Boris Brezillon [Tue, 8 Dec 2020 10:38:50 +0000 (11:38 +0100)]
pan/mdg: Add support for multi sample iteration writeout
Some MSAA+fmt combination require writeout to be split. Right now, it
only impacts blend shaders since we only support MSAA 4x, and the only
formats that could exceed the 128bit/pixel limit in MSAA 4x are
not supported by the fixed-function blend unit. We thus rely on the
blend shader to split things properly. Things will change once we add
MSAA 8x/16x to the mix, since even the blendable formats will exceed
the 128b/pixel limit in that case.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7984>
Boris Brezillon [Mon, 30 Nov 2020 10:07:33 +0000 (11:07 +0100)]
panfrost: Fix fencing
Commit
64d6f56ad26f ("panfrost: Allocate syncobjs in panfrost_flush")
aimed at optimizing the fencing logic but it looks it also broke the
fence-based synchronization in subtle ways.
Indeed, now that the fence only waits on a single syncobj, we're not
guaranteed that all jobs queued in panfrost_flush_all_batches() will
be done when the fence is signaled, because jobs at the top level
(those stored in the batches hashmap) have not inter-dependencies.
Commit
9e397956b092 ("panfrost: signal syncobj if nothing is going to
be flushed") made this even more apparent by signaling the fence right
away if nothing was left to be drawn in the current context, thus
ignoring any of the batches left to flushed in the ->batches map.
If we want to keep relying the existing kernel APIs there's clearly no
ideal solution here. We can either go back to the original fencing
mechanism where each fence contained an array of syncobjs to be tested
or serialize jobs that have no explicit dependencies so we know the last
submitted job will also be the last one to return. The orginal approach
has proven to add quite a significant overhead (caused by the amount of
ioctls and the time spent in kernel space to gather dma fences attached
to those syncobjs and test them). So let's go for the simple solution
where we have a single syncobj bound to the context which we update to
point to the last job out_sync every time we submit a top-level job.
This approach implies reworking the way we create fences since we
need to capture the syncobj state at the time the fence is created.
Unfortunately, there's not SYNCOBJ_CLONE ioctl, which forces us to
export/create/import a fence so we have a new object that's not
subject to changes done to the context syncobj.
If we want to further optimize the logic, we should probably explore
some of those options:
1/ Adding array based SYNCOBJ ioctls (SYNCOBJ_{CREATE,DESTROY,CLONE}_ARRAY)
so we can mitigate the cost of ioctls when we need to manipulate
arrays of syncobjs
2/ Support synchronization jobs. That is, jobs that have a NULL job chain
but an array of sync_in and a sync_out to allow creating
synchronization points
3/ Add syncobj aggregators so we only have to wait on one syncobj from
userspace. The syncobj aggregator would wait for all sub syncobjs to
be signaled before signaling the top-level one.
Fixes:
64d6f56ad26f ("panfrost: Allocate syncobjs in panfrost_flush")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7831>
Boris Brezillon [Mon, 30 Nov 2020 09:23:32 +0000 (10:23 +0100)]
panfrost: Make sure we always add a reader -> write dependency when needed
We shouldn't reset the ->writer field when a reader comes in because we
want subsequent readers to have a dependency on the writer too. Let's
add a new field encoding the last access type and use it to replace the
writer != NULL test.
Reported-by: Roman Elshin
Fixes:
c6ebff3ecdde ("panfrost: Remove panfrost_bo_access type")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7831>
Marek Olšák [Mon, 21 Sep 2020 13:08:31 +0000 (09:08 -0400)]
st/mesa: enable compute shader derivatives in SPIR-V
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6799>
Marek Olšák [Mon, 21 Sep 2020 11:54:12 +0000 (07:54 -0400)]
ac,radeonsi: implement GL_NV_compute_shader_derivatives
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6799>
Marek Olšák [Tue, 8 Dec 2020 19:43:07 +0000 (14:43 -0500)]
winsys/amdgpu: use VRAM for command buffers if all VRAM is visible
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7951>
Marek Olšák [Sun, 6 Dec 2020 19:49:28 +0000 (14:49 -0500)]
radeonsi: map PIPE_USAGE_STREAM to VRAM if all VRAM is visible
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7951>
Marek Olšák [Sun, 6 Dec 2020 19:46:20 +0000 (14:46 -0500)]
radeonsi: unify uploaders and upload to VRAM if all VRAM is visible
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7951>
Marek Olšák [Sun, 6 Dec 2020 19:03:46 +0000 (14:03 -0500)]
radeonsi: only use staging for linear textures when all VRAM is not visible
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7951>
Marek Olšák [Sun, 6 Dec 2020 19:01:15 +0000 (14:01 -0500)]
radeons: only force staging uploads for VRAM when all VRAM is not visible
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7951>
Marek Olšák [Sun, 6 Dec 2020 18:59:53 +0000 (13:59 -0500)]
ac: add radeon_info::all_vram_visible for Smart Access Memory
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7951>
Marek Olšák [Sun, 6 Dec 2020 18:49:21 +0000 (13:49 -0500)]
radeonsi: initialize ctx and gfx_cs first, then allocators
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7951>
Rhys Perry [Tue, 1 Dec 2020 14:24:53 +0000 (14:24 +0000)]
compiler: update gl_access_qualifier comments
Clarify ACCESS_NON_{READABLE,WRITEABLE} and update ACCESS_CAN_REORDER.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6483>
Rhys Perry [Tue, 25 Aug 2020 16:36:14 +0000 (17:36 +0100)]
nir/opt_access: infer writeonly
This isn't always done for GL because it could cause ImageAccess to be
incorrect: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3278
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6483>
Rhys Perry [Tue, 25 Aug 2020 15:35:44 +0000 (16:35 +0100)]
nir/opt_access: consider global stores
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6483>
Rhys Perry [Tue, 25 Aug 2020 15:50:24 +0000 (16:50 +0100)]
nir/opt_access: handle variable pointers
We might not be able to determine what variable a instruction accesses.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6483>
Rhys Perry [Tue, 25 Aug 2020 16:03:03 +0000 (17:03 +0100)]
nir/opt_access: add basic Vulkan support
This involves determining the variables referenced by intrinsics, setting and
using the access qualifier correctly and considering that images and buffers
can alias.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6483>
Rhys Perry [Tue, 25 Aug 2020 13:30:52 +0000 (14:30 +0100)]
nir/opt_access: rename can_reorder() and set ACCESS_NON_WRITEABLE in it
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6483>
Rhys Perry [Tue, 1 Dec 2020 14:38:32 +0000 (14:38 +0000)]
nir/opt_access: don't check restrict in can_reorder()
ACCESS_NON_WRITEABLE means that the memory is read-only, not the variable,
so we don't have to check for aliasing.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6483>
Rhys Perry [Tue, 1 Dec 2020 14:34:48 +0000 (14:34 +0000)]
nir/opt_access: check restrict before marking a variable as readonly
ACCESS_NON_WRITEABLE means that the memory is read-only, not the variable,
so we have to check for aliasing first.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6483>
Rhys Perry [Tue, 1 Dec 2020 14:30:22 +0000 (14:30 +0000)]
nir/opt_access: ignore barriers and coherent qualifier
It shouldn't matter whether an access/variable is coherent or not, just
that it's not written. The coherent qualifier doesn't mean anything with
read-only memory.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6483>
Rhys Perry [Tue, 1 Dec 2020 14:26:40 +0000 (14:26 +0000)]
nir/opt_access: don't ignore image arrays in process_variable()
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6483>
Andrii Simiklit [Mon, 30 Nov 2020 18:15:10 +0000 (20:15 +0200)]
iris: update depth value for stages after fast clear depth
This should fix an outdated depth value visible from the shader side.
v1: (Nanley Chery)
- fix commit message
v2: (Nanley Chery)
- replace `fixes` tag by `mesa-stable`
Closes: #3883
Cc: 20.3 20.2 <mesa-stable>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Signed-off-by: Andrii Simiklit <andrii.simiklit@globallogic.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7838>
Juan A. Suarez Romero [Tue, 1 Dec 2020 08:45:39 +0000 (09:45 +0100)]
v3d: remove old tile blit code
This code was a direct copy from VC4, and it was disabled, as it did not
work.
Now that we have an implementation, let's remove it.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7816>
Juan A. Suarez Romero [Tue, 1 Dec 2020 08:39:52 +0000 (09:39 +0100)]
v3d: implement tile-based blit operation
This implements fast-path blit using the TLB to blit from one buffer to
another, if conditions for allowing this are met.
v1:
- Move checks in the code (Iago)
v2:
- Use function to compute tile width and height (Iago)
- Fix commit message (Iago)
- Use surface size to compute draw_tiles_{x,y} (Iago)
- Move checks (Iago)
- Fix tile draw parameters (Iago)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7816>
Juan A. Suarez Romero [Thu, 3 Dec 2020 10:32:40 +0000 (11:32 +0100)]
v3d: refactor set tile buffer size function
Change it to not require a v3d job to compute the tile buffer
parameters.
v1:
- Pass nr_cbufs parameter to get tile buffer size function (Iago)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7816>
Juan A. Suarez Romero [Fri, 27 Nov 2020 13:23:43 +0000 (14:23 +0100)]
v3d: implement tile buffer blits
This implements blit operation using the TLB.
It uses a source color buffer (bbuf) which will be blitted to color
buffer 0.
It also takes in account the number of samples for the input and output
so it can perform multisample resolve.
v1:
- Fix comment (Iago)
- Removed needless brackets (Iago)
- Ensure msaa is correctly set (Iago)
- Get rid of job->resolve (Iago)
- Add rbuf as part of job's key (Iago)
- Rename rbuf/rsurf by bbuf/bsurf (Iago)
- Revert needless change (Iago)
v2:
- Remove spurious change (Iago)
- Add assert for safety reasons (Iago)
- Add brackets in condition (Iago)
- Fix commit message and title (Iago)
- Do tile blit only for version >=4.0 (Iago)
v3:
- Add assertion (Iago)
- Fix comment (Iago)
- Change commit title (Iago)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7816>
Juan A. Suarez Romero [Fri, 27 Nov 2020 13:17:59 +0000 (14:17 +0100)]
v3d: add helper to check if format supports TLB resolve
It checks if the TLB can perform multisample resolve for the specified
format.
v1:
- Fix commit title (Iago)
v2:
- Fix identation (Iago)
- Fix prototype style (Iago)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7816>
Juan A. Suarez Romero [Fri, 27 Nov 2020 13:14:00 +0000 (14:14 +0100)]
v3d: split binning start from draw
Separate the binning part in the draw and make it public, so we can
invoke it later from a different part.
v1:
- Ensure the job needs flush (Iago)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7816>
Juan A. Suarez Romero [Wed, 2 Dec 2020 12:53:17 +0000 (13:53 +0100)]
v3d: store number of color buffers in job
Avoids requiring to calculate in other places.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7816>
Juan A. Suarez Romero [Fri, 27 Nov 2020 13:10:18 +0000 (14:10 +0100)]
v3d: make set tile buffer size function public
This will be used later when implementing Tile-based blit function.
v1:
- Fix commit title (Iago)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7816>
Juan A. Suarez Romero [Fri, 27 Nov 2020 12:07:43 +0000 (13:07 +0100)]
v3d: force alpha to 1 when rendering RGBX formats
This adds an enum to the load tile buffer that forces the alpha channel
to be set to 1. This will be required later to load RGBX formats.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7816>
Juan A. Suarez Romero [Fri, 27 Nov 2020 12:03:16 +0000 (13:03 +0100)]
util: function to check for rgbX format
Function that checks if the format has its alpha component forced to 1.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7816>
Marcin Ślusarz [Tue, 8 Dec 2020 18:38:01 +0000 (19:38 +0100)]
intel/tools/aubinator_error_decode: fix small memory leaks
Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7994>
Marcin Ślusarz [Tue, 8 Dec 2020 18:27:43 +0000 (19:27 +0100)]
intel/tools/aubinator_error_decode: cleanup path/file handling
Should help some compilers/static analyzers understand this code and avoid
things like this:
../src/intel/tools/aubinator_error_decode.c:850:19: warning: "path" may be used uninitialized in this function [-Wmaybe-uninitialized]
850 | ret = asprintf(&filename, "%s/%d/i915_error_state", path, minor);
Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7994>
Vinson Lee [Thu, 3 Dec 2020 23:12:07 +0000 (15:12 -0800)]
r600/sfn: Initialize ShaderInputVarying members in constructors.
Fix defects reported by Coverity Scan.
uninit_member: Non-static class member m_interpolate is not initialized in this constructor nor in any functions that it calls.
uninit_member: Non-static class member m_lds_pos is not initialized in this constructor nor in any functions that it calls.
uninit_member: Non-static class member m_mask is not initialized in this constructor nor in any functions that it calls.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7909>
Vinson Lee [Wed, 25 Nov 2020 21:18:03 +0000 (13:18 -0800)]
meson: Fix build with llvm-12.
This patch fixes this Meson build error.
$ meson builddir \
-Dshared-llvm=disabled
-Ddri-drivers=''
-Dbuild-tests=true \
-Dgallium-drivers=swrast \
-Dvulkan-drivers=''
[...]
/usr/bin/ld: src/gallium/auxiliary/libgallium.a(gallivm_lp_bld_misc.cpp.o): in function `llvm::InitializeNativeTarget()':
llvm/Support/TargetSelect.h:118: undefined reference to `LLVMInitializeX86TargetInfo'
/usr/bin/ld: llvm/Support/TargetSelect.h:119: undefined reference to `LLVMInitializeX86Target'
/usr/bin/ld: llvm/Support/TargetSelect.h:120: undefined reference to `LLVMInitializeX86TargetMC'
/usr/bin/ld: src/gallium/auxiliary/libgallium.a(gallivm_lp_bld_misc.cpp.o): in function `llvm::InitializeNativeTargetAsmPrinter()':
llvm/Support/TargetSelect.h:132: undefined reference to `LLVMInitializeX86AsmPrinter'
/usr/bin/ld: src/gallium/auxiliary/libgallium.a(gallivm_lp_bld_misc.cpp.o): in function `llvm::InitializeNativeTargetDisassembler()':
llvm/Support/TargetSelect.h:156: undefined reference to `LLVMInitializeX86Disassembler'
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7777>
Eric Anholt [Tue, 8 Dec 2020 23:18:34 +0000 (15:18 -0800)]
ci/vc4: Skip VS dynamic loops tests that cause GPU hangs.
We've been getting spurious failures from the new VC4 CI, which I believe
are due to this set of tests (which have been showing up along with a GPU
hang report in the list of flaky tests in the failing jobs). This was a
known issue I had in vc4.
Reviewed-by: Jose Maria Casanova Crespo <jmcasanova@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7998>
Marek Olšák [Tue, 8 Dec 2020 21:58:29 +0000 (16:58 -0500)]
radeonsi: fix the blit test for SW_64KB_R_X
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7908>
Marek Olšák [Tue, 8 Dec 2020 21:50:03 +0000 (16:50 -0500)]
radeonsi: rename SI_TEST_DMA to SI_TEST_BLIT
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7908>
Marek Olšák [Thu, 3 Dec 2020 22:57:45 +0000 (17:57 -0500)]
radeonsi: remove SDMA support
There are many issues with SDMA across many generations of hardware.
A recent example is that gfx10.3 suffers from random GPU hangs if
userspace uses SDMA.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7908>
Marek Olšák [Thu, 3 Dec 2020 23:01:11 +0000 (18:01 -0500)]
radeonsi: rename buffer functions so as not to reference rings
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7908>
Marek Olšák [Thu, 3 Dec 2020 22:10:06 +0000 (17:10 -0500)]
radeonsi: move si_screen_clear_buffer into si_compute_blit.c w/o SDMA option
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7908>
Eric Anholt [Tue, 8 Dec 2020 18:29:06 +0000 (10:29 -0800)]
ci/freedreno: Detect the cheza power management bus error and restart.
This is an issue on the cheza platform, the theory is due to some old
firmware bug that will be fixed in future platforms. Given that cheza was
a target that didn't get released and we expect future platforms to be
fixed, just detect the issue and restart.
I've noticed this error in my CI monitoring less than once a week.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7993>
Mauro Rossi [Mon, 7 Dec 2020 22:26:30 +0000 (23:26 +0100)]
android: spirv: fix '::' typo in gen rules
among all Android gen rules '::' was used only here to declare dependencies;
mesa development and stable branch are worth receiving the fix
Fixes the following building errors with Android 7:
obj/STATIC_LIBRARIES/libmesa_nir_intermediates/spirv/gl_spirv.P:184: *** target file
gen/STATIC_LIBRARIES/libmesa_nir_intermediates/spirv/vtn_generator_ids.h' has both : and :: entries. Stop.
Cc: "20.3" <mesa-stable@lists.freedesktop.org>
Fixes:
1070bba19e26 ("android: fix SPIR-V -> NIR build")
Reported-by: youling257 <youling257@gmail.com>
Mike Blumenkrantz [Sat, 5 Dec 2020 16:42:54 +0000 (11:42 -0500)]
mesa/st: set reserved storage for params+values to 16
zink apparently has a lot of uniform data, and 8x8 isn't enough to avoid realloc
fixes mesa/mesa#3930
Fixes:
eda37fb2 ("mesa: properly disallow param list reallocation")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7990>
Michel Zou [Sun, 6 Dec 2020 11:35:42 +0000 (12:35 +0100)]
meson: fix multiline string warning
gets rid of: WARNING: Newline character in a string detected, use '''
(three single quotes) for multiline strings instead.
Reviewed-by: Yevhenii Kharchenko <yevhenii.kharchenko@globallogic.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7946>
Ian Romanick [Tue, 8 Dec 2020 03:39:57 +0000 (19:39 -0800)]
util: Add cnd_monotonic to Makefile.sources
Fixes:
33a78948283 ("util,radv: Cross-platform monotonic condition variable")
Acked-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7980>
Marek Olšák [Sat, 5 Dec 2020 17:20:07 +0000 (12:20 -0500)]
mesa: remove MAX_3D_TEXTURE_LEVELS, MAX_CUBE_TEXTURE_LEVELS
they are redundant with MAX_TEXTURE_LEVELS
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7956>
Marek Olšák [Sat, 5 Dec 2020 17:03:42 +0000 (12:03 -0500)]
mesa: remove code for old (mostly unsupported) GL_NV_point_sprite
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7956>
Marek Olšák [Sat, 5 Dec 2020 16:47:50 +0000 (11:47 -0500)]
mesa: fix glPopAttrib for GL_COORD_REPLACE for r200
Fixes:
959380dc "mesa: more optimizations in glPopAttrib (colormask, drawbuffers, coord replace)"
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7956>
Jonathan Marek [Thu, 3 Dec 2020 05:05:37 +0000 (00:05 -0500)]
turnip: always set LRZ registers to zero for 3d clear/blit
Apparently LRZ will be read/written regardless of depth being enabled or
not, so we have to make sure these registers are zero.
Fixes:
1d83f5ae8435 ("turnip: disable LRZ on vkCmdClearattachments() 3D fallback path")
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7899>
Jonathan Marek [Thu, 3 Dec 2020 05:03:14 +0000 (00:03 -0500)]
turnip: move up LRZ invalidate in CmdClearAttachments
There is an early return if cmd->state.predication_active is true, so do
the LRZ invalidate before that.
Fixes:
2f79e0066405 ("turnip: disable LRZ on vkCmdClearAttachments()")
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7899>
Jonathan Marek [Thu, 3 Dec 2020 04:59:24 +0000 (23:59 -0500)]
turnip: do not emit draw states in draw_cs outside of renderpass
This avoids a possible issue with MSAA sysmem clears, which use a 3D clear
path which assumes draw states are disabled, and are emitted in draw_cs in
BeginRenderPass.
(checking for TU_CMD_DIRTY_DRAW_STATE also allows not emitting the draw
states if they will be re-emitted on the next draw anyway. the previous
patch makes it so TU_CMD_DIRTY_DRAW_STATE is always set outside of
renderpasses)
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7899>
Jonathan Marek [Thu, 3 Dec 2020 04:49:26 +0000 (23:49 -0500)]
turnip: correctly disable draw states outside of renderpasses
* do the disable in EndRenderPass2 to fix the missing disable for sysmem
* we don't need a disable at the end of every tile, or between binning pass
and gmem pass (the first draw in draw_cs emits all the draw states)
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7899>
Jonathan Marek [Thu, 3 Dec 2020 04:32:13 +0000 (23:32 -0500)]
turnip: always emit LRZ draw state in DIRTY_DRAW_STATE path
The packet size is constant and assumes all states, except for the 2 input
attachment states. (this means we get an invalid packet if DIRTY_LRZ isn't
set when DIRTY_DRAW_STATE is set).
Fixes:
3c07a1499863 ("turnip: enable LRZ")
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7899>
Jonathan Marek [Thu, 3 Dec 2020 04:28:38 +0000 (23:28 -0500)]
turnip: do not include compute stage in pipeline_builder
This avoids emitting compute-related state in the graphics pipeline
(tu6_emit_xs_config was being called for compute stage).
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7899>
Jonathan Marek [Thu, 3 Dec 2020 04:23:20 +0000 (23:23 -0500)]
turnip: no linear_to_srgb for alpha channel for gmem clear value packing
Alpha channel is always linear (oops).
Fixes:
ddac5933f8f3 ("turnip: call packing functions directly for pack_gmem_clear_value")
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7899>
Marek Olšák [Sun, 29 Nov 2020 06:25:01 +0000 (01:25 -0500)]
gallium/u_threaded: optimize set_constant_buffer
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7719>
Marek Olšák [Sat, 28 Nov 2020 10:24:06 +0000 (05:24 -0500)]
gallium/u_threaded: don't make a local copy of pipe_draw_start_count
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7719>
Marek Olšák [Sat, 28 Nov 2020 10:22:06 +0000 (05:22 -0500)]
gallium/u_threaded: don't copy the indexbuf pointer if we overwrite it
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7719>
Marek Olšák [Sat, 28 Nov 2020 10:18:38 +0000 (05:18 -0500)]
gallium/u_threaded: set has_user_indices = false in the driver thread
to remove some overhead from the main thread.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7719>
Marek Olšák [Sat, 28 Nov 2020 10:03:31 +0000 (05:03 -0500)]
gallium/u_threaded: don't pass index bounds to the driver to decrease overhead
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7719>
Marek Olšák [Sat, 10 Oct 2020 09:09:53 +0000 (05:09 -0400)]
glthread: count batch space in units of uint64_t elements
This removes one x86 shr instruction from _mesa_glthread_allocate_command.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7719>
Marek Olšák [Sat, 10 Oct 2020 03:05:08 +0000 (23:05 -0400)]
glthread: change sizes to unsigned or size_t where needed
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7719>
Marek Olšák [Sat, 10 Oct 2020 02:37:28 +0000 (22:37 -0400)]
glthread: use uint64_t to declare the batch buffer instead of align(8)
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7719>
Marek Olšák [Sat, 10 Oct 2020 02:16:29 +0000 (22:16 -0400)]
glthread: use glthread->used instead of glthread->next_batch->used
remove one pointer dereference by having "used" in glthread too
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7719>
Mike Blumenkrantz [Mon, 7 Dec 2020 18:41:03 +0000 (13:41 -0500)]
zink: use shader keys for samplemask
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7801>
Mike Blumenkrantz [Mon, 7 Dec 2020 20:43:33 +0000 (15:43 -0500)]
zink: change a memcmp==0 to !memcmp
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7801>
Mike Blumenkrantz [Thu, 26 Nov 2020 17:08:46 +0000 (12:08 -0500)]
zink: initial shader key implementation
this is a squashed version of a previously-reviewed series
ref https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7193
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7801>
Witold Baryluk [Mon, 7 Dec 2020 20:45:42 +0000 (20:45 +0000)]
vulkan/device_select: Store Vulkan vendorID and deviceID as uint32_t
Vulkan uses 32-bit IDs, compared to PCI/USB 16-bit ones.
Some driver vendorIDs do exceed 0xffff, including MESA (used by lavapipe).
Without this, the value will be truncated, and device select layer
will not match expected (or any) device.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3935
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7977>
Rob Clark [Mon, 7 Dec 2020 17:08:10 +0000 (09:08 -0800)]
gallium/aux: Update scons build for u_tracepoints.[ch]
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7974>
Rob Clark [Mon, 7 Dec 2020 17:05:29 +0000 (09:05 -0800)]
gallium/aux: Split u_tracepoints.[ch] generation
Allow for separate rules to be used to generate the .c and .h (partially
to make it easier for scons build, and partially because that is just a
better practice).
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7974>
Rob Clark [Sat, 5 Dec 2020 18:39:45 +0000 (10:39 -0800)]
gallium/aux: Avoid creating queue when traces not enabled
Avoids unnecessarily creating an extra thread when tracing is not
enabled. Which also side-steps a chrome/ium sandbox issue with
the setscheduler call in u_queue.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7974>
Rhys Perry [Fri, 22 May 2020 12:39:06 +0000 (13:39 +0100)]
nir: improve divergence analysis for loads with non-uniform resources
If ACCESS_NON_UNIFORM is not specified, we can assume the resource is
uniform. This requires nir_lower_non_uniform_access to remove that flag.
A few Detroit: Become Human shaders use a index sourced from a fragment
input which is expected to be uniform.
shader-db (Navi):
Totals from 8 (0.01% of 127638) affected shaders:
SGPRs: 224 -> 384 (+71.43%)
VGPRs: 208 -> 112 (-46.15%)
CodeSize: 5360 -> 5344 (-0.30%); split: -1.49%, +1.19%
Instrs: 1036 -> 1028 (-0.77%); split: -1.93%, +1.16%
VMEM: 1320 -> 608 (-53.94%)
SMEM: 384 -> 336 (-12.50%); split: +14.58%, -27.08%
VClause: 24 -> 16 (-33.33%)
SClause: 48 -> 56 (+16.67%)
PreSGPRs: 124 -> 216 (+74.19%)
PreVGPRs: 168 -> 88 (-47.62%)
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5201>
Rhys Perry [Tue, 6 Oct 2020 16:08:33 +0000 (17:08 +0100)]
nir/lower_non_uniform: remove non_uniform flags after lowering
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5201>
yshi18 [Thu, 3 Dec 2020 05:54:14 +0000 (13:54 +0800)]
iris: fix memleak for query_buffer_uploader
In the Chrome WebGL Aquarium stress test, 20 instances of Chrome will run
Aquarium simultaneously over 20+ hours. That causes Chrome crash.
During the stress, glBeginQueryIndexed is called frequently.
1.Each query will only use 32 bytes from query_buffer_uploader. After the offset
exceed 4096, it will alloc new buffer for query_buffer_uploader->buffer
and release the old buffer.
2.But iris_begin_query will call u_upload_alloc when the offset changed, and it
will increase the query_buffer_uploader->buffer->reference.count every time
when it called u_upload_alloc.
3.So when u_upload_release_buffer try to release the resource of
query_buffer_uploader->buffer, its reference.count is
already equal to 129. pipe_reference_described will only decrease its reference
count to 128.So it never called old_dst->screen->resource_destroy.
4.The old resouce bo will never be freeed. And chrome will called mmap every time
when it alloc new resource bo.
5. Chrome process map too many vmas in its process. Its map count exceed the
sysctl_max_map_count which is 65530 defined in kernel.
6. When iris_begin_query want to alloc new resource bo, it will meet NULL pointer
because mmap return failed. Finally chrome crashed when it access this NULL resource
bo.
The fix is decrease the reference count in iris_destroy_query.
Patch is verified by chrome webgl Aquarium test case for more than 72 hours.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Signed-off-by: Yang Shi <yang.a.shi@intel.com>
Reviewed-by: Alex Zuo <alex.zuo@intel.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7890>
Jonathan Gray [Tue, 4 Aug 2020 07:22:32 +0000 (17:22 +1000)]
aco: use UINT64_C on 64 bit constant arguments
avoids errors seen when building on OpenBSD/amd64
../src/amd/compiler/aco_instruction_selection.cpp:1677:62: error: ambiguous conversion for functional-style cast from 'unsigned long' to 'aco::Operand'
bld.vop3(aco_opcode::v_mul_f64, Definition(dst), Operand(0x3FF0000000000000lu), tmp);
^~~~~~~~~~~~~~~~~~~~~~~~~~~
glibc uses unsigned long for uint64_t on LP64 archs and unsigned long long for
uint64_t on ILP32 archs. On OpenBSD unsigned long long is used for uint64_t
on all archs.
The Operand constructors are uint8_t uint16_t uint32_t uint64_t
use UINT64_C so lu or llu suffix will be used as needed.
Fixes:
df645fa369d ("aco: implement VK_KHR_shader_float_controls")
Signed-off-by: Jonathan Gray <jsg@jsg.id.au>
Reviewed-by: Tony Wasserka <tony.wasserka@gmx.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7944>
Gert Wollny [Sun, 25 Oct 2020 18:28:22 +0000 (19:28 +0100)]
mesa/st: lower 64 bit ops to scalar before lowering to soft-float
The fp64 emulation is not prepared for vectorized 64 bit code, so
if the driver doesn't ask for lowering to scalar by itself, do it before
lowering to soft-fb, and run a vectorizazion pass afterwards.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7961>
Samuel Pitoiset [Mon, 7 Dec 2020 11:38:46 +0000 (12:38 +0100)]
radv: disable alphaToOne feature
The feature was exposed but completely ignored by the driver. Other
AMD drivers don't expose it as well, probably because it's complicated
to implement alpha-to-coverage properly. Let's disable it.
Cc: mesa-stable.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7966>
Pierre-Eric Pelloux-Prayer [Wed, 4 Nov 2020 09:19:18 +0000 (10:19 +0100)]
driconf: add allow_incorrect_primitive_id option
And enable it for SPECviewperf.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7078>
Pierre-Eric Pelloux-Prayer [Tue, 20 Oct 2020 08:57:04 +0000 (10:57 +0200)]
vbo/dlist: only use merged primitives when it's ok to do so
Merging primitives generates incorrect gl_PrimitiveID[In] values.
So make merged primitives construction non-destructive and fallback
to drawing with original primitives if a program reads gl_PrimitiveId.
This commit adds _mesa_update_primitive_id_is_unused modeled after
_mesa_update_allow_draw_out_of_order to update ctx->_PrimitiveIDIsUnused
each time shaders are updated.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7078>
Pierre-Eric Pelloux-Prayer [Tue, 20 Oct 2020 08:57:04 +0000 (10:57 +0200)]
vbo/dlist: avoid splitting draw commands in multiple draws
For (Multi)DrawArrays and (Multi)DrawElements commands, the storage size
needed are known early so we can make sure that the prim_store/vertex_store
will be big enough to store the whole command.
This reduces the amount of drawcalls in snx03 tests. For instance in test10:
| Num draw calls | GPU-load |
------|----------------|-----------------|
| Before | After | Before | After |
------|--------|-------|---------|-------|
test10| 35k | 8k | 58% | 80% |
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7078>
Pierre-Eric Pelloux-Prayer [Tue, 20 Oct 2020 08:38:37 +0000 (10:38 +0200)]
vbo/dlist: refactor prim_store/vertex_store allocations
This will be used in the next commit.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7078>
Pierre-Eric Pelloux-Prayer [Wed, 7 Oct 2020 15:31:51 +0000 (17:31 +0200)]
mesa: optimize _mesa_program_resource_location
- xxhash is faster than sha1.
- remove superfluous calls to strlen
Using SPECviewperf13 snx-03 first subtest and "perf -e cycles -g", perf report says:
Before | After | Function
---------|--------|---------------
47.39% | 47.36% | _mesa_CallList
5.00% | 3.03% | _mesa_program_resource_location
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7078>
Pierre-Eric Pelloux-Prayer [Thu, 8 Oct 2020 14:49:01 +0000 (16:49 +0200)]
util/hash_table: add _mesa_hash_data_with_seed function
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7078>
Pierre-Eric Pelloux-Prayer [Thu, 8 Oct 2020 12:49:29 +0000 (14:49 +0200)]
vbo/dlist: implement primitive merging
Merge consecutive primitives using the same mode while constructing the index buffer.
This improves performance a lot (x3 - x10) SPECviewperf13 snx-03 test by reducing the
number of draw calls per frame.
Here are some numbers for 4 of the tests:
| Num draw calls | GPU-load |
------|----------------|-----------------|
| Before | After | Before | After |
------|--------|-------|---------|-------|
test1 | 390k | 16k | 68% | 90% |
test2 | 370k | 16k | 40% | 90% |
test3 | 1.2M | 35k | 38% | 78% |
test10| 3.5M | 35k | 36% | 58% |
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7078>
Pierre-Eric Pelloux-Prayer [Thu, 8 Oct 2020 12:48:25 +0000 (14:48 +0200)]
vbo/dlist: convert LINE_STRIPS to LINES
Less primitive modes allows for better primitive merging.
Lines are always used (instead of picking dynamically lines or line
strips for instance) because:
- they don't need primitive restarts to be merged
- they perform better (at least on radeonsi) - SPECviewperf13 snx subtests
with lines (like 4 or 10) are 1.5x-2x faster.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7078>
Pierre-Eric Pelloux-Prayer [Thu, 8 Oct 2020 12:43:57 +0000 (14:43 +0200)]
vbo/dlist: create an index buffer in compile_vertex_list
This will be useful to implement other optimizations.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7078>
Pierre-Eric Pelloux-Prayer [Tue, 6 Oct 2020 09:58:29 +0000 (11:58 +0200)]
dlist: do not call _mesa_lookup_list twice
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7078>
Mauro Rossi [Sat, 5 Dec 2020 01:58:32 +0000 (02:58 +0100)]
android: radv: add libcutils shared dependency
Fixes the following building error:
FAILED: out/target/product/x86_64/obj/SHARED_LIBRARIES/vulkan.android-x86_intermediates/LINKED/vulkan.android-x86.so
...
ld.lld: error: undefined symbol: property_get
>>> referenced by os_misc.c:193 (external/mesa/src/util/os_misc.c:193)
>>> os_misc.o:(os_get_option) in archive out/target/product/x86_64/obj/STATIC_LIBRARIES/libmesa_util_intermediates/libmesa_util.a
Fixes:
eeecc21d ("util: Add property_get() fallback for android")
Reviewed-by: Marijn Suijten <marijn.suijten@somainline.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7861>
Marijn Suijten [Tue, 1 Dec 2020 18:19:29 +0000 (19:19 +0100)]
mesa/math: Fix address of array always returning true
external/mesa3d/src/mesa/math/m_matrix.c:1403:13: error: address of array 'mat->inv' will always evaluate to 'true' [-Werror,-Wpointer-bool-conversion]
if (mat->inv && (mat->flags & MAT_DIRTY_INVERSE)) {
~~~~~^~~ ~~
Fixes:
3175b63a0df ("mesa: don't allocate matrices with malloc")
Reviewed-by: Rob Clark <robclark@freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7861>
Marijn Suijten [Tue, 1 Dec 2020 18:13:14 +0000 (19:13 +0100)]
android: util: Add libcutils to Android.mk shared libs
Otherwise:
external/mesa3d/src/util/os_misc.c:59:12: fatal error: 'cutils/properties.h' file not found
# include <cutils/properties.h>
^~~~~~~~~~~~~~~~~~~~~
And:
ld.lld: error: undefined symbol: property_get
>>> referenced by os_misc.c:193 (external/mesa3d/src/util/os_misc.c:193)
>>> os_misc.o:(os_get_option) in archive out/target/product/discovery/obj/STATIC_LIBRARIES/libmesa_util_intermediates/libmesa_util.a
Fixes:
eeecc21d935 ("util: Add property_get() fallback for android")
Acked-by: Rob Clark <robclark@freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7861>
Witold Baryluk [Mon, 7 Dec 2020 20:30:32 +0000 (20:30 +0000)]
zink: Cap PIPE_SHADER_CAP_MAX_CONST_BUFFERS to 32
PIPE_MAX_CONSTANT_BUFFERS is 32, however many Vulkan implementations
has maxPerStageDescriptorUniformBuffers that exceeds it, for example:
radv 8388606,
anv 64
nvidia 1048580 for RTX 2000 and up.
and, together with the current zink logic, the returned value
will exceed the maximum allowed value for the cap.
This causes cso_destroy_context to pass big values back to zink
(via zink_set_constant_buffer), resulting in access beyond end of
allocated buffer for all UBOs.
Cap the cap to PIPE_MAX_CONSTANT_BUFFERS (32), not INT_MAX.
Add an assert to verify future drivers.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Fixes:
daaf5f1d186 ("gallium: Fix leak of currently bound UBOs at CSO context destruction.")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7976>
Marek Olšák [Sat, 28 Nov 2020 07:25:49 +0000 (02:25 -0500)]
gallium: fix the PIPE_SHADER_CAP_SUPPORTED_IRS value for all drivers
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7826>
Marek Olšák [Tue, 24 Nov 2020 05:11:48 +0000 (00:11 -0500)]
st/mesa: remove less useful debug options in hot paths
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7826>
Marek Olšák [Sat, 28 Nov 2020 07:41:46 +0000 (02:41 -0500)]
st/mesa: don't generate TGSI for the draw VS because it now supports NIR too
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7826>
Marek Olšák [Sat, 28 Nov 2020 07:38:16 +0000 (02:38 -0500)]
draw: add NIR support to draw_create_vertex_shader
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7826>
Erik Faye-Lund [Mon, 7 Dec 2020 11:27:23 +0000 (12:27 +0100)]
zink: fix channel ordering in format-mapping
This looks like a typo. Packed vulkan formats should always map to the
inverse order of the corresponding gallium notation. Besides, it makes
no sense that unsigned and signed formats have different ordering.
Fixes:
cdfb1d925f3 ("zink: add last few format maps for ARB_vertex_type_2_10_10_10_rev")
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7964>
Gert Wollny [Mon, 7 Dec 2020 12:01:35 +0000 (13:01 +0100)]
r600/sfn: Add support for shader_clock
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7973>