
---
src/gallium/auxiliary/util/u_format.h | 42 ++-
1 file changed, 41 insertions(+), 1 deletion(-)
diff --git a/src/gallium/auxiliary/util/u_format.h
b/src/gallium/auxiliary/util/u_format.h
index e4b9c36..db6c290 100644
--- a/src
The new field eats up all the remaining bits of the bitfield,
but it could easily be shrunk in future.
Signed-off-by: Richard Sandiford
---
src/gallium/auxiliary/util/u_format.h| 1 +
src/gallium/auxiliary/util/u_format_table.py | 4 ++--
2 files changed, 3 insertions(+), 2 deletions(-)
The old code tried to emulate little-endian format layout on big-endian
targets by byte-swapping each pixel value. This was OK for packed byte
formats, but isn't what we want for multibyte channels. E.g. the floats
in R32G32B32A32_FLOAT should all have native endianness, and the red
component sho
Store the shift value for a bitmask channel in the Channel object,
rather than working it out on the fly. This allows the shift to
depend on endianness.
There is no change to the generated u_format_table.c.
Signed-off-by: Richard Sandiford
---
src/gallium/auxiliary/util/u_format_pack.py | 8 +
Splits out the code that parses the channel list, so that we
can have different lists for little and big endian.
There is no change to the generated u_format_table.c.
Signed-off-by: Richard Sandiford
---
src/gallium/auxiliary/util/u_format_parse.py | 96 +++-
1 file chan
Rather than iterate over format.channels and format.swizzles directly,
use Python subfunctions that take the channel and swizzle lists as
arguments. This allow the channel and swizzle lists to depend on
endianness.
There is no change to the generated u_format_table.c.
Signed-off-by: Richard Sand
The main aim is to reduce the number of places that access channels[0],
swizzles[0] and swizzles[1] directly.
There is no change to the generated u_format_table.c.
Signed-off-by: Richard Sandiford
---
src/gallium/auxiliary/util/u_format_pack.py | 27 ++-
src/gallium/auxiliary/util/
On 06/12/2013 03:41 AM, kristof.ralov...@gmail.com wrote:
On Mon, Apr 29, 2013 at 4:21 PM, Brian Paul wrote:
On 04/28/2013 06:54 AM, kristof.ralov...@gmail.com wrote:
Dear All,
please review and merge this demo.
In patch 2:
* Let's pass the GS input/output info as parameters to the funct
This series is a replacement for the util part of:
http://lists.freedesktop.org/archives/mesa-dev/2013-May/039419.html
It doesn't include any of the controversial format name parts of
those patches (although of course they're still needed in some form).
The original version took the endianne
- Original Message -
> Am 11.06.2013 05:39, schrieb Zack Rusin:
> > There are strict limits on those registers. Define the maximums
> > and use them instead of magic numbers. Also allows us to add
> > some extra sanity checks.
> > Suggested by Brian.
> >
> > Signed-off-by: Zack Rusin
>
https://bugs.freedesktop.org/show_bug.cgi?id=65676
Priority: medium
Bug ID: 65676
Assignee: mesa-dev@lists.freedesktop.org
Summary: Piglit: port remaining glean tests to piglit framework
Severity: minor
Classification: Unclassified
https://bugs.freedesktop.org/show_bug.cgi?id=64386
Tollef Fog Heen changed:
What|Removed |Added
Assignee|sitewranglers@lists.freedes |mesa-dev@lists.freedesktop.
I've just completed a piglit run of kwg/g45-math16 picked to master on my
ironlake. No regressions vs master.
Helps cs:s video stress test by about 2%.
-- Chris
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailma
https://bugs.freedesktop.org/show_bug.cgi?id=61919
Quentin Glidic changed:
What|Removed |Added
CC||sardemff7+freedesktop@sarde
https://bugs.freedesktop.org/show_bug.cgi?id=61919
--- Comment #8 from Quentin Glidic ---
Created attachment 80713
--> https://bugs.freedesktop.org/attachment.cgi?id=80713&action=edit
Patch to fix CFLAGS
I didn’t test it as I don’t get this failure myself, but from my Autotools
reading, this s
36 matches
Mail list logo
