- 27 Nov, 2020 37 commits
-
-
Connor Abbott authored
nir_lower_io_to_temporaries should generate better code. Part-of: <mesa/mesa!7274>
-
Connor Abbott authored
Do all the necessary lowering in one place, during finalization, and stop uselessly calling nir_lower_indirect_derefs in turnip. Splitting i/o to elements should no longer be necessary since we use the i/o semantics instead of variables now. This has the side effect that we no longer generate enormous if-ladders for tess/GS shaders with turnip. Part-of: <mesa/mesa!7274>
-
Danylo Piliaiev authoredSigned-off-by:
Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <mesa/mesa!7810> -
Samuel Pitoiset authored
Like GFX10.3 which is currently broken. Cc: 20.2 20.3 Signed-off-by:
Samuel Pitoiset <samuel.pitoiset@gmail.com> Reviewed-by:
Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl> Part-of: <mesa/mesa!7786>
-
Rhys Perry authoredSigned-off-by:
Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev> Part-of: <!7291>
-
Rhys Perry authored
Signed-off-by:
Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: -
Rhys Perry authored
Signed-off-by:
Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: -
Rhys Perry authored
Signed-off-by:
Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: -
Daniel Stone authoredTrying this again after reworking the virtualised network setup on the Windows runner. Signed-off-by:
Daniel Stone <daniels@collabora.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com> Part-of: <mesa/mesa!7537>
-
Daniel Stone authored
Signed-off-by:
Daniel Stone <daniels@collabora.com>
Reviewed-by: -
Daniel Stone authored
Signed-off-by:
Daniel Stone <daniels@collabora.com>
Reviewed-by: -
Daniel Stone authored
The full end-to-end job now completes in around 45min, which is far more reasonable than the previous 3.5h. Signed-off-by:
Daniel Stone <daniels@collabora.com>
Reviewed-by: -
Daniel Stone authored
The physical machine has 64, and the bottleneck is mostly I/O and network, so let's just go ahead and smash the CPUs to bits. Signed-off-by:
Daniel Stone <daniels@collabora.com>
Reviewed-by: -
Daniel Stone authored
We need these builds in Windows in order to build the microsoft-clc parts. Signed-off-by:
Daniel Stone <daniels@collabora.com>
Reviewed-by: -
Daniel Stone authored
Saturated conversions from f32 -> i64/u64 need to be reworked in the upstream tree. Disable this test until we've done so, since it doesn't pass currently. Signed-off-by:
Daniel Stone <daniels@collabora.com>
Reviewed-by: -
Daniel Stone authored
clang's version argument will return '1.2.3git' when built from a non-release source tree, but the resource directory will be '1.2.3'. Strip the 'git' suffix if it's there. Signed-off-by:
Daniel Stone <daniels@collabora.com>
Reviewed-by: -
Rhys Perry authored
These are no longer necessary and break the build. Signed-off-by:
Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: 
Erik Faye-Lund <erik.faye-lund@collabora.com>
Fixes: c9bcad25 ("nir: add generated intrinsic builders")
Part-of: <mesa/mesa!7808> -
Rhys Perry authored
MSVC C++ can't do designated initializers without /std:c++latest. These helpers will likely be removed soon anyway, so just don't use the intrinsic builders here. This should also fix the GCC7 build, which doesn't implement non-trivial designated initializers. Signed-off-by:
Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Fixes: c9bcad25 ("nir: add generated intrinsic builders")
Part-of: <mesa/mesa!7808> -
Rhys Perry authored
Fix this error: error C4576: a parenthesized type followed by an initializer list is a non-standard explicit type conversion syntax Signed-off-by:
Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Fixes: c9bcad25 ("nir: add generated intrinsic builders")
Part-of: <mesa/mesa!7808> -
Gert Wollny authored
Signed-off-by:
Gert Wollny <gert.wollny@collabora.com> Part-of: <mesa/mesa!7715>
-
Gert Wollny authored
Signed-off-by:
-
Gert Wollny authored
Signed-off-by:
-
Gert Wollny authored
Signed-off-by:
-
Gert Wollny authored
Signed-off-by:
-
Gert Wollny authored
Signed-off-by:
-
Gert Wollny authored
Signed-off-by:
-
Gert Wollny authored
Signed-off-by:
-
Gert Wollny authored
Signed-off-by:
-
Gert Wollny authored
Signed-off-by:
-
Gert Wollny authored
the values are loaded directly from a register and the specification of parycentrics is only in the way. Signed-off-by:
-
Gert Wollny authored
Fixes: 18e97817 r600/sfn: Use load_ubo_vec4 lowering pass Signed-off-by:
-
Boris Brezillon authored
If we don't do that the stride in texture/framebuffer descriptors is wrong, leading to page faults when those buffers are accessed. Fixes: 00360cd5 ("panfrost: Calculate the row stride at resource creation time") Signed-off-by:
Boris Brezillon <boris.brezillon@collabora.com> Reviewed-by:
Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <mesa/mesa!7797> -
Iago Toral authored
Also, update the hardware version assert check to match what we are really requiring for Vulkan. Reviewed-by:
Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <mesa/mesa!7788> -
Iago Toral authored
When copying multiple regions that have the same image subresource we are effectively copying various rects across the same layer range, so we can batch together all the rects to copy for each layer in a single job. This allows us to significantly reduce CPU overhead when recording the command, as we need to produce less jobs and allocate less descriptor sets. It also offers smaller gains in execution time due to the reduced job count. A stress test where we copy 10 subrects of an image in a loop 100 time, choosing regions that will involve the texel buffer path, we get these results: | Recording Time | Execution Time | ----------|----------------|----------------| master | 3.021s | 0.112s | ----------|----------------|----------------| patch | 0.163s | 0.080s | ----------|----------------|----------------| Reviewed-by:Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <mesa/mesa!7782> -
Iago Toral authored
Fixes: ba69c36a ("v3dv: add a buffer to image copy path using a texel buffer") Reviewed-by:
Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <mesa/mesa!7782> -
Marek Olšák authored
This increases performance for indexed triangle strips up to +100%. In practice, it's limited by memory bandwidth and compute power, so 256-bit memory bus and a lot of CUs are recommended. Acked-by:
Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com> Part-of: <mesa/mesa!7681>
-
Bas Nieuwenhuizen authored
This workaround fixes a hang while loading a renderdoc trace for me. Since the workload does 1 mip per cmdbuffer it is quite hard to confirm what exactly the conditions for the hang are but this is the most restrictive set I found and it corresponds to a workaround in AMDVLK as well. CC: mesa-stable Reviewed-by:
-
- 26 Nov, 2020 3 commits
-
-
Erik Faye-Lund authoredWe forgot to document this previously, so let's add it now. Fixes: feb9462b ("zink: Added inbuilt debug logging from the VK_LAYER_LUNARG_standard_validation layer.") Reviewed-By:
Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <mesa/mesa!7802> -
Rhys Perry authored
Make it consistent with nir_intrinsics.py, the unlabelled indices just before it and the intrinsic builders. Signed-off-by:
Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: 
Jason Ekstrand <jason@jlekstrand.net>
Part-of: <mesa/mesa!6587> -
Rhys Perry authored
Signed-off-by:
Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by:
-
