Alyssa Rosenzweig
65789854c5
agx: lower votes to ballots
...
not optimal but passes the tests.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
c037fa376d
agx: wire up some ballots
...
lots of possibly optimizations but that's for another day
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
93d15c2bd2
agx: optimize fcmp like fcmpsel
...
total instructions in shared programs: 1810310 -> 1807236 (-0.17%)
instructions in affected programs: 269043 -> 265969 (-1.14%)
helped: 1553
HURT: 17
Instructions are helped.
total bytes in shared programs: 11851492 -> 11834830 (-0.14%)
bytes in affected programs: 1786394 -> 1769732 (-0.93%)
helped: 1540
HURT: 37
Bytes are helped.
total halfregs in shared programs: 474698 -> 474333 (-0.08%)
halfregs in affected programs: 6239 -> 5874 (-5.85%)
helped: 146
HURT: 22
Inconclusive result (%-change mean confidence interval includes 0).
total threads in shared programs: 19012032 -> 19012096 (<.01%)
threads in affected programs: 768 -> 832 (8.33%)
helped: 1
HURT: 0
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
210c6931ff
agx: fix early-z + discard together
...
don't trigger tests twice.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
eab145e223
agx: use more mem->tex barriers even on g13g
...
fixes KHR-GLES31.core.texture_buffer.texture_buffer_atomic_functions
with bigger subgroups on g13g.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
2feea36b37
asahi: allow multiple compute dispatches in a batch
...
glmark2 terrain 50% faster. yes, really.
requires setting more barriers to fix flakes.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
c455a27f54
agx: allow bindful arrays if not clamping
...
for blit shaders mostly.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
611fcfc5ca
asahi: do not use compression blits for uncompressed levels
...
pointless, it's much faster to just twiddle on the cpu. use the helpers to
detect compression per-level the right way.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
11a180c6e1
ail: add ail_is_level_twiddled_uncompressed
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
8dcdf5040b
ail: use is_level_compressed
...
to get unit testing coverage. identified several bugs!
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
67d92e0bb6
ail: add is_level_compressed query
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
2728995f12
asahi: fix eMRT + background load interaction
...
for when some render targets are spilled only. prevents regression in
KHR-GLES31.core.draw_buffers_indexed.color_masks
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26963 >
2024-01-10 08:44:37 -04:00
Alyssa Rosenzweig
e7f3112eb9
asahi: Implement lod queries
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26861 >
2024-01-04 01:51:07 +00:00
Daniel Schürmann
bdbf873b0f
nir: remove redundant passes from nir_opt_if()
...
These are now covered by nir_opt_loop():
- opt_if_loop_last_continue()
- opt_merge_breaks()
- opt_if_loop_terminator()
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24940 >
2024-01-03 20:48:05 +00:00
Yonggang Luo
0210b554d6
treewide: Replace the include of nir_types.h with glsl_types.h
...
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26753 >
2023-12-30 15:08:11 +00:00
Alyssa Rosenzweig
d11c9f9836
asahi: clamp draw count for mdi
...
spec req. KHR-GL43.indirect_parameters_tests.MultiDrawArraysIndirectCount
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
982e644d18
agx: don't produce split of immediate
...
can't be consumed, affects KHR-GL42.shader_image_load_store.basic-allTargets-load-ms
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
6e2c71dc3b
agx: Lower 64-bit I/O to 32-bit
...
Fixes KHR-GL42.vertex_attrib_64bit.*
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
d6474be46b
agx: fix fp64 lowering options
...
we have no fp64 whatsoever.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
8d451fe9f2
asahi: fix integer RT clamping
...
fixes the rest of KHR-GL33.packed_pixels.*
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
6d05bf8761
asahi: fix dupe rgb65 formats
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
b68caa13e4
asahi: add more BGR formats
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
c89f0becf7
asahi: Implement ARB_cull_distance
...
Passes KHR-GL33.cull_distance.* and the piglits.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
ba077d899e
asahi: handle compact clip/cull in gs component gather
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
f02190b688
asahi: prepare gs copy shaders for compact clip/cull
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
0dc004888c
asahi: handle some components/offsets in GS lowering
...
occurs for compact clip/cull xfb. not hard to deal with, do so instead of
getting caught up in assertions.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
17f348f73a
asahi: Implement ARB_viewport_array
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
af9241ce92
asahi: add XML for multiple viewports
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
bb59c787ec
agx: report if we have a nonzero viewport
...
so driver can optimize.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
bc7afbf23d
agx: allocate varying slot if writing viewport only
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
643acacd1b
agx: note that sample_mask runs occlusion queries
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
79e58e1d77
agx: handle force early-z + discard
...
fixes ./arb_shader_image_load_store-early-z. experimentally, an opaque pass type
works too but better match what the blob does.
also, I now have proof that sample_mask triggers occlusion query updates because
if you run it multiple times, you get >1 hits per fragment in a counting query
:p
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
3987c8a35f
agx: remove spurious z/s writes in force early-z shaders
...
fixes crash in arb_shader_image_load_store-early-z
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
c43c90a5fa
asahi: rewrite pointsize handling
...
In the wise words of Mike Blumenkrantz, "I hate gl_PointSize and so can you".
The mesa/st lowering won't mesh well with vertex shader epilogues, and it falls
over in various circumstances. I am too tired to go against the grain, so let's
just pretend to be a normal gallium driver and trust in the rasterizer CSO,
lowering point size internally. This properly handles transform feedback without
any hacks, both GL and GLES behaviours, etc.
Fixes:
KHR-GL31.transform_feedback.capture_vertex_separate_test
gl-2.0-large-point-fs
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
23f216d6e7
asahi: Lower edge flags
...
With the common geometry shader based lowering added for zink. Fixes edge flag
related piglits.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
f4a648c607
asahi: Unroll GS/XFB primitive restart on the GPU
...
..and fix bugs versus the CPU unroll while we're at it. CPU based unrolling is
invalid in Vulkan, but this slow-as-dogs GPU unroll is ok.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
15957219ad
agx: rework libagx linking a bit
...
for correctness with generic ptrs, avoids splat with the next patch.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
4a4c6d1bc6
agx: Avoid scratch mem with tri strip w/ adjacency
...
with restart unroll kernel where nir fails to optimize this away without a
little help from us.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
7f9ef5b176
agx: fix VARYING_SLOT_COL0 getting flatshaded
...
it's a bit mask.
didn't fix the piglit i was debugging :-(
but did fix the shadow glitching out in neverball.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
6437578435
asahi: add xml for flatshading fans
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
403a20a9f2
asahi: rework cf binding xml
...
for flat shading of fans.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
2fec6bf951
asahi: fix tri fan enum
...
from tracing macos. fixes piglits:
spec@!opengl 1.1@gl-1.1-drawarrays-vertex-count 100000 varray gl_triangle_fan
spec@!opengl 1.1@gl-1.1-drawarrays-vertex-count 100000 vbo gl_triangle_fan
spec@!opengl 1.4@triangle-rasterization-overdraw
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
6ae72a2f2a
asahi: Support L/A/I formats for texture buffers
...
mesa/st doesn't know how to emulate them. Fixes
arb_texture_buffer_object-formats.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
5b33c74626
agxdecode: fix stack smash with border colour
...
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
0fa7252d8a
asahi: Implement multidraw indirect
...
GS only for now (inserting a passthru GS if needed). This should be optimized
later, but it's ~correct.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
5987e47a29
asahi: rework GS input assembly
...
in prep for tessellation (which will share the IA lowering), and for multidraw
indirect (which greatly complicates IA lowering with geom/tess).
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
0805fef082
asahi: make UB easier to see
...
would've saved us
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Alyssa Rosenzweig
09e46aa168
asahi: add agx_push macro
...
Thanks, Ella!
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 12:08:39 -04:00
Asahi Lina
25d185a501
ail: Fix tile size & strides for compressed textures
...
Compressed textures have two additional quirks that affect the tiling
code (but not the mip offsets): they get extra stride padding in some
cases for the large miptree, and the tile size is based on the POT size
and not the real size for the small miptree.
Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 10:56:45 -04:00
Asahi Lina
be3890a898
ail: Add explicit specification of mip level strides
...
For compressed textures, mip levels > 0 can have additional stride
padding. This (in some cases) affects the tile stride calculation, so it
cannot be implicitly represented with the existing members.
Add an explicit array containing the stride, in elements, of each
miptree level. The tiling code uses this instead of the minified and
element-aligned width when computing tile addressing.
This commit should be a functional no-op.
Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26614 >
2023-12-09 10:56:45 -04:00