Commit graph

549 commits

Author SHA1 Message Date
Alyssa Rosenzweig
ad2f096a8b agx/lower_vbo: dce as we go
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:19 +00:00
Alyssa Rosenzweig
f0e1ccc8d4 asahi: rewrite varying linking
Lower store_output to store_uvs_agx + math. Link UVS indices at draw-time
instead of compile-time to get efficient separate shaders. Also picks up varying
compaction along the way.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:19 +00:00
Alyssa Rosenzweig
a2265ff588 asahi: drop =varyings debug
hasn't been relevant.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:19 +00:00
Alyssa Rosenzweig
9c2df9e814 asahi: add agx_push_packed
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:19 +00:00
Alyssa Rosenzweig
42a43bbdad libagx: parallelize prefix sum over 1024 threads
instead of 32. small % win on a synthetic GS benchmark

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:18 +00:00
Alyssa Rosenzweig
3319d4fdba libagx: deal with silly NIR
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:18 +00:00
Alyssa Rosenzweig
13ecef56d0 libagx: accelerate prim restart unroll across wg
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:18 +00:00
Alyssa Rosenzweig
890a96e2a7 libagx: accelerate restart unroll across a subgroup
before implementing hard stream compaction algorithms, let's do the easy
accelleration.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:18 +00:00
Alyssa Rosenzweig
b51282500d libagx: polyfill glsl ballot()
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:18 +00:00
Alyssa Rosenzweig
2e28998017 libagx: improve static assert message
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:18 +00:00
Alyssa Rosenzweig
72ef80dfc8 asahi: stop merging VS and TCS
unfortunately, shader stage merging is bogus when coherent images are used, so
we need an unmerged path. i'd rather not maintain two paths, so let's just
stop merging. as a bonus this makes ESO a lot easier, and lets us reuse the same
VS for both VS->GS and VS->TCS.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:17 +00:00
Alyssa Rosenzweig
f2a1d38096 asahi: fix unit mismatch with unroll path
Fixes GS line/point expansion in Dolphin, as well as corruption in Blender when
the dancing ants show up

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:17 +00:00
Alyssa Rosenzweig
e952189443 asahi: fix overread with samplers
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:17 +00:00
Alyssa Rosenzweig
007a440a33 asahi/lib: fix overread with stateful
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:17 +00:00
Alyssa Rosenzweig
0a8d0217c9 asahi: move more code out of agx_preprocess_nir
we need to gather tex masks  / lower mediump io before lowering textures for our
detection to work ... also we want driver-side i/o soon lowering for Marek's
thing anyway. do some code motion / pass reordering to make this doable.

in doing so, we get rid of agx_uncompiled_shader_info which is probably good.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28483>
2024-03-30 00:26:17 +00:00
Alyssa Rosenzweig
da59672e2b asahi: derive ht
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28056>
2024-03-12 19:41:49 +00:00
Alyssa Rosenzweig
a3f647fde1 asahi: precompile helper program
1. avoids cluttering shaders,internal print
2. reduces screen create overhead. this cuts average "glxinfo" runtime by about
   25%, and seems to shave maybe 0.5% off the CTS job.
3. enforces helper program builds successfully as a build-time assertion
4. reduces the libagx generated blob for now, since g13 binaries are more
   compact than serialized NIR

Closes https://gitlab.freedesktop.org/asahi/mesa/-/issues/37

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
4a3b905bb8 agx: move texture lowering into lib
This is a bit annoying, but it gets rid of the libagx dep which is preventing
precompiling helper programs.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
1f465d72dd asahi: split out genxml/ directory
src/asahi/lib is creating dependency hell, let's start splitting it up so we can
ease some of the clc related pains.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
35c671ecef asahi: clean up format table renderability
infer renderability from whether we specify a renderable format. nfc now that
rgb9e5 is fixed.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
3333c2cdeb asahi/lib: use #pragma once
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
b39de8bb4a asahi: Implement ARB_texture_barrier by decompression
Your perf goes not stonks, sorry. Passes
KHR-GL40.texture_barrier_ARB.*.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
2dc1189a9c agx: rm unused backend nr_samples
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
27ddcea2b3 agx: call agx_nir_lower_sample_mask earlier
A given sample needs depth testing to happen before writing its colour, which
requires shuffling pass order.

To do so, merge agx_nir_opt_ixor_bcsel into regular late alg pass
now that we can, which is actually a small shader-db win.

Closes https://gitlab.freedesktop.org/asahi/mesa/-/issues/30

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
0eb9b62199 asahi: rm ia key
fully dynamic in GS now.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
beb7ddba2b asahi: always support ARB_clip_control
perf impact has been neglible.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
6673924b7e asahi: make gs topology dynamic
even with shobjs, we know the class of topology statically, so we just need to
select between the (up to) 3 compatible topologies, and luckily there are common
subexpressions we can factor out when calculating all 3 at once.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
049434f89c asahi: make provoking vertex dynamic
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
dbb2b25ef9 asahi: eliminate tri fan %
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
43dd8b87cb asahi: simplify expressions involving xfb
post-invert cleanup

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
b44630aa94 asahi: rm dead
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
86782156e7 asahi: invert geometry shaders
instead of dumping GS outputs to memory and using a GS copy vertex shader,
invert the GS to get a hardware vertex shader to use directly. theoretically,
reduces mem bandwidth at the cost of repeated work if we don't optimze well.
perhaps more importantly, it should reduce heap usage.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:32 +00:00
Alyssa Rosenzweig
1ac3e48fcc asahi: fix vertex out size calc
expression underallocates if var31 is actually used!

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
141e084f55 asahi: use load_instance_id in gs lowering
it will lower differently in the prepass vs the rast shader

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
17896f1699 nir: rm load_vert_id_in_prim_agx
now unused since we separate vs/gs

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
e13d001160 asahi: rm arrayed output lowering
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
d96fbd4618 asahi: separate GS from VS
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
0b30777028 asahi: add geometry parameters for separable GS
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
cdd1b012e9 asahi: simplify IA mode handling
we'll solve the IA variants problem a different way, for now make more
straightforward.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
e1783fc083 libagx: generalize vertex_id_for_topology
so we can use it in separable GS

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
c9f55166dc libagx: static assert some sizes
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
aaa609f030 libagx: use real PACKED macro
hopefully end these alignment bugs for good

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
fc929d9318 libagx: use native static_assert on host
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
bfaa3691e1 asahi: make GS flatshade_first more dynamic
baby step

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
4036c710bd libagx: fix buggy align macro
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
784323cf0f asahi: rip out existing MDI+GS implementation
I don't love deleting working code without a working replacement. But this impl
is deeply questionable as-is, and will only balloon in complexity for shader
objects. We need to go back to the drawing board with MDI and lower to real
draws, possibly with control stream looping.

Good news is that this is GL so we can prototype all this.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
bb37b072a5 asahi,agx: use intrinsics pass
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:31 +00:00
Alyssa Rosenzweig
8ab443d549 asahi: enable compblit behind dbg flag
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:30 +00:00
Alyssa Rosenzweig
7eac18b1bc asahi: force inline ppp update logic
benefits a lot from constant folding.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:30 +00:00
Alyssa Rosenzweig
c6c8262ce1 asahi: implement pipeline stats as a checkbox
real impl is blocked on uapi to plumb thru hw perf counters.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
2024-02-14 21:02:30 +00:00