On 25/9/18 10:35 am, Marek Olšák wrote:
Do you know what's broken in LLVM? Or is it a problem with the ISA?
I haven't actually dug any further. Adding Samuel to see if he remembers
more.
However according to the original bug report this is a regression when
going from LLVM 6 to LLVM 7. I
https://bugs.freedesktop.org/show_bug.cgi?id=106833
--- Comment #9 from Tapani Pälli ---
(In reply to Mark Janes from comment #8)
> Tapani, is there a piglit test for this?
Nope but I can make a test, will send also the fix to mesa-dev unless I come up
with something better.
--
You are
For the remaining patches (3-6), it would be better to precompute the
register values in struct si_shader, so that the emit functions are as
cheap as possible.
Marek
On Mon, Sep 24, 2018 at 10:51 PM, Marek Olšák wrote:
> On Tue, Sep 18, 2018 at 4:21 PM, Sonny Jiang wrote:
>> Signed-off-by:
On Tue, Sep 18, 2018 at 4:21 PM, Sonny Jiang wrote:
> Signed-off-by: Sonny Jiang
> ---
> src/gallium/drivers/radeonsi/si_gfx_cs.c | 15 ++-
> src/gallium/drivers/radeonsi/si_state.h | 19 ++-
> .../drivers/radeonsi/si_state_shaders.c | 112 --
> 3 files
On Tue, Sep 18, 2018 at 4:21 PM, Sonny Jiang wrote:
> Signed-off-by: Sonny Jiang
> ---
> src/gallium/drivers/radeonsi/si_gfx_cs.c | 1 +
> src/gallium/drivers/radeonsi/si_pm4.c | 3 +++
> src/gallium/drivers/radeonsi/si_pm4.h | 11 ++
>
Reviewed-by: Marek Olšák
Marek
On Thu, Sep 20, 2018 at 4:27 AM, Erik Faye-Lund
wrote:
> In st_renderbuffer_alloc_storage, we avoid allocating storage for
> zero-sized buffers, leading to this pointer being NULL. We already
> take care to avoid dereferencing these pointers for color-buffers,
>
Do you know what's broken in LLVM? Or is it a problem with the ISA?
Marek
On Fri, Sep 21, 2018 at 10:38 PM, Timothy Arceri wrote:
> This ports the fix from 3d41757788ac. Both LLVM 7 & 8 continue
> to have this problem.
>
> It fixes rendering issues in some menu and loading screens of
> Civ VI
Looks good to me.
Marek
On Mon, Sep 24, 2018 at 2:29 AM, Axel Davy wrote:
> Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=105333
> Fixes: https://github.com/iXit/Mesa-3D/issues/314
>
> For this application, NaN is passed to KILL_IF and is expected to
> pass.
>
> v2: Explain in the code
This patch also handles all types, just differently. If you change the
typedefs in the header, you'll get a different type and the code is
exactly the same for all types, but that's not important (to me
anyway).
It also supports signed division. (not important to me either, but may
be important
Assuming testing comes back ok.
Reviewed-by: Timothy Arceri
On 25/9/18 1:19 am, Emil Velikov wrote:
This reverts commit 3d81e11b49366b5636b8524ba0f8c7076e3fdf34.
As reported by Federico, some games require the 'sort by year' since
they truncate the extensions which do not fit the fixed size
On 25/9/18 4:18 am, Nicholas Kazlauskas wrote:
Some programs start with the path and command line arguments in
argv[0] (program_invocation_name). Chromium is an example of
an application using mesa that does this.
This tries to query the real path for the symbolic link /proc/self/exe
to find
Did you copy the code from the same author?
Does your version also have an interface for dividing by a uniform
instead of a compile time constant?
Note that this algorithm was originally only written for
non-power-of-two divisors and I extended it to support 1 and
power-of-two divisors in order
Quoting Chuck Atkins (2018-09-24 13:24:29)
> Hi Dylan,
>
>
> I think you could simplify this:
>
> > + if dep_llvm.found() and with_gallium_swr and dep_llvm.version
> ().version_compare('>= 7')
> > + _llvm_2pass = true
> > + llvm_modules += ['ipo', 'objcarcopts']
>
Hi Dylan,
I think you could simplify this:
>
> > + if dep_llvm.found() and with_gallium_swr and
> dep_llvm.version().version_compare('>= 7')
> > +_llvm_2pass = true
> > +llvm_modules += ['ipo', 'objcarcopts']
> > + endif
> > + if _llvm_2pass
> > +dep_llvm = dependency(
> > +
Quoting Chuck Atkins (2018-09-24 12:58:26)
> Signed-off-by: Chuck Atkins
> CC:
> CC: Dylan Baker
> CC: Bruce Cherniak
> CC: Tim Rowley
> ---
> meson.build | 37 -
> 1 file changed, 24 insertions(+), 13 deletions(-)
>
> diff --git a/meson.build
v2: Fix typo in autoconf
Signed-off-by: Chuck Atkins
CC:
CC: Bruce Cherniak
CC: Tim Rowley
---
configure.ac| 5 +
src/gallium/drivers/swr/rasterizer/jitter/blend_jit.cpp | 4
Signed-off-by: Chuck Atkins
CC:
CC: Dylan Baker
CC: Bruce Cherniak
CC: Tim Rowley
---
meson.build | 37 -
1 file changed, 24 insertions(+), 13 deletions(-)
diff --git a/meson.build b/meson.build
index cbf88b5013..a7e03c29dc 100644
--- a/meson.build
+++
On Thu, Sep 20, 2018 at 11:01 PM Topi Pohjolainen
wrote:
>
> Fixes gpu hangs with Carchase and Manhattan.
>
> Cc: Anuj Phogat
> Signed-off-by: Topi Pohjolainen
> ---
> src/intel/compiler/brw_fs_visitor.cpp | 16 +---
> 1 file changed, 13 insertions(+), 3 deletions(-)
>
> diff --git
You should try building with MSVC. ☹
Quoting Jason Ekstrand (2018-09-21 11:47:32)
> You should try building with clang.
>
> On September 21, 2018 15:51:31 Eric Engestrom
> wrote:
>
> > After this, I only see one last warning [1] in all of Mesa \o/
> > (building with GCC8, YMMV of course,
Quoting Eric Engestrom (2018-09-21 06:50:40)
> Signed-off-by: Eric Engestrom
> ---
> meson.build | 9 ++---
> 1 file changed, 6 insertions(+), 3 deletions(-)
>
> diff --git a/meson.build b/meson.build
> index 7b7bf4d8e609c92bbd06..863962186caa6afe8ae4 100644
> --- a/meson.build
> +++
Quoting Eric Engestrom (2018-09-24 09:44:37)
> Signed-off-by: Eric Engestrom
> ---
> v2: Use the util_* wrapper to make sure MSVC stays happy (Dylan)
> ---
> src/gallium/auxiliary/gallivm/lp_bld_printf.c | 4 +---
> 1 file changed, 1 insertion(+), 3 deletions(-)
>
> diff --git
Quoting Chuck Atkins (2018-09-24 10:34:22)
> Hi Dylan (and others?)
>
> SWR has some checks to try to determine which compiler option is needed to
> enable a given instruction set. The way this is implemented in Meson seems to
> be incorrect currently. For example, the attempt to detect the
On Fri, Sep 21, 2018 at 08:23:04PM +0200, Jason Ekstrand wrote:
> Rb. This also fixes simulation errors on gen9; might be worth mentioning that.
>
Thanks for the review! It turns out that I had an old build of the gen11
simulator. It was later updated to remove this error. As such, I'm
guessing
The DDX driver can be notified of adaptive sync suitability by
flagging the application's window with the _VARIABLE_REFRESH property.
This property is set on the first swap the application performs
when adaptive_sync_enable is set to true in the drirc.
It's performed here instead of when the
It's better to let most applications make use of adaptive sync
by default. Problematic applications can be placed on the blacklist
or the user can manually disable the feature.
Signed-off-by: Nicholas Kazlauskas
---
src/gallium/drivers/radeonsi/driinfo_radeonsi.h | 4
1 file changed, 4
Applications that don't present at a predictable rate (ie. not games)
shouldn't have adaptive sync enabled. This list covers some of the
common desktop compositors, web browsers and video players.
Signed-off-by: Nicholas Kazlauskas
---
src/util/00-mesa-defaults.conf | 79
Some programs start with the path and command line arguments in
argv[0] (program_invocation_name). Chromium is an example of
an application using mesa that does this.
This tries to query the real path for the symbolic link /proc/self/exe
to find the program name instead.
Signed-off-by: Nicholas
These patches are part of a proposed new interface for supporting variable
refresh rate via DRM properties.
It adds a new option for supporting adaptive sync to drirc along
with the implementation of notifying the window manager/DDX of
the support via a window property.
The option is enabled by
Hi Dylan (and others?)
SWR has some checks to try to determine which compiler option is needed to
enable a given instruction set. The way this is implemented in Meson seems
to be incorrect currently. For example, the attempt to detect the correct
compiler option to enable the AVX512 KNL
On Monday, 2018-09-24 09:34:34 -0700, Dylan Baker wrote:
> by adding `_llvm == 'true'` to the required argument we can check the
> 'auto' and 'true' case in one path.
Reviewed-by: Eric Engestrom
> ---
> meson.build | 23 ---
> 1 file changed, 8 insertions(+), 15
Signed-off-by: Eric Engestrom
---
v2: Use the util_* wrapper to make sure MSVC stays happy (Dylan)
---
src/gallium/auxiliary/gallivm/lp_bld_printf.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_printf.c
by adding `_llvm == 'true'` to the required argument we can check the
'auto' and 'true' case in one path.
---
meson.build | 23 ---
1 file changed, 8 insertions(+), 15 deletions(-)
diff --git a/meson.build b/meson.build
index 7a5bcf9004b..d098f96d949 100644
--- a/meson.build
Quoting Chuck Atkins (2018-09-24 09:01:37)
> Hi Dylan,
>
>
> > xswr)
> > llvm_require_version $LLVM_REQUIRED_SWR "swr"
> > + llvm_add_default_components "swr"
> > + if test $LLVM_VERSION_MAJOR -ge 7; then
> > +
Hi List,
Due to me just forgetting to send out the release on Friday :/, it's one work
day late. Mesa 18.1.9 is now available for general consumption, it is the last
release in the 18.1.x series, consider upgrading to 18.2.x for further updates.
This has been another busy cycle, with roughly 35
+Cc Keith, Jason & Daniel, who know this code best.
On Monday, 2018-09-24 08:46:22 +1000, Stuart Young wrote:
> From: Maxime
>
> Since the Randr lease code was added, compiling against libxcb 1.12 no
> longer works.
>
> CC: mesa-sta...@lists.freedesktop.org
> Bugzilla:
Hi Dylan,
> xswr)
> > llvm_require_version $LLVM_REQUIRED_SWR "swr"
> > +llvm_add_default_components "swr"
> > +if test $LLVM_VERSION_MAJOR -ge 7; then
> > +llvm_add_component "ipo" "swr"
> > +llvm_add_component
On 24 September 2018 at 01:55, Marek Olšák wrote:
> On Fri, Sep 21, 2018 at 11:34 AM, Emil Velikov
> wrote:
>> On 21 September 2018 at 00:42, Timothy Arceri wrote:
>>> On 20/9/18 11:09 pm, Ian Romanick wrote:
On 09/19/2018 11:36 PM, Federico Dossena wrote:
>
> As most of you
This reverts commit 3d81e11b49366b5636b8524ba0f8c7076e3fdf34.
As reported by Federico, some games require the 'sort by year' since
they truncate the extensions which do not fit the fixed size string
array.
Seemingly I did not consider that as the documentation (both Mesa and
Nvidia) mention
https://bugs.freedesktop.org/show_bug.cgi?id=102204
--- Comment #7 from mirh ---
Could the comments here be of any help?
https://github.com/acomminos/wine-pba/blob/master/patches/0009-wined3d-Add-quirk-to-use-GL_CLIENT_STORAGE_BIT-for-m.patch
I.e. did anybody try to check for
Very similar And mine handles 8, 16, and 64-bit types. :-D
--Jason
On Mon, Sep 24, 2018 at 8:53 AM Ian Romanick wrote:
> I didn't look really closely at either set, but this seems really
> similar to something Jason sent out a week or two. Perhaps you guys
> could unify these?
>
> On
On 9/24/18 4:38 PM, Ian Romanick wrote:
On 09/24/2018 06:23 AM, Tapani Pälli wrote:
Cc: mesa-sta...@lists.freedesktop.org
Signed-off-by: Tapani Pälli
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108012
---
src/compiler/glsl/ast_to_hir.cpp | 5 +
1 file changed, 5
I didn't look really closely at either set, but this seems really
similar to something Jason sent out a week or two. Perhaps you guys
could unify these?
On 09/23/2018 09:57 AM, Marek Olšák wrote:
> From: Marek Olšák
>
> Compilers can use this to generate optimal code for integer division
> by
On 09/24/2018 06:23 AM, Tapani Pälli wrote:
> Cc: mesa-sta...@lists.freedesktop.org
> Signed-off-by: Tapani Pälli
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108012
> ---
> src/compiler/glsl/ast_to_hir.cpp | 5 +
> 1 file changed, 5 insertions(+)
>
> diff --git
https://bugs.freedesktop.org/show_bug.cgi?id=108012
Tapani Pälli changed:
What|Removed |Added
Status|NEW |ASSIGNED
--
You are receiving this
Cc: mesa-sta...@lists.freedesktop.org
Signed-off-by: Tapani Pälli
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108012
---
src/compiler/glsl/ast_to_hir.cpp | 5 +
1 file changed, 5 insertions(+)
diff --git a/src/compiler/glsl/ast_to_hir.cpp b/src/compiler/glsl/ast_to_hir.cpp
index
From: Scott D Phillips
Rename the (un)map_gtt functions to (un)map_map (map by
returning a map) and add new functions (un)map_tiled_memcpy that
return a shadow buffer populated with the intel_tiled_memcpy
functions.
Tiling/detiling with the cpu will be the only way to handle Yf/Ys
tiling, when
From: Scott D Phillips
The reference for MOVNTDQA says:
For WC memory type, the nontemporal hint may be implemented by
loading a temporary internal buffer with the equivalent of an
aligned cache line without filling this data to the cache.
[...] Subsequent MOVNTDQA reads to
Hi;
Here's another try to inline sse41 code and get rid of gtt maps
in intel_miptree_map (revert 58fb613a519). To be able to safely
utilize sse41 we separate sse41 functionality as a library and
then choose run time if we want to use it.
Couple of different approaches were tried, this one
There is currently no use of returned memcpy functions outside
intel_tiled_memcpy. Patch changes intel_get_memcpy to return memcpy
type instead of actual function. This makes it easier later to separate
streaming load copy in to own static library.
Signed-off-by: Tapani Pälli
---
Reviewed-by: Bas Nieuwenhuizen
On Mon, Sep 24, 2018 at 10:19 AM Timothy Arceri wrote:
>
> This allows NIR to CSE more operations. LLVM does this also so the
> impact is limited, however doing this in NIR allows other opts to
> make progress. For example in radeonsi more loops are unrolled in
>
On Sun, 2018-09-23 at 17:01 +0200, Jason Ekstrand wrote:
>
> On September 23, 2018 16:17:36 Emil Velikov wrote:
>
> > Hi Jason,
> >
> > On 20 September 2018 at 11:33, Jason Ekstrand wrote:
> > > This was added as part of 1.1 but it's very hard to track exactly what
> > > extension added it.
This allows NIR to CSE more operations. LLVM does this also so the
impact is limited, however doing this in NIR allows other opts to
make progress. For example in radeonsi more loops are unrolled in
Civilization Beyond Earth.
The actual pipeline-db stats are not overwhelming but even in the
Do you know of other applications that are affected by the order or
extensions? This is the first time I encounter this problem.
On 2018-09-21 13:56, Ian Romanick wrote:
On 09/20/2018 04:42 PM, Timothy Arceri wrote:
On 20/9/18 11:09 pm, Ian Romanick wrote:
On 09/19/2018 11:36 PM, Federico
Any updates on this? I see the code is still in master
On 2018-09-21 17:34, Emil Velikov wrote:
On 21 September 2018 at 00:42, Timothy Arceri wrote:
On 20/9/18 11:09 pm, Ian Romanick wrote:
On 09/19/2018 11:36 PM, Federico Dossena wrote:
As most of you are probably aware of, id2 and id3
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=105333
Fixes: https://github.com/iXit/Mesa-3D/issues/314
For this application, NaN is passed to KILL_IF and is expected to
pass.
v2: Explain in the code why UGE is used.
Signed-off-by: Axel Davy
Reviewed-by: Marek Olšák
CC:
---
On 24/09/18 01:55, Marek Olšák wrote:
On Fri, Sep 21, 2018 at 11:34 AM, Emil Velikov wrote:
On 21 September 2018 at 00:42, Timothy Arceri wrote:
On 20/9/18 11:09 pm, Ian Romanick wrote:
On 09/19/2018 11:36 PM, Federico Dossena wrote:
As most of you are probably aware of, id2 and id3
56 matches
Mail list logo
