AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
169,126 Commits 1 Branch 0 Tags
6826a0ab14d4af4e41db6bf550a799bcdfd76d5b
Commit Graph

169126 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Danylo Piliaiev
6826a0ab14 freedreno/computerator: C++ proofing 
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22148>
 2023-03-30 23:40:48 +00:00
Danylo Piliaiev
5d2ddce99f freedreno/registers: More a7xx regs 
Based on 011c54b0 from Jonathan Marek.

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22148>
 2023-03-30 23:40:48 +00:00
Danylo Piliaiev
899d142336 freedreno/registers: Document new CP_EVENT_WRITE::SEQNO 
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22148>
 2023-03-30 23:40:48 +00:00
Alyssa Rosenzweig
1e67f71324 panfrost: Add a v9 fast path for no images 
The usual case.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21848>
 2023-03-30 23:21:59 +00:00
Alyssa Rosenzweig
e6529d6dcc panfrost: Don't update access with a single batch 
drawoverhead test 25 from 462->492

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21848>
 2023-03-30 23:21:59 +00:00
Alyssa Rosenzweig
c224bc6f70 panfrost: Mark packs as ALWAYS_INLINE 
As Intel does. These functions are written with the expectation that they will
be inlined away, allowing gcc's copy-prop and constant folding to eliminate the
template struct and any unused fields.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21848>
 2023-03-30 23:21:59 +00:00
Alyssa Rosenzweig
f8b29f47a0 panfrost: Don't redundantly call emit_const_buf 
On Valhall, we were calling emit_const_buf in two places:

1. The main "handle dirty flags" code shared with Bifrost
2. A Valhall-specific shader environment emitter

The latter was not dirty tracked, and the former was not used. That meant we
were calling emit_const_buf way too much. It's not a cheap routine, either.

Instead, use the results from the dirty tracked function in the shader
environment emitter, to avoid the redundant call and get the expected dirty
tracking.

In a Dolphin trace I'm looking at, fps increases 27->33.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21848>
 2023-03-30 23:21:59 +00:00
Alyssa Rosenzweig
6ba62be633 panfrost: Print perf debug on seqnum overflow 
Another unexpected source of flushes.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21830>
 2023-03-30 22:53:16 +00:00
Alyssa Rosenzweig
9d3e01ddef panfrost: Print perf debug when flushing everything 
..Even if the only batch is the one that's currently bound.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21830>
 2023-03-30 22:53:16 +00:00
Mike Blumenkrantz
70b7c24206 zink: stop caching vertex states 
I tried to be too clever and ended up wasting cpu cycles. it's
much, much, much, much faster to just generate this one struct array
every time than it is to do set lookups with thousands of members

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22116>
 2023-03-30 22:28:38 +00:00
Mike Blumenkrantz
91ddfe55b5 zink: use fast popcnt for vstate draws 
also delete some unused stubs for no dynamic vertex input since I'm never
gonna implement that path

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22116>
 2023-03-30 22:28:38 +00:00
Mike Blumenkrantz
11a61ab424 zink: don't swizzle velems state for vstate draws 
this isn't ever used, so don't touch it

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22116>
 2023-03-30 22:28:38 +00:00
Mike Blumenkrantz
f676704fca zink: explicitly pass null velems when creating pipelines with dynamic vinput 
this may or may not be a usable pointer, and it's not being read, so
don't pass it at all

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22116>
 2023-03-30 22:28:38 +00:00
Mike Blumenkrantz
1ead8f7375 zink: add another vstate draw template for popcnt presence 
matching radeonsi

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22116>
 2023-03-30 22:28:38 +00:00
Mike Blumenkrantz
41983630c4 zink: bind vertex state directly from draw hook 
this is more streamlined and readable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22116>
 2023-03-30 22:28:38 +00:00
Mike Blumenkrantz
837168db20 zink: use search_or_add for masking vstate 
this should be a significant perf boost instead of multiple lookups

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22116>
 2023-03-30 22:28:38 +00:00
Mike Blumenkrantz
4be5caba67 zink: flag vertex buffers for rebind after vstate draws 
vstate draws bind their own vertex buffers unrelated to the bound
gallium buffers, so any draw occurring after a vstate draw must
rebind vertex buffers to ensure the correct ones are bound

cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22116>
 2023-03-30 22:28:38 +00:00
Mike Blumenkrantz
6c3b5921b2 zink: omit VkPipelineVertexInputStateCreateInfo with dynamic vinput 
this should never be used/needed

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22116>
 2023-03-30 22:28:38 +00:00
Konstantin Seurer
f6147051e2 radv: Stop counting user SGPRS separately 
Renames radv_declare_shader_args to declare_shader_args and runs it
twice to first gather the user SGPR count without push constants and
descriptor sets.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22119>
 2023-03-30 21:52:03 +00:00
Konstantin Seurer
0c915ba501 radv: Set user SGPR locations when declaring args 
Merge shader arg declaration with setting up the user data locations.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22119>
 2023-03-30 21:52:03 +00:00
Harri Nieminen
fd767a4517 bin: Fix typos 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22189>
 2023-03-30 21:37:00 +00:00
Mike Blumenkrantz
db582e5e7d zink: block resolves where src extents > dst extents 
vulkan resolves only provide "extents" instead of src and dst regions like
GL, which means vk resolves can't be used to downscale images, as such
operations will instead just crop the image

fixes #8655

cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22195>
 2023-03-30 21:13:40 +00:00
Samuel Pitoiset
373c6346f5 radv: add push constant state to the cmdbuf state 
Push constants are handled per bind point internally. Using a separate
structure in the cmdbuf state would allow us to update it easily
without relying on bound pipelines.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22209>
 2023-03-30 20:41:23 +00:00
Samuel Pitoiset
a0baefa033 radv: copy need_indirect_descriptor_sets to radv_cmd_state 
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22209>
 2023-03-30 20:41:23 +00:00
Samuel Pitoiset
eeefe18f05 radv: add a helper to convert a VkPipelineBindPoint 
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22209>
 2023-03-30 20:41:23 +00:00
Rob Clark
88f3676019 freedreno: Optimize repeated finishes 
Sometimes apps (glances at stk) spin on a syncobj with very short
timeouts.  But ensuring the fence is flushed all the way through to
the kernel (including handling TC unflushed fences) only needs to
be done once.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22098>
 2023-03-30 19:42:01 +00:00
Rob Clark
8416bc1c60 freedreno/drm: Disable threaded-submit for msm 
We've had drm/sched support on the kernel side for more than a year and
a half.  This makes submit ioctl async by handling fence waits from the
sched's kthread, which is what threaded submit was originally working
around.  For now, threaded submit is only used for virtgpu, which does
not (yet?) have drm/sched support.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22098>
 2023-03-30 19:42:01 +00:00
Rob Clark
a16533c43e freedreno/drm: Make threaded-submit optional 
We've had gpu-sched support in the kernel for a while now, so our fence
waits are not synchronous in the ioctl path.  The only reason this path
still exists is that virtgpu does not have gpu-sched.  So lets disable
it on msm.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22098>
 2023-03-30 19:42:01 +00:00
Rob Clark
cacbbfd6a8 mesa: Add a few more function traces 
Sprinkle around a few more traces that were useful in locating fence
waits.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22098>
 2023-03-30 19:42:01 +00:00
Rob Clark
c2194552e7 freedreno/drm: Stop cleanup at first active BO 
Buffers are added to the deferred freelist at the tail.  And frequently
the last reference is dropped immediately after the submit.  So almost
always, once we see a still-busy BO, the remaining in the list will also
still be busy.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22098>
 2023-03-30 19:42:01 +00:00
Rob Clark
712c26e2b6 freedreno/drm: Fast path for idle check 
If already idle, no need to cleanup_fences() (and take related lock).

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22098>
 2023-03-30 19:42:00 +00:00
Mike Blumenkrantz
77c7198d76 zink: fix quads emulation gs with array variables 
this was broken for e.g., gl_ClipDistance, which uses explicit array
types and therefore cannot be directly read/written

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22208>
 2023-03-30 19:21:52 +00:00
Rob Clark
f9a074dd55 dri2/android: Bypass throttling 
The android window system (SurfaceFlinger, et al) already does it's own
throttling.  Trying to do this also in mesa's egl is counterproductive.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22197>
 2023-03-30 18:46:04 +00:00
Konstantin Seurer
816f434efc radv/rq: Rematerialize inv_dir before proceed 
Helps with register pressure.

Quake II RTX:
Totals from 7 (14.29% of 49) affected shaders:

VGPRs: 688 -> 672 (-2.33%)
CodeSize: 167496 -> 167560 (+0.04%); split: -0.01%, +0.05%
MaxWaves: 70 -> 72 (+2.86%)
Instrs: 31716 -> 31760 (+0.14%); split: -0.02%, +0.16%
Latency: 385343 -> 386040 (+0.18%); split: -0.01%, +0.19%
InvThroughput: 78878 -> 78045 (-1.06%); split: -1.22%, +0.17%
VClause: 596 -> 600 (+0.67%)
Copies: 4774 -> 4747 (-0.57%); split: -0.98%, +0.42%
PreVGPRs: 617 -> 592 (-4.05%)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20469>
 2023-03-30 18:15:11 +00:00
Mike Blumenkrantz
9c73312248 zink: set src access when rebinding buffers, unset unordered_* 
this ensures that the buffer is marked active and prevents promotion
in cases where reordering would break rendering

unordered_read prohibits write reordering for buffers, so setting
this flag must be done when the buffer is actually used, ideally as
late as possible

setting it at the time of (re)bind catches all the buffer rebind cases
which might otherwise erroneously permit reordering

fixes #8381

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22205>
 2023-03-30 17:54:11 +00:00
Danylo Piliaiev
2cc9364c20 tu/drm: Support cached non-coherent memory 
Requires some hand rolled assembly:
- DC CVAC / DC CIVAC for aarch64
- DCCMVAC / DCCIMVAC for arm32, unfortunately it seems that it is
  illegal to call them from userspace.
- clflush for x86-64

We handle x86-64 case because Turnip may run in x86-64 guest
e.g. in FEX-Emu or Box64.

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20550>
 2023-03-30 15:50:47 +00:00
Danylo Piliaiev
5a59410962 turnip: add cached and cached-coherent memory types 
vkd3d requires cached memory type.

MSM backend doesn't have a special ioctl for memory
flushing/invalidation, we'd have to use cvac and civac
arm assembly instructions (would be done in following commit).

KGSL has an the ioctl for this, which is used in this commit.

Note, CTS tests doesn't seem good at testing flushing and
invalidating, the ones I found passed on KGSL with both
functions being no-op.

Based on the old patch from Jonathan Marek.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7636

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20550>
 2023-03-30 15:50:47 +00:00
Erik Faye-Lund
bd816084c6 zink: enable spir-v 1.6 for vulkan 1.3 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18274>
 2023-03-30 14:06:54 +00:00
Erik Faye-Lund
99bd1eaf3d zink: use spir-v 1.6 local-size when needed 
The WorkgroupSize built-in is deprecated in SPIR-V 1.6, so let's switch
to using LocalSizeId instead, like the spec recommends.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18274>
 2023-03-30 14:06:54 +00:00
Erik Faye-Lund
da895596da zink: use demote from spir-v 1.6 when possible 
With SPIR-V 1.6, we don't need to enable the extension for demote any
more.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18274>
 2023-03-30 14:06:54 +00:00
Erik Faye-Lund
636bfc6358 zink: emit terminate for spir-v 1.6 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18274>
 2023-03-30 14:06:54 +00:00
Mike Blumenkrantz
67462ed014 zink: add spirv builder function for terminate 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18274>
 2023-03-30 14:06:54 +00:00
Mike Blumenkrantz
db1371cce1 llvmpipe: fix handling of unused color attachments 
if an attachment doesn't have blending or color output from the shader,
nothing should touch the attachment

this is consistent with vulkan spec and needed for upcoming cts coverage

cc: mesa-stable

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22135>
 2023-03-30 12:04:22 +00:00
Mike Blumenkrantz
6a5dcd2776 llvmpipe: fix linear fs analysis with nonzero fs outputs 
linear fs only works with output 0

cc: mesa-stable

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22135>
 2023-03-30 12:04:22 +00:00
Samuel Pitoiset
15f1d5cc8f radv: copy ia_multi_vgt_param to the cmdbuf state 
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22194>
 2023-03-30 11:43:06 +00:00
Samuel Pitoiset
0daffade14 radv: copy uses_{drawid,baseinstance} to the cmdbuf state 
As well as the vertex user sgpr info. This also needs to be copied
for merged shaders (ie. VS+TCS).

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22194>
 2023-03-30 11:43:06 +00:00
Samuel Pitoiset
8fd915098d radv: copy rast_prim to the cmdbuf state 
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22194>
 2023-03-30 11:43:06 +00:00
Samuel Pitoiset
6ee8d33ef4 radv: add a helper that returns the current rasterized primitive 
For tess/geom/mesh the rasterized primitive is part of the shader
info and primitive topology should be ignored.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22194>
 2023-03-30 11:43:06 +00:00
Samuel Pitoiset
b0a007583d radv: copy custom blend mode to the cmdbuf state 
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22194>
 2023-03-30 11:43:06 +00:00
Samuel Pitoiset
bc39fa9500 radv: rework emitting inner coverage when a fragment shader is bound 
To stop relying on the pipeline. The state is now flagged when a new
fragment shader with a different inner coverage setting is bound.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22194>
 2023-03-30 11:43:06 +00:00