Instead of lowering it in NIR, use the lookup tables as inputs to a
second-order Taylor expansion. shader-db results aren't amazing but keep
in mind this is without backend CSE yet.
total instructions in shared programs: 115913 -> 115707 (-0.18%)
instructions in affected programs: 3151 -> 2945 (-6.54%)
helped: 12
HURT: 0
Instructions are helped.
total nops in shared programs: 84045 -> 84041 (<.01%)
nops in affected programs: 1571 -> 1567 (-0.25%)
helped: 1
HURT: 7
Inconclusive result (value mean confidence interval includes 0).
total clauses in shared programs: 20498 -> 20489 (-0.04%)
clauses in affected programs: 188 -> 179 (-4.79%)
helped: 6
HURT: 0
Clauses are helped.
total quadwords in shared programs: 90395 -> 90291 (-0.12%)
quadwords in affected programs: 2287 -> 2183 (-4.55%)
helped: 12
HURT: 0
Quadwords are helped.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9420>
When we were only able to pipeline smooth varyings, if we had to disable
ldvary pipelining in the middle of a sequence it would stay disabled for
the rest of the program, to prevent us from prioritizing scheduling of
ldvary instructions that we would not be able to pipeline effectively.
Now that we can pipeline all ldvary sequences we can change this.
This change re-enables ldvary pipelining upon finding the next
ldvary in the program in the hopes that we can continue pipelining
succesfully. To do this, we track the number of ldvary instructions we
emitted so far and compare that to the number of inputs in the fragment
shader we are scheduling. This also allows us to simplify our ldvary
tracking at nir to vir time, since that is all now handled in the QPU
scheduler.
total instructions in shared programs: 13817048 -> 13810783 (-0.05%)
instructions in affected programs: 810114 -> 803849 (-0.77%)
helped: 4843
HURT: 591
Instructions are helped.
total max-temps in shared programs: 2326612 -> 2326300 (-0.01%)
max-temps in affected programs: 4689 -> 4377 (-6.65%)
helped: 285
HURT: 7
Max-temps are helped.
total sfu-stalls in shared programs: 30942 -> 30865 (-0.25%)
sfu-stalls in affected programs: 207 -> 130 (-37.20%)
helped: 120
HURT: 42
Sfu-stalls are helped.
total inst-and-stalls in shared programs: 13847990 -> 13841648 (-0.05%)
inst-and-stalls in affected programs: 825378 -> 819036 (-0.77%)
helped: 4899
HURT: 590
Inst-and-stalls are helped.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9404>
Prior to SKL, the mipmaps for 3D surfaces are laid out in a way
that make it impossible to represent in the way that
VkSubresourceLayout expects. Since we can't tell users how to make
sense of them, don't report them as available.
"Fixes" dEQP-VK.image.subresource_layout.3d.*
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9419>
This implements most of the remaining u_threaded_context support. Most
of the heavy lifting was done in the previous patches which fixed things
up for the new thread safety requirements. Only a few things remain.
u_threaded_context support can be disabled via an environment variable:
GALLIUM_THREAD=0
On Felix's Tigerlake with the GPU at fixed frequency, enabling
u_threaded_context improves performance of several games:
- Civilization VI: +17%
- Shadow of Mordor: +6%
- Bioshock Infinite +6%
- Xonotic: +6%
Various microbenchmarks improve substantially as well:
- GfxBench5 gl_driver2: +58%
- SynMark2 OglBatch6: +54%
- Piglit drawoverhead: +25%
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8964>
pipe->transfer_map can be called from u_threaded_context's thread
rather than the driver thread. We need to use two different slab
allocators, one for each thread. transfer_unmap, on the other hand,
is only ever called from the driver thread.
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8964>
u_threaded_context requires various objects to inherit from a new
threaded_foo base class rather than directly from pipe_foo. This
patch does most of the mechanical changes required for that.
It also initializes the new threaded_resource fields.
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8964>
When we enable u_threaded_context, the pipe->create_*_state hooks
(precompile variants) are going to be called from one thread, while
iris_update_compiled_shaders (on-the-fly variants) are going to be
called from a driver thread. BLORP shaders also happen from
clear, blit, and so on in the driver thread.
u_upload_mgr isn't thread-safe, so use an uploader for each purpose.
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8964>
This enables us to replace the backing storage of resources that have
been used as stream output targets, in case we're invalidating their
entire contents. This can avoid stalls. We simply hadn't supported it
because it was going to be tricky to re-emit 3DSTATE_SO_BUFFER without
screwing up "reset offset to zero" vs. "keep appending". But that
should be working fine with the previous patch's refactor.
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8964>
The previous mechanism was a bit fragile. We stored the zero offset
in the pre-baked packet, and used an flag to override 0xFFFFFFFF
(append) offsets until our first emit - then prohibited anyone from
trying to re-emit the packet by flagging IRIS_DIRTY_SO_BUFFERS,
because that would re-emit the version with the zeroing of the offset.
Now, we always store 0xFFFFFFFF in the pre-baked packet, and use a
flag to override it to zero on the first emit. That way, we can
re-emit that packet at any time, and it'll just keep appending.
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8964>
In the future, Marek is planning to make u_threaded_context call
create_stream_output_target() from a different thread than the main
driver thread, which means that we can't safely use uploaders there.
To prepare for this eventual future, just defer the allocation of
the offset BO 'til later. It's a very small amount of overhead.
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8964>
With u_threaded_context, create_surface and create_sampler_view will
be called from a different thread than the driver thread. They aren't
allowed to access the context, which means that they can't use the
uploaders there to upload our SURFACE_STATE entries.
Thanks to backing-storage replacement and iris_rebind_buffer, we already
reworked things to maintain CPU-side copies of the SURFACE_STATE entries
and added the ability to upload or re-upload them later. So we can skip
the upload at object creation time, and add a simple resource-is-NULL
check at binding table upload time to ensure that they get uploaded by
the time we need them. (They might get uploaded earlier due to rebinds
or clear color updates, but this is the last moment to do so.)
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8964>
It shouldn't affect bound program state, and the current context state
shouldn't be relevant for shader creation precompiles anyway (level load
isn't going to have the eventual set of sampler views bound when you go to
draw with that shader).
Closes: #4306
Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9089>
mi_ is already a unique prefix in Mesa so the gen_ isn't really gaining
us anything except extra characters. It's possible that MI_ may
conflict a tiny bit with GenXML but it doesn't seem to be a problem
today and we can deal with that in the future if it's ever an issue.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9393>
Writes to global memory should not be moved over discard,
otherwise we could have unintended side-effects or lack of
side-effects where they should be observed.
Fixes tests:
dEQP-VK.rasterization.frag_side_effects.color_at_beginning.kill
dEQP-VK.rasterization.frag_side_effects.color_at_end.kill
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9365>
