Re: Moving code around, post classic

On Tue, 2021-12-07 at 08:19 -0500, Alyssa Rosenzweig wrote: > > > > > 2. Move src/compiler/glsl into src/gallium/frontends/mesa as well > > > >     Given that there are now no? drivers that use GLSL-IR directly, > > it > >     might make sense to move the glsl compiler into the mesa > >    

Re: [Mesa-dev] [RFC] Concrete proposal to split classic

On 6/16/21 1:16 PM, Jason Ekstrand wrote: On Tue, Jun 15, 2021 at 8:46 PM Timothy Arceri wrote: On 6/16/21 11:03 AM, Jason Ekstrand wrote: I'm bringing this up via e-mail so it gets a wider audience. Given how will crocus is working at this point, is like to propose we hold off for about

Re: [Mesa-dev] [RFC] Concrete proposal to split classic

On 6/16/21 11:03 AM, Jason Ekstrand wrote: I'm bringing this up via e-mail so it gets a wider audience. Given how will crocus is working at this point, is like to propose we hold off for about three more releases before we drop classic. This next release, 21.2, we'll have crocus as an option

Re: [Mesa-dev] [RFC] Concrete proposal to split classic

On 3/23/21 7:26 PM, Ian Romanick wrote: I would like to wait a couple more releases to do this. I have a couple things that I've been gradually working on for some of the non-i965 classic drivers that I'd like to land before they're put out to pasture. I talked to ajax about this a few weeks

Re: [Mesa-dev] Small problem with my Mesa install (via Re: [ANNOUNCE] mesa 20.2.2)

Looks like it might have been this bug https://gitlab.freedesktop.org/mesa/mesa/-/issues/3990 which should be fixed by https://gitlab.freedesktop.org/mesa/mesa/-/commit/58e43594fc457eaaf1b1e01e48948959a82080bc On 2/11/21 10:40 PM, Ed B wrote: Dear Dylan / Mesa Developer(s), I have just 

Re: [Mesa-dev] remaining 20.2 patches

On 2/12/20 7:05 am, Dylan Baker wrote: 9659384744 ac/nir: fix a typo in ac_are_tessfactors_def_in_all_invocs I'm surprised this doesn't apply cleanly as I don't believe this code has been changed in quite a while. However it is only a single line change so should be pretty easy to resolve

Re: [Mesa-dev] Mesa 20.2.x and GL_RG8_SNORM/GL_NONE

So I found your thread [1] Forcing no error is generally not recommended or a good idea. The game seems to work fine for me. [1] https://forum.warthunder.com/index.php?/topic/500816-unsupported-formattype-gl_rg8_snormgl_none/ On 16/10/20 11:11 am, Daniel Mota Leite wrote: Hi

Re: [Mesa-dev] radv: Handle failing to create .cache dir. BISECTED - This one is broken

This should fix it: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5249 On 28/5/20 1:16 pm, Dieter Nützel wrote: Hello Bas, here is something lost/broken. Part-of: If you have NO ~/.cache/radv_builtin_shaders64 it

Re: [Mesa-dev] Requiring a full author name when contributing to mesa?

PM Timothy Arceri wrote: On 13/12/19 1:54 am, Daniel Stone wrote: On Wed, 11 Dec 2019 at 22:35, Timothy Arceri wrote: So it seems lately we have been increasingly merging patches with made up names, or single names etc [1]. The latest submitted patch has the name Icecream95. This seems wrong

Re: [Mesa-dev] Requiring a full author name when contributing to mesa?

On 13/12/19 1:54 am, Daniel Stone wrote: On Wed, 11 Dec 2019 at 22:35, Timothy Arceri wrote: So it seems lately we have been increasingly merging patches with made up names, or single names etc [1]. The latest submitted patch has the name Icecream95. This seems wrong to me from a point

Re: [Mesa-dev] Requiring a full author name when contributing to mesa?

On 12/12/19 10:38 am, Eric Engestrom wrote: On 2019-12-11 at 23:09, Eric Anholt wrote: On Wed, Dec 11, 2019 at 2:35 PM Timothy Arceri wrote: Hi, So it seems lately we have been increasingly merging patches with made up names, or single names etc [1]. The latest submitted patch has the name

Re: [Mesa-dev] Requiring a full author name when contributing to mesa?

On 12/12/19 10:09 am, Eric Anholt wrote: On Wed, Dec 11, 2019 at 2:35 PM Timothy Arceri wrote: Hi, So it seems lately we have been increasingly merging patches with made up names, or single names etc [1]. The latest submitted patch has the name Icecream95. This seems wrong to me from a point

[Mesa-dev] Requiring a full author name when contributing to mesa?

Hi, So it seems lately we have been increasingly merging patches with made up names, or single names etc [1]. The latest submitted patch has the name Icecream95. This seems wrong to me from a point of keeping up the integrity of the project. I'm not a legal expert but it doesn't seem ideal

Re: [Mesa-dev] Is it time to stop using the mailing list for patch review?

On 11/12/19 8:54 am, Alex Deucher wrote: On Tue, Dec 10, 2019 at 1:13 PM Dylan Baker wrote: Do we have those and does anyone notice? I personally rarely look at the list now unless I'm CC'd on something. That seems really bad for drive by contributors. But frankly I wouldn't submit to a

Re: [Mesa-dev] Is it time to stop using the mailing list for patch review?

On 11/12/19 5:58 am, Zebediah Figura wrote: On 12/10/19 12:21 PM, Matt Turner wrote: On Mon, Dec 9, 2019 at 6:07 PM Dylan Baker wrote: Hi everyone, I think its time we discussed whether we're going to continue to do patch review on the mailing list, or if it it should all go through

Re: [Mesa-dev] Remove classic drivers or fork src/mesa for gallium?

On 4/12/19 3:45 pm, Jason Ekstrand wrote: On Tue, Dec 3, 2019 at 8:19 PM Dave Airlie > wrote: On Wed, 4 Dec 2019 at 10:39, Marek Olšák mailto:mar...@gmail.com>> wrote: > > Hi, > > Here are 2 proposals to simplify and better optimize the

[Mesa-dev] [PATCH] mesa: remove super old TODOs from shaderapi.c

--- src/mesa/main/shaderapi.c | 5 - 1 file changed, 5 deletions(-) diff --git a/src/mesa/main/shaderapi.c b/src/mesa/main/shaderapi.c index 1351f7fd06f..607d30e2bd4 100644 --- a/src/mesa/main/shaderapi.c +++ b/src/mesa/main/shaderapi.c @@ -29,11 +29,6 @@ * * Implementation of

Re: [Mesa-dev] [PATCH] drirc: Add workaround for Divinity: Original Sin Enhanced Edition

I ended up creating a merge request for this and another patch: https://gitlab.freedesktop.org/mesa/mesa/merge_requests/1550 On 2/8/19 10:09 am, Timothy Arceri wrote: This adds an additional work around for the game to fix the blocky shaders as reported in bug 105282 Bugzilla: https

[Mesa-dev] [PATCH] egl: fix OpenGL 3.1 context creation

From the EGL_KHR_create_context spec: "* If OpenGL 3.1 is requested, the context returned may implement any of the following versions: * Version 3.1. The GL_ARB_compatibility extension may or may not be implemented, as determined by the implementation. *

[Mesa-dev] [PATCH] drirc: Add workaround for Divinity: Original Sin Enhanced Edition

This adds an additional work around for the game to fix the blocky shaders as reported in bug 105282 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105282 --- src/util/00-mesa-defaults.conf | 1 + 1 file changed, 1 insertion(+) diff --git a/src/util/00-mesa-defaults.conf

Re: [Mesa-dev] [PATCH] mesa: save/restore SSO flag when using ARB_get_program_binary

Ping! The spec bug was updated and they have agreed to update the spec to define this behavior. I've also sent a piglit test for this: https://patchwork.freedesktop.org/patch/317112/ On 1/7/19 12:25 pm, Timothy Arceri wrote: Without this the restored program will fail the pipeline validation

[Mesa-dev] [PATCH] radv: fix memory leak when restoring from cache

Fixes: 726a31df705b ("radv: Add the concept of radv shader binaries.") --- src/amd/vulkan/radv_pipeline_cache.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/amd/vulkan/radv_pipeline_cache.c b/src/amd/vulkan/radv_pipeline_cache.c index 2b3fda6eb8e..b773de30c32 100644 ---

Re: [Mesa-dev] [PATCH] nir/lower_io_to_temporaries: Fix hash table leak

Reviewed-by: Timothy Arceri On 9/7/19 2:20 am, Connor Abbott wrote: Fixes: c45f5db527252384395e55fb1149b673ec7b5fa8 ("nir/lower_io_to_temporaries: Handle interpolation intrinsics") --- Whoops... src/compiler/nir/nir_lower_io_to_temporaries.c | 2 ++ 1 file changed, 2 insertion

Re: [Mesa-dev] [PATCH] gallium: remove boolean from state tracker APIs

Looks good to me. Reviewed-by: Timothy Arceri On 5/7/19 5:36 am, Ilia Mirkin wrote: Signed-off-by: Ilia Mirkin --- It's actually a bit tricky to compile all the bits involved, so some of this is untested. However if it so happens that I missed a spot, it should be trivial to fix. src

Re: [Mesa-dev] [PATCH] mesa: save/restore SSO flag when using ARB_get_program_binary

On 1/7/19 2:24 pm, Timothy Arceri wrote: FYI since the spec doesn't specifically cover what to do in this case I have filed a spec bug suggesting it be updated to follow this behavior. https://gitlab.khronos.org/opengl/GLSL/issues/46 Whoops filed that one against GLSL rather than the API

Re: [Mesa-dev] [PATCH] mesa: save/restore SSO flag when using ARB_get_program_binary

FYI since the spec doesn't specifically cover what to do in this case I have filed a spec bug suggesting it be updated to follow this behavior. https://gitlab.khronos.org/opengl/GLSL/issues/46 On 1/7/19 12:25 pm, Timothy Arceri wrote: Without this the restored program will fail the pipeline

[Mesa-dev] [PATCH] mesa: save/restore SSO flag when using ARB_get_program_binary

Without this the restored program will fail the pipeline validation checks when we attempt to use an SSO program. Fixes: c20fd744fef1 ("mesa: Add Mesa ARB_get_program_binary helper functions") Cc: Jordan Justen Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=111010 ---

[Mesa-dev] [PATCH] st/glsl: fix silly regression finding gs/tes variants

Fixes: d19fe5e67a39 ("st/glsl: support clamping color outputs in compat for gs/tes") --- src/mesa/state_tracker/st_program.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c index

Re: [Mesa-dev] [PATCH 1/2] nir/loop_analyze: used nir_alu_src to track loop limit

On 20/6/19 5:57 am, Jason Ekstrand wrote: On Wed, Jun 19, 2019 at 3:09 AM Timothy Arceri <mailto:tarc...@itsqueeze.com>> wrote: This helps reduce the amount of abstraction in this pass and allows us to retain more information about the src such as any swizzles.

Re: [Mesa-dev] [PATCH 1/2] nir/loop_analyze: used nir_alu_src to track loop limit

On 20/6/19 5:57 am, Jason Ekstrand wrote: On Wed, Jun 19, 2019 at 3:09 AM Timothy Arceri <mailto:tarc...@itsqueeze.com>> wrote: This helps reduce the amount of abstraction in this pass and allows us to retain more information about the src such as any swizzles.

Re: [Mesa-dev] [PATCH 1/2] nir/loop_analyze: used nir_alu_src to track loop limit

On 19/6/19 11:55 pm, Brian Paul wrote: On 06/19/2019 02:08 AM, Timothy Arceri wrote: This helps reduce the amount of abstraction in this pass and allows us to retain more information about the src such as any swizzles. Retaining the swizzle information is required for a bugfix in the following

[Mesa-dev] [PATCH 2/2] nir/loop_analyze: handle swizzles on the loop limit

Fixes: 6772a17acc8e ("nir: Add a loop analysis pass") --- src/compiler/nir/nir_loop_analyze.c | 8 ++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/src/compiler/nir/nir_loop_analyze.c b/src/compiler/nir/nir_loop_analyze.c index 57d2d94cad2..79ed5bb2712 100644 ---

[Mesa-dev] [PATCH 1/2] nir/loop_analyze: used nir_alu_src to track loop limit

This helps reduce the amount of abstraction in this pass and allows us to retain more information about the src such as any swizzles. Retaining the swizzle information is required for a bugfix in the following patch. Fixes: 6772a17acc8e ("nir: Add a loop analysis pass") ---

[Mesa-dev] [PATCH 2/2] nir/loop_analyze: handle swizzles on invariant

Fixes: 6772a17acc8e ("nir: Add a loop analysis pass") --- src/compiler/nir/nir_loop_analyze.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/src/compiler/nir/nir_loop_analyze.c b/src/compiler/nir/nir_loop_analyze.c index ff73b32c51d..e85a404da1b 100644 ---

[Mesa-dev] [PATCH 1/2] nir/loop_analyze: use nir_src to track the invariant

This helps reduce the amount of abstraction in this pass and allows us to retain more information about the src such as any swizzles. Retaining the swizzle information is required for a bugfix in a following patch. Fixes: 6772a17acc8e ("nir: Add a loop analysis pass") ---

[Mesa-dev] [PATCH] st/glsl: make sure to propagate initialisers to driver storage

This essentially reverts 20234cfe3a20. Fixes piglit test: tests/spec/arb_get_program_binary/execution/uniform-after-restore.shader_test Fixes: 20234cfe3a20 "st/mesa: don't propagate uniforms when restoring from cache" Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=110784 ---

[Mesa-dev] [PATCH] Revert "st/mesa: expose 0 shader binary formats for compat profiles for Qt"

This reverts commit 55376cb31e2f495a4d872b4ffce2135c3365b873. It's been over a year and both QT 5.9.5 and 5.11.0 contained a fix for the original issue. It seems i965 only ever applied this workaround to the 18.0 branch. --- src/mesa/state_tracker/st_context.c| 2 +-

[Mesa-dev] [PATCH] radeonsi: add drirc workaround for American Truck Simulator

Cc: "19.0" "19.1" Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=110711 --- src/util/00-mesa-defaults.conf | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/util/00-mesa-defaults.conf b/src/util/00-mesa-defaults.conf index 3c459f3dec4..e36190f61c8 100644 ---

[Mesa-dev] [PATCH] util: add missing include to build_id.h

Required to use uint8_t --- src/util/build_id.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/util/build_id.h b/src/util/build_id.h index 86d611d8db7..1872ca5c7e5 100644 --- a/src/util/build_id.h +++ b/src/util/build_id.h @@ -26,6 +26,8 @@ #ifdef HAVE_DL_ITERATE_PHDR +#include +

[Mesa-dev] [PATCH] glsl_to_nir: remove unused type_is_int()

This was missed in e00fa99b08b3. Cc: Christian Gmeiner --- src/compiler/glsl/glsl_to_nir.cpp | 9 - 1 file changed, 9 deletions(-) diff --git a/src/compiler/glsl/glsl_to_nir.cpp b/src/compiler/glsl/glsl_to_nir.cpp index 47159ebd5e8..a51d39c4753 100644 ---

Re: [Mesa-dev] [PATCH] radv: call constant folding before opt algebraic

please double check before pushing because of the flrp changes that landed around? No change. On 5/7/19 7:14 AM, Timothy Arceri wrote: ping! On 2/5/19 1:38 pm, Timothy Arceri wrote: The pattern of calling opt algebraic first seems to have originated in i965. The order in OpenGL drivers

Re: [Mesa-dev] [PATCH] Revert "glx: Fix synthetic error generation in __glXSendError"

On 7/5/19 6:27 pm, Michel Dänzer wrote: On 2019-05-07 5:55 a.m., Timothy Arceri wrote: This reverts commit e91ee763c378d03883eb88cf0eadd8aa916f7878. This seems to have broken a number of wine games. Cc: Adam Jackson Cc: Ian Romanick Cc: Hal Gentz Bugzilla: https://bugs.freedesktop.org

Re: [Mesa-dev] [PATCH] radv: call constant folding before opt algebraic

ping! On 2/5/19 1:38 pm, Timothy Arceri wrote: The pattern of calling opt algebraic first seems to have originated in i965. The order in OpenGL drivers generally doesn't matter because the GLSL IR optimisations do constant folding before opt algebraic. However in Vulkan drivers calling opt

[Mesa-dev] [PATCH] Revert "glx: Fix synthetic error generation in __glXSendError"

This reverts commit e91ee763c378d03883eb88cf0eadd8aa916f7878. This seems to have broken a number of wine games. Cc: Adam Jackson Cc: Ian Romanick Cc: Hal Gentz Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=110632 Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=110590 ---

[Mesa-dev] [PATCH] radeonsi: add an AMD_TEX_ANISO environment variable

This brings it inline with the recently added AMD_DEBUG. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109619 --- src/gallium/drivers/radeonsi/si_pipe.c | 4 1 file changed, 4 insertions(+) diff --git a/src/gallium/drivers/radeonsi/si_pipe.c

[Mesa-dev] [PATCH] radeonsi: add config entry for Counter-Strike Global Offensive

This fixes rendering issues with gun scopes which is rather important. Cc: "19.0" "19.1" Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=100239 --- src/util/00-mesa-defaults.conf | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/util/00-mesa-defaults.conf

[Mesa-dev] [PATCH] util/drirc: add workarounds for bugs in Doom 3: BFG

This makes the game playable on radeonsi. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=110143 Cc: "19.0" "19.1" --- src/util/00-mesa-defaults.conf | 5 + 1 file changed, 5 insertions(+) diff --git a/src/util/00-mesa-defaults.conf b/src/util/00-mesa-defaults.conf index

[Mesa-dev] [PATCH] radv: call constant folding before opt algebraic

The pattern of calling opt algebraic first seems to have originated in i965. The order in OpenGL drivers generally doesn't matter because the GLSL IR optimisations do constant folding before opt algebraic. However in Vulkan drivers calling opt algebraic first can result in missed constant folding

Re: [Mesa-dev] [PATCH] glsl: fix and clean up NV_compute_shader_derivatives support

I feel like I've asked this before but can you please try to add commit proper commit messages. Not adding a proper commit message might save you a minute or two but it cost much more than a minute or two for those trying to review a patch or that have bisect to the commit, used git blame, git

Re: [Mesa-dev] [PATCH] nir: fix nir_remove_unused_varyings()

now if it looks okay? Looks good to me. Thanks! Thanks, Dylan Quoting Timothy Arceri (2019-04-24 18:17:42) We were only setting the used mask for the first component of a varying. Since the linking opts split vectors into scalars this has mostly worked ok. However this causes an issue wh

[Mesa-dev] [PATCH 1/2] radeonsi/nir: create si_nir_opts() helper

We will make use of this in the following commit. --- src/gallium/drivers/radeonsi/si_shader.h | 1 + src/gallium/drivers/radeonsi/si_shader_nir.c | 78 +++- 2 files changed, 43 insertions(+), 36 deletions(-) diff --git a/src/gallium/drivers/radeonsi/si_shader.h

[Mesa-dev] [PATCH 2/2] radeonsi/nir: call radeonsi nir opts before the scan pass

Some of the opts are not called in the general optimastion loop in the state trackers glsl -> nir conversion. We need to call the radeonsi specific optimisation once before scanning over the nir otherwise we can end up gathering info on code that is later removed. Fixes an assert in the piglit

[Mesa-dev] [PATCH] nir: fix nir_remove_unused_varyings()

We were only setting the used mask for the first component of a varying. Since the linking opts split vectors into scalars this has mostly worked ok. However this causes an issue where for example if we split a struct on one side of the interface but not the other, then we can possibly end up

Re: [Mesa-dev] [PATCH 1/2] nir: add an option for skipping split_alu_of_phi

On 24/4/19 10:59 am, Jason Ekstrand wrote: On Tue, Apr 23, 2019 at 7:39 PM Timothy Arceri <mailto:tarc...@itsqueeze.com>> wrote: On 24/4/19 1:45 am, Samuel Pitoiset wrote: > > On 4/23/19 5:16 PM, Jason Ekstrand wrote: >> On Tue, Apr 23, 2019 at 7:

Re: [Mesa-dev] [PATCH 1/2] nir: add an option for skipping split_alu_of_phi

On 24/4/19 1:45 am, Samuel Pitoiset wrote: On 4/23/19 5:16 PM, Jason Ekstrand wrote: On Tue, Apr 23, 2019 at 7:46 AM Samuel Pitoiset mailto:samuel.pitoi...@gmail.com>> wrote: On 4/23/19 10:45 AM, Bas Nieuwenhuizen wrote: > On Tue, Apr 23, 2019 at 9:35 AM Samuel Pitoiset >

[Mesa-dev] [PATCH] st/mesa/radeonsi: fix race between destruction of types and shader compilation

Commit 624789e3708c moved the destruction of types out of atexit() and made use of a ref count instead. This is useful for avoiding a crash where drivers such as radeonsi are still compiling in a thread when the app exits and has not called MakeCurrent to change from the current context. While

Re: [Mesa-dev] [PATCH 3/3] ac/nir: use the new raw/struct SSBO atomic intrisics for comp_swap

Series: Reviewed-by: Timothy Arceri On 18/4/19 5:23 pm, Samuel Pitoiset wrote: This is actually fixed now. This change requires LLVM r358579. Make sure to have it in your tree, otherwise the following piglit will hang: tests/spec/arb_shader_storage_buffer_object/execution/ssbo

Re: [Mesa-dev] [PATCH 1/3] Revert "radv: add VK_KHR_shader_atomic_int64 but disable it for now"

On 18/4/19 4:39 pm, Samuel Pitoiset wrote: On 4/18/19 8:38 AM, Samuel Pitoiset wrote: On 4/18/19 8:34 AM, Timothy Arceri wrote: On 18/4/19 4:34 pm, Samuel Pitoiset wrote: On 4/18/19 8:16 AM, Timothy Arceri wrote: Meant to add this was tested on LLVM 9. What do you mean? I meant to say

Re: [Mesa-dev] [PATCH] Revert "compiler/glsl: handle case where we have multiple users for types"

On 18/4/19 4:38 pm, Tapani Pälli wrote: On 4/18/19 8:50 AM, Timothy Arceri wrote: On 18/4/19 3:08 pm, Tapani Pälli wrote: On 4/18/19 8:05 AM, Timothy Arceri wrote: This reverts commit 624789e3708c87ea2a4c8d2266266b489b421cba. It caused 400+ piglit tests to crash on radeonsi, I haven't been

Re: [Mesa-dev] [PATCH 1/3] Revert "radv: add VK_KHR_shader_atomic_int64 but disable it for now"

On 18/4/19 4:34 pm, Samuel Pitoiset wrote: On 4/18/19 8:16 AM, Timothy Arceri wrote: Meant to add this was tested on LLVM 9. What do you mean? I meant to say in the commit messages: This first patch in this series caused a piglit regression with radeonsi NIR on my VEGA64 with LLVM 9

Re: [Mesa-dev] [PATCH 1/3] Revert "radv: add VK_KHR_shader_atomic_int64 but disable it for now"

Meant to add this was tested on LLVM 9. On 18/4/19 4:15 pm, Timothy Arceri wrote: This reverts commit 9cf55b022dfa43f8fe3163edeb87a1c25ebf5a16. This first patch in this series caused a piglit regression with radeonsi NIR on my VEGA64. tests/spec/arb_shader_storage_buffer_object/execution/ssbo

[Mesa-dev] [PATCH 1/3] Revert "radv: add VK_KHR_shader_atomic_int64 but disable it for now"

This reverts commit 9cf55b022dfa43f8fe3163edeb87a1c25ebf5a16. This first patch in this series caused a piglit regression with radeonsi NIR on my VEGA64. tests/spec/arb_shader_storage_buffer_object/execution/ssbo-atomicAdd-int.shader_test --- src/amd/vulkan/radv_device.c | 10 --

[Mesa-dev] [PATCH 3/3] Revert "ac/nir: use new LLVM 8 intrinsics for SSBO atomics except cmpswap"

This reverts commit 78c551aca1c785470e3c0480e33a072b0b5f8928. This patch caused a piglit regression with radeonsi NIR on my VEGA64. tests/spec/arb_shader_storage_buffer_object/execution/ssbo-atomicAdd-int.shader_test --- src/amd/common/ac_nir_to_llvm.c | 31 +-- 1

[Mesa-dev] [PATCH 2/3] Revert "ac/nir: add 64-bit SSBO atomic operations support"

This reverts commit d118e382dd037696ff904e230b11028abf214e80. This first patch in this series caused a piglit regression with radeonsi NIR on my VEGA64. tests/spec/arb_shader_storage_buffer_object/execution/ssbo-atomicAdd-int.shader_test --- src/amd/common/ac_nir_to_llvm.c | 10 +++--- 1

Re: [Mesa-dev] [PATCH] Revert "compiler/glsl: handle case where we have multiple users for types"

On 18/4/19 3:08 pm, Tapani Pälli wrote: On 4/18/19 8:05 AM, Timothy Arceri wrote: This reverts commit 624789e3708c87ea2a4c8d2266266b489b421cba. It caused 400+ piglit tests to crash on radeonsi, I haven't been able to track down the reason yet. Could you share the backtrace? It's more

[Mesa-dev] [PATCH] Revert "compiler/glsl: handle case where we have multiple users for types"

This reverts commit 624789e3708c87ea2a4c8d2266266b489b421cba. It caused 400+ piglit tests to crash on radeonsi, I haven't been able to track down the reason yet. --- src/amd/vulkan/radv_device.c | 3 --- src/compiler/glsl/glsl_parser_extras.cpp | 20 ++-

[Mesa-dev] [PATCH] radeonsi/nir: fix scanning of bindless images

Fixes: d62d434fe920 ("ac/nir_to_llvm: add image bindless support") --- src/gallium/drivers/radeonsi/si_shader_nir.c | 75 ++-- 1 file changed, 37 insertions(+), 38 deletions(-) diff --git a/src/gallium/drivers/radeonsi/si_shader_nir.c

Re: [Mesa-dev] [PATCH] nir: optimize gl_SampleMaskIn to gl_HelperInvocation for radeonsi when possible

On 16/4/19 1:14 pm, Marek Olšák wrote: On Mon, Apr 15, 2019, 10:41 PM Timothy Arceri <mailto:tarc...@itsqueeze.com>> wrote: On 16/4/19 7:03 am, Marek Olšák wrote: > ping > > On Tue, Apr 9, 2019 at 10:03 PM Marek Olšák mailto:mar...@gmail.com&g

Re: [Mesa-dev] [PATCH] nir: optimize gl_SampleMaskIn to gl_HelperInvocation for radeonsi when possible

On 16/4/19 7:03 am, Marek Olšák wrote: ping On Tue, Apr 9, 2019 at 10:03 PM Marek Olšák > wrote: From: Marek Olšák mailto:marek.ol...@amd.com>> ---  src/compiler/nir/nir.h                    |  8 +  src/compiler/nir/nir_opt_intrinsics.c     |

Re: [Mesa-dev] [PATCH] nir: fix packing components with arrays

rrays when there is space. Hopefully that makes sense. On 4/15/19 7:00 AM, Timothy Arceri wrote: When gathering info for unmovable types we need to handle arrays. While we dont support packing/moving arrays we do support packing scalar components with these arrays. Fixes piglit:

[Mesa-dev] [PATCH] nir: fix packing components with arrays

When gathering info for unmovable types we need to handle arrays. While we dont support packing/moving arrays we do support packing scalar components with these arrays. Fixes piglit: tests/spec/arb_enhanced_layouts/execution/component-layout/vs-fs-array-interleave-range.shader_test Fixes:

Re: [Mesa-dev] [PATCH] ac: fix possibly incorrect bindless atomic code in visit_image_atomic

Thanks I didn't see this and sent the same patch. Please add the following to the commit message: Coverity: CID 1444664 Fixes: d62d434fe920 ("ac/nir_to_llvm: add image bindless support") With those: Reviewed-by: Timothy Arceri On 13/4/19 1:40 am, Marek Olšák wrote: From: M

[Mesa-dev] [PATCH] ac/nir_to_llvm: use correct intrinsic type for bindless atomic_{min, max}

Coverity: CID 1444664 Fixes: d62d434fe920 ("ac/nir_to_llvm: add image bindless support") --- 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 61b33c74e6c..8266e7e9f68

Re: [Mesa-dev] Mesa compatibility range

Not really answering your question directly but Desktop chips have had 2.1 support for the past 12+ years. IMO trying to support systems older than that is going to be more trouble than its worth, llvmpipe is unlikely to give you very good performance on a system that old. On 11/4/19 1:37 am,

Re: [Mesa-dev] [PATCH 10/10] ac/nir: remove some useless integer casts for ALU operations

Before I attempt to review, have you run piglit on this series with radeonsi nir? On 11/4/19 1:16 am, Samuel Pitoiset wrote: Sources are always casted to integers. Signed-off-by: Samuel Pitoiset --- src/amd/common/ac_nir_to_llvm.c | 16 1 file changed, 16 deletions(-)

[Mesa-dev] [PATCH] nir: initialise some variables in opt_if_loop_last_continue()

Fixes a couple of Coverity warnings CID 1444626. Fixes: e30804c6024f ("nir/radv: remove restrictions on opt_if_loop_last_continue()") --- src/compiler/nir/nir_opt_if.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/compiler/nir/nir_opt_if.c

Re: [Mesa-dev] [PATCH v2 6/6] glsl/linker: check for xfb_offset aliasing

s expected to fail, but it succeeds: " ... layout (xfb_offset = 0) out vec2 a; layout (xfb_offset = 0) out vec4 b; ... " v2: use a data structure to track the used components instead of a nested loop (Ilia). Cc: Timothy Arceri Cc: Ilia Mirkin Signed-

Re: [Mesa-dev] [PATCH] glsl: allow the #extension directive within code blocks for the dri option

Acked-by: Timothy Arceri On 10/4/19 9:39 am, Marek Olšák wrote: From: Marek Olšák for Viewperf 13 --- src/compiler/glsl/glsl_parser.yy | 9 + 1 file changed, 9 insertions(+) diff --git a/src/compiler/glsl/glsl_parser.yy b/src/compiler/glsl/glsl_parser.yy index 8d6c47fb6a3

Re: [Mesa-dev] [PATCH] glsl: fix shader_storage_blocks_write_access for SSBO block arrays

Reviewed-by: Timothy Arceri On 9/4/19 7:21 am, Marek Olšák wrote: From: Marek Olšák CTS: GL45-CTS.compute_shader.resources-max Fixes: 4e1e8f684bf "glsl: remember which SSBOs are not read-only and pass it to gallium" --- src/compiler/glsl/link_uniforms.cpp | 8 ++-- 1 file

[Mesa-dev] [PATCH] nir/radv: remove restrictions on opt_if_loop_last_continue()

When I implemented opt_if_loop_last_continue() I had restricted this pass from moving other if-statements inside the branch opposite the continue. At the time it was causing a bunch of spilling in shader-db for i965. However Samuel Pitoiset noticed that making this pass more aggressive

Re: [Mesa-dev] [PATCH v2] nir: do not pack varying with different types

Reviewed-by: Timothy Arceri On 5/4/19 8:26 pm, Samuel Pitoiset wrote: The current algorithm only supports packing 32-bit types. If a shader uses both 16-bit and 32-bit varyings, we shouldn't compact them together. v2 - fix regressions spotted by Intel CI Cc: Timothy Arceri Signed-off

Re: [Mesa-dev] [PATCH] nir: do not pack varying with different types

Seems ok for now. Reviewed-by: Timothy Arceri On 3/4/19 10:37 pm, Samuel Pitoiset wrote: The current algorithm only supports packing 32-bit types. If a shader uses both 16-bit and 32-bit varyings, we shouldn't compact them together. Cc: Timothy Arceri Signed-off-by: Samuel Pitoiset

[Mesa-dev] [PATCH v3] nir: propagate known constant values into the if-then branch

Helps Max Waves / VGPR use in a bunch of Unigine Heaven shaders. shader-db results radeonsi (VEGA): Totals from affected shaders: SGPRS: 5505440 -> 5505872 (0.01 %) VGPRS: 3077520 -> 3077296 (-0.01 %) Spilled SGPRs: 39032 -> 39030 (-0.01 %) Spilled VGPRs: 16326 -> 16326 (0.00 %) Private memory

[Mesa-dev] [PATCH] nir: disable opt_for_known_values() for non scalar constants

We don't handle swizzles and non-scalar backends don't split these constants so we need to skip these too. Fixes: 4218b6422cf1 ("nir: propagate known constant values into the if-then branch") https://bugs.freedesktop.org/show_bug.cgi?id=110311 --- src/compiler/nir/nir_opt_if.c | 3 ++- 1 file

[Mesa-dev] [PATCH v2] nir: propagate known constant values into the if-then branch

Helps Max Waves / VGPR use in a bunch of Unigine Heaven shaders. shader-db results radeonsi (VEGA): Totals from affected shaders: SGPRS: 5505440 -> 5505872 (0.01 %) VGPRS: 3077520 -> 3077296 (-0.01 %) Spilled SGPRs: 39032 -> 39030 (-0.01 %) Spilled VGPRs: 16326 -> 16326 (0.00 %) Private memory

Re: [Mesa-dev] [PATCH] ac: fix return type for llvm.amdgcn.frexp.exp.i32.64

Reviewed-by: Timothy Arceri On 29/3/19 6:39 pm, Samuel Pitoiset wrote: This fixes the following piglit with RadeonSI tests/spec/arb_gpu_shader_fp64/execution/built-in-functions/fs-frexp-dvec4.shader_test Signed-off-by: Samuel Pitoiset --- src/amd/common/ac_llvm_build.c | 2 +- 1 file

Re: [Mesa-dev] [PATCH 2/4] ac: add ac_build_frex_exp() helper ans 16-bit/32-bit support

This change broke radeonsi for: tests/spec/arb_gpu_shader_fp64/execution/built-in-functions/fs-frexp-dvec4.shader_test LLVM ERROR: Cannot select: intrinsic %llvm.amdgcn.frexp.exp On 23/3/19 12:52 am, Samuel Pitoiset wrote: Signed-off-by: Samuel Pitoiset --- src/amd/common/ac_llvm_build.c

Re: [Mesa-dev] [PATCH] nir: Lock around validation fail shader dumping

Reviewed-by: Timothy Arceri On 28/3/19 3:22 am, Jason Ekstrand wrote: --- src/compiler/nir/nir_validate.c | 10 ++ 1 file changed, 10 insertions(+) diff --git a/src/compiler/nir/nir_validate.c b/src/compiler/nir/nir_validate.c index ef2e2b62783..f8327c0dd45 100644 --- a/src

Re: [Mesa-dev] [PATCH] st/glsl_to_nir: fix incorrect arrary access

On 26/3/19 3:43 am, Emil Velikov wrote: Hi Timothy, On Fri, 1 Mar 2019 at 10:36, Timothy Arceri wrote: This fixes a segfault when we try to access the array using a -1 when the array wasn't allocated in the first place. Before 7536af670b75 we would just access a pre-allocated array

Re: [Mesa-dev] [PATCH v4 1/6] glsl/linker: location aliasing requires types to have the same width

On 21/3/19 2:07 am, Andres Gomez wrote: On Wed, 2019-03-20 at 16:46 +0200, Andres Gomez wrote: On Thu, 2019-03-21 at 00:20 +1100, Timothy Arceri wrote: On 20/3/19 9:31 pm, Andres Gomez wrote: On Wed, 2019-03-20 at 20:46 +1100, Timothy Arceri wrote: On 2/2/19 5:05 am, Andres Gomez wrote

Re: [Mesa-dev] [PATCH v4 1/6] glsl/linker: location aliasing requires types to have the same width

On 21/3/19 1:38 am, Andres Gomez wrote: On Wed, 2019-03-20 at 20:46 +1100, Timothy Arceri wrote: On 2/2/19 5:05 am, Andres Gomez wrote: From: Iago Toral Quiroga Regarding location aliasing requirements, the OpenGL spec says: "Further, when location aliasing, the aliases sh

Re: [Mesa-dev] [PATCH v4 1/6] glsl/linker: location aliasing requires types to have the same width

On 20/3/19 9:31 pm, Andres Gomez wrote: On Wed, 2019-03-20 at 20:46 +1100, Timothy Arceri wrote: On 2/2/19 5:05 am, Andres Gomez wrote: From: Iago Toral Quiroga Regarding location aliasing requirements, the OpenGL spec says: "Further, when location aliasing, the aliases sh

Re: [Mesa-dev] [PATCH 1/1] nir: add a pass that removes continue blocks

On 20/3/19 9:41 pm, Samuel Pitoiset wrote: 28717 shaders in 14931 tests Totals: SGPRS: 1267317 -> 1267549 (0.02 %) VGPRS: 896876 -> 895920 (-0.11 %) Spilled SGPRs: 24701 -> 26367 (6.74 %) Code Size: 48379452 -> 48507880 (0.27 %) bytes Max Waves: 241159 -> 241190 (0.01 %) Totals from affected

Re: [Mesa-dev] [PATCH 1/1] nir: add a pass that removes continue blocks

On 20/3/19 9:41 pm, Samuel Pitoiset wrote: 28717 shaders in 14931 tests Totals: SGPRS: 1267317 -> 1267549 (0.02 %) VGPRS: 896876 -> 895920 (-0.11 %) Spilled SGPRs: 24701 -> 26367 (6.74 %) Code Size: 48379452 -> 48507880 (0.27 %) bytes Max Waves: 241159 -> 241190 (0.01 %) Totals from affected

Re: [Mesa-dev] [PATCH v4 1/6] glsl/linker: location aliasing requires types to have the same width

On 2/2/19 5:05 am, Andres Gomez wrote: From: Iago Toral Quiroga Regarding location aliasing requirements, the OpenGL spec says: "Further, when location aliasing, the aliases sharing the location must have the same underlying numerical type (floating-point or integer)." Khronos

Re: [Mesa-dev] [PATCH] nir: only override previous alu during loop analysis if supported

Piglit test to trigger the crash: https://patchwork.freedesktop.org/patch/292855/ On 19/3/19 12:09 pm, Timothy Arceri wrote: Users of this function expect alu to be a supported comparision if the induction variable is not NULL. Since we attempt to override the return values if the first limit

[Mesa-dev] [PATCH] nir: only override previous alu during loop analysis if supported

Users of this function expect alu to be a supported comparision if the induction variable is not NULL. Since we attempt to override the return values if the first limit is not a const, we must make sure we are dealing with a valid comparision before overriding the alu instruction. Fixes an

[Mesa-dev] [PATCH] ac/nir_to_llvm: make sure to we are comparing ints in a couple of places

Fixes llvm validation error in a Deus Ex shader on radeonsi. Both operands to ICmp instruction are not of the same type! --- src/amd/common/ac_nir_to_llvm.c | 7 ++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c

[Mesa-dev] [PATCH] ac/nir_to_llvm: add assert to emit_bcsel()

nir to llvm assumes we have already split vectors to scalars via nir_lower_alu_to_scalar(). --- src/amd/common/ac_nir_to_llvm.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c index 18297ed99b1..0ca3f83a248 100644 ---

Re: [Mesa-dev] [PATCH] radv: remove sisched hack for talos

is back from vacation. Running all the time with AMD_DEBUB (R600_DEBUG)=nir,sisched 'cause it's worth it... Can you point to specific games that are helped? Dieter On 3/15/19 11:25 AM, Timothy Arceri wrote: This was added in 8a7d4092d260 but no longer seems to have any impact on performance

  1   2   3   4   5   6   7   8   9   10   >