fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-16 18:28:05 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1
167796 commits 66 branches 935 tags 1 GiB
Commit graph

249 commits

Author SHA1 Message Date
Georg Lehmann
77afe7d960 aco: treat VINTERP_INREG as VALU 
It's just v_fma with fixed DPP8 and builtin s_waitcnt_expcnt, so it can mostly
be handled as a pure VALU instruction.

Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21023>
 2023-03-07 11:53:23 +00:00
Rhys Perry
75d9a4a6ce aco: always update orig_names in get_reg_phi() 
No idea why this wasn't done if pc.first was a renamed temporary.

Fixes navi10 RA validation error with
dEQP-VK.binding_model.descriptor_buffer.multiple.graphics_geom_buffers1_sets3_imm_samplers

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8349
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21501>
 2023-02-27 15:10:22 +00:00
Rhys Perry
20e670d060 aco/ra: don't swap create_vector operand with definition blocker for SGPRs 
There is no SGPR swap instruction, we always need 3 XORs.

fossil-db (navi21):
Totals from 76 (0.06% of 135636) affected shaders:
Instrs: 58400 -> 58347 (-0.09%); split: -0.10%, +0.01%
CodeSize: 312580 -> 312368 (-0.07%); split: -0.08%, +0.01%
Latency: 843333 -> 843180 (-0.02%); split: -0.02%, +0.00%
InvThroughput: 126431 -> 126412 (-0.02%)
Copies: 4008 -> 3955 (-1.32%); split: -1.47%, +0.15%

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/20240>
 2022-12-09 15:58:43 +00:00
Rhys Perry
a05dd58309 aco/ra: don't swap p_create_vector operand with definition blocker for scc 
SCC is 1-bit, and we can't copy a 32-bit value into it.

Fixes dEQP-VK.spirv_assembly.type.scalar.i32.iequal_tesse with
ACO_DEBUG=noopt.

No fossil-db changes.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fixes: 9476986e6f ("aco/ra: special-case get_reg_for_create_vector_copy()")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20240>
 2022-12-09 15:58:43 +00:00
Tatsuyuki Ishi
327c906424 aco: Migrate RA to use std::optional 
The use of std::optional simplifies expressions and would be useful for some
upcoming RA tweaks.

C++17 has been available since the merge of rusticl and should be safe to use as
far as packaging is concerned.

A few style choices are:
- Testing for emptiness uses implicit bool conversion.
- Constructing an empty value uses {}.
- Constructing a filled value uses the implicit conversion constructor.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20125>
 2022-12-08 12:08:01 +00:00
Rhys Perry
50073d6135 aco/gfx11: increase gfx1100/gfx1101 physical vgprs 
https://reviews.llvm.org/D134522

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18825>
 2022-11-02 17:09:32 +00:00
Rhys Perry
6113ee650a aco/gfx11: fix FS input loads in quad-divergent control flow 
This is not ideal and it would be great to somehow make it better some
day.

fossil-db (gfx1100):
Totals from 5208 (3.86% of 135032) affected shaders:
MaxWaves: 127058 -> 126962 (-0.08%); split: +0.01%, -0.09%
Instrs: 3983440 -> 4072736 (+2.24%); split: -0.00%, +2.24%
CodeSize: 21872468 -> 22230852 (+1.64%); split: -0.00%, +1.64%
VGPRs: 206688 -> 206984 (+0.14%); split: -0.05%, +0.20%
Latency: 37447383 -> 37491197 (+0.12%); split: -0.05%, +0.17%
InvThroughput: 6421955 -> 6422348 (+0.01%); split: -0.03%, +0.03%
VClause: 71579 -> 71545 (-0.05%); split: -0.09%, +0.04%
SClause: 148289 -> 147146 (-0.77%); split: -0.84%, +0.07%
Copies: 259011 -> 258084 (-0.36%); split: -0.61%, +0.25%
Branches: 101366 -> 101314 (-0.05%); split: -0.10%, +0.05%
PreSGPRs: 223482 -> 223460 (-0.01%); split: -0.21%, +0.20%
PreVGPRs: 184448 -> 184744 (+0.16%)

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/19370>
 2022-11-01 12:42:43 +00:00
Daniel Schürmann
6fcabb3b02 aco/ra: only rename fixed Operands if the copy-location matches 
With epilogs, the same Operand can be fixed to multiple registers.

Fixes: 061b8bfd29 ('aco/ra: rework fixed operands')
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18485>
 2022-10-24 13:12:34 +00:00
Rhys Perry
aadb7aef01 aco: add VINTERP instruction format 
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/17333>
 2022-09-26 14:49:56 +00:00
Yonggang Luo
091249dff4 aco: Fixes compiling error about char8_t with c++20 
The error is:
../mesa/src/amd/compiler/aco_register_allocation.cpp:382:7: error: no matching function for call to 'printf'
      printf(u8"☐");

Fixes: 209a89e51d ("aco: Convert to use u8 literal for Unicode character to fixes msvc warning")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7318

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Tested-by: Marcus Seyfarth <m.seyfarth@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18796>
 2022-09-26 13:28:33 +00:00
Yonggang Luo
209a89e51d aco: Convert to use u8 literal for Unicode character to fixes msvc warning 
Warning:
aco_register_allocation.cpp(383): warning C4819: The file contains a character that cannot be represented in the current code page (0). Save the file in Unicode format to prevent data loss

This warning was treated as error with compiling with msvc
u8 is belongs to c11 standard so it's safe to use it

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18682>
 2022-09-20 18:40:50 +00:00
Rhys Perry
6df5ff7f19 aco: DCE ra_ctx::defs_done 
This was used to distinguish definitions fixed before and during RA, but
it seems it isn't used anymore.

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/18547>
 2022-09-20 12:24:03 +00:00
Rhys Perry
061b8bfd29 aco/ra: rework fixed operands 
This moves all fixed operands at once, so they don't interfere with one
another.

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/17493>
 2022-09-01 11:22:46 +00:00
Rhys Perry
ec867ef0e7 aco/ra: remove bounds parameter from get_regs_for_copies() 
I don't think it makes sense for this to be anything but get_reg_bounds(),
and this change makes this function usuable with a mix of SGPRs and VGPRs.

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/17493>
 2022-09-01 11:22:46 +00:00
Rhys Perry
efcbccaf0e aco/ra: handle empty def_reg interval in get_regs_for_copies 
If def_reg is empty, then def_reg.lo() may be lower than bounds.lo() if
we're moving VGPRs and info.bounds will be invalid.

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/17493>
 2022-09-01 11:22:46 +00:00
Daniel Schürmann
3d6ea4f666 aco: use std::vector::reserve() more often 
This removes the majority of vector re-allocations.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18105>
 2022-08-30 16:03:26 +00:00
Georg Lehmann
e06773281b aco/ra: Optimize some SOP2 instructions with literal to SOPK. 
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15999>
 2022-07-06 09:54:54 +00:00
Georg Lehmann
87b4f3daa1 aco/ra: Move mac encoding optimization to its own function. 
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15999>
 2022-07-06 09:54:54 +00:00
Georg Lehmann
c9490436b6 aco/ra: Static assert that changing instruction type to VOP2 is valid. 
It's not obvious that this is correct.

Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15999>
 2022-07-06 09:54:54 +00:00
Rhys Perry
84f04fd080 aco/ra: update register file when updating phi definition 
update_renames() fills in the wrong temp id.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fixes: 302cb5c900 ("aco/ra: remove some redundant code")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17295>
 2022-06-30 19:30:55 +00:00
Rhys Perry
cd2b4c319a aco/ra: fix usage of invalidated iterator 
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fixes: 58bd9a379e ("aco/ra: fix live-range splits of phi definitions")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16454>
 2022-05-30 16:41:32 +00:00
Marek Olšák
39800f0fa3 amd: change chip_class naming to "enum amd_gfx_level gfx_level" 
This aligns the naming with PAL.

Acked-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Pierre-Eric Pellou-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16469>
 2022-05-13 14:56:22 -04:00
Daniel Schürmann
5e6e47ecea aco/ra: improve split_vector register assignment if the operand is not killed 
This allows for more coalescing when lowering the copies.

Totals from 44801 (33.21% of 134913) affected shaders: (GFX10.3)
VGPRs: 1513264 -> 1513248 (-0.00%)
CodeSize: 113354240 -> 113172872 (-0.16%); split: -0.16%, +0.00%
Instrs: 21648793 -> 21603397 (-0.21%); split: -0.21%, +0.00%
Latency: 95762290 -> 95757403 (-0.01%); split: -0.01%, +0.00%
InvThroughput: 15427354 -> 15427341 (-0.00%); split: -0.00%, +0.00%
Copies: 2065330 -> 2019933 (-2.20%); split: -2.20%, +0.00%

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15414>
 2022-05-06 14:52:07 +00:00
Rhys Perry
2f0bb39e16 aco: ensure that definitions fixed to operands have matching regclasses 
If the operand is not killed, the definition needs to be large enough so
that the new location for the operand does not intersect with the old
location.

Fixes with zink:
KHR-GL45.shader_image_load_store.basic-allTargets-atomicCS
KHR-GL45.shader_image_load_store.basic-allTargets-atomicGS
KHR-GL45.shader_image_load_store.basic-allTargets-atomicVS

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Gitlab: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6276
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16106>
 2022-05-05 19:56:48 +01:00
Rhys Perry
1b639a0ce5 aco/ra: fix vgpr_limit 
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Fixes: b98a4d4dd7 ("aco: refactor GPR limit calculation")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16297>
 2022-05-04 11:12:13 +00:00
Daniel Schürmann
58bd9a379e aco/ra: fix live-range splits of phi definitions 
It could happen that at the time of a live-range split,
a phi was not yet placed in the new instruction vector,
and thus, instead of renamed, a new phi was created.

Fixes: dEQP-VK.subgroups.ballot_broadcast.compute.subgroupbroadcast_i8vec2
Cc: mesa-stable

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16248>
 2022-05-03 09:54:30 +00:00
Daniel Schürmann
415a3820fc aco/ra: omit VCC affinity on VOPC_SDWA for GFX9+ 
VOPC_SDWA can also use arbitrary SGPR pairs on GFX9+.

Totals from 5607 (4.16% of 134913) affected shaders: (GFX10.3)
CodeSize: 42470760 -> 42452988 (-0.04%)
Instrs: 7943174 -> 7942883 (-0.00%)
Latency: 102887029 -> 102886305 (-0.00%); split: -0.00%, +0.00%
InvThroughput: 20454456 -> 20454338 (-0.00%); split: -0.00%, +0.00%
Copies: 376818 -> 376865 (+0.01%); split: -0.00%, +0.01%

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15408>
 2022-04-13 21:52:43 +00:00
Daniel Schürmann
6ebc61d71b aco/ra: create VCC-affinities during RA 
instead of using register hints.

Totals from 88367 (65.50% of 134913) affected shaders: (GFX10.3)
CodeSize: 322492184 -> 322252912 (-0.07%); split: -0.08%, +0.01%
Instrs: 60615809 -> 60541260 (-0.12%); split: -0.12%, +0.00%
Latency: 557067980 -> 557009210 (-0.01%); split: -0.01%, +0.00%
InvThroughput: 109676757 -> 109674804 (-0.00%); split: -0.00%, +0.00%
SClause: 1939703 -> 1939924 (+0.01%); split: -0.01%, +0.02%
Copies: 4557567 -> 4487530 (-1.54%); split: -1.54%, +0.00%
Branches: 1941123 -> 1937453 (-0.19%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15408>
 2022-04-13 21:52:43 +00:00
Daniel Schürmann
44fb9ba84a aco/ra: only use VCC if program->needs_vcc == true 
A future commit will make VCC register assignment independent
from register hints. Up to GFX9, VCC can alternatively be used
as regular SGPR, so prevent overlap.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15408>
 2022-04-13 21:52:43 +00:00
Daniel Schürmann
db8c401f71 aco/ra: fix stride check on subdword parallelcopies for create_vector 
On GFX6/7, info.rc is in full dwords.

Fixes: 9476986e6f ('aco/ra: special-case get_reg_for_create_vector_copy()')
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15393>
 2022-03-31 08:55:07 +00:00
Daniel Schürmann
007cb02db9 aco: use branch definition as scratch register for SSA lowering 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15505>
 2022-03-28 07:36:46 +00:00
Daniel Schürmann
2d1e6b756e aco: remove 'high' parameter from can_use_opsel() 
No fossil-db changes.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15551>
 2022-03-25 22:02:50 +00:00
Daniel Schürmann
70aea6b41a aco/ra: refactor collect_vars() to return a sorted vector 
The vector of IDs is sorted with decreasing sizes,
and by increasing assigned registers.
This decouples register assingment from ssa IDs.

Totals from 12694 (9.41% of 134913) affected shaders: (GFX10.3)
VGPRs: 757864 -> 757848 (-0.00%); split: -0.00%, +0.00%
CodeSize: 72350540 -> 72348688 (-0.00%); split: -0.02%, +0.02%
MaxWaves: 237018 -> 237020 (+0.00%); split: +0.00%, -0.00%
Instrs: 13545494 -> 13544699 (-0.01%); split: -0.03%, +0.02%
Latency: 148539203 -> 148533292 (-0.00%); split: -0.01%, +0.00%
InvThroughput: 30319086 -> 30320382 (+0.00%); split: -0.01%, +0.01%
VClause: 326875 -> 327028 (+0.05%); split: -0.05%, +0.09%
SClause: 479833 -> 479837 (+0.00%); split: -0.00%, +0.00%
Copies: 862152 -> 860914 (-0.14%); split: -0.43%, +0.28%
Branches: 317775 -> 317777 (+0.00%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11526>
 2022-03-14 08:32:10 +00:00
Daniel Schürmann
61c36b6dc0 aco/ra: refactor find_vars() to return a vector 
instead of std::set<>

No fossil-db changes.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11526>
 2022-03-14 08:32:10 +00:00
Daniel Schürmann
9476986e6f aco/ra: special-case get_reg_for_create_vector_copy() 
This function implements separate handling for
p_create_vector during get_regs_for_copies().
This simplifies some code and lets more precisely select
swap instructions if possible.

Totals from 876 (0.65% of 134913) affected shaders: (GFX10.3)
VGPRs: 53312 -> 53336 (+0.05%)
CodeSize: 3792936 -> 3788160 (-0.13%); split: -0.15%, +0.03%
MaxWaves: 16084 -> 16078 (-0.04%)
Instrs: 707449 -> 706385 (-0.15%); split: -0.19%, +0.04%
Latency: 6288293 -> 6286677 (-0.03%); split: -0.03%, +0.01%
InvThroughput: 4264450 -> 4263671 (-0.02%); split: -0.02%, +0.00%
VClause: 18655 -> 18679 (+0.13%); split: -0.20%, +0.33%
Copies: 55397 -> 54353 (-1.88%); split: -2.45%, +0.57%
Branches: 12426 -> 12415 (-0.09%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11526>
 2022-03-14 08:32:10 +00:00
Daniel Schürmann
9181e8ceba aco/ra: count constant moves in get_reg_create_vector() 
Also implements a correct_pos_mask to keep track which
operands are already at the right target position.
To mitigate some regressions, call get_reg_impl() less often.

Totals from 1229 (0.91% of 134913) affected shaders: (GFX10.3)
VGPRs: 60216 -> 59848 (-0.61%)
CodeSize: 3716496 -> 3711268 (-0.14%); split: -0.19%, +0.05%
MaxWaves: 27952 -> 28004 (+0.19%)
Instrs: 685983 -> 685035 (-0.14%); split: -0.20%, +0.06%
Latency: 6727587 -> 6725340 (-0.03%); split: -0.06%, +0.02%
InvThroughput: 9289043 -> 9289866 (+0.01%); split: -0.02%, +0.03%
VClause: 17730 -> 17740 (+0.06%); split: -0.25%, +0.30%
Copies: 54352 -> 53420 (-1.71%); split: -2.46%, +0.75%
Branches: 12122 -> 12121 (-0.01%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11526>
 2022-03-14 08:32:10 +00:00
Daniel Schürmann
ccf4bcd162 aco/ra: don't immediately assign a register for p_branch 
These get now assigned after handling phis.

Totals from 564 (0.42% of 134913) affected shaders: (GFX10.3)
CodeSize: 5519744 -> 5515308 (-0.08%)
Instrs: 1063045 -> 1061936 (-0.10%)
Latency: 11880452 -> 11875904 (-0.04%)
InvThroughput: 2259933 -> 2259581 (-0.02%); split: -0.02%, +0.00%
Copies: 86908 -> 85799 (-1.28%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13432>
 2022-03-03 20:21:08 +00:00
Rhys Perry
d5349a99c2 aco/ra: fix register allocation of branch definitions 
fossil-db (Sienna Cichlid):
Totals from 704 (0.52% of 134913) affected shaders:
CodeSize: 7177288 -> 7182072 (+0.07%); split: -0.00%, +0.07%
Instrs: 1371781 -> 1372977 (+0.09%); split: -0.00%, +0.09%
Latency: 17993572 -> 18001344 (+0.04%); split: -0.00%, +0.04%
InvThroughput: 4198996 -> 4199569 (+0.01%); split: -0.00%, +0.01%
Copies: 122456 -> 123516 (+0.87%); split: -0.01%, +0.88%
Branches: 43815 -> 43818 (+0.01%); split: -0.02%, +0.03%

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/13432>
 2022-03-03 20:21:08 +00:00
Rhys Perry
608d48b787 aco/ra: add get_reg_phi() helper 
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/13432>
 2022-03-03 20:21:08 +00:00
Rhys Perry
ba44634e4d aco: fix v_mac_legacy_f32 
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fixes: f68797ead7 ("aco: create v_mac_legacy_f32/v_fmac_legacy_f32")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/5952
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14820>
 2022-02-01 15:08:04 +00:00
Rhys Perry
f68797ead7 aco: create v_mac_legacy_f32/v_fmac_legacy_f32 
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13436>
 2022-01-20 22:54:42 +00:00
Daniel Schürmann
d36a43598c aco/ra: fix get_reg_for_operand() in case of stride mismatches 
We have to clear the register file from the previous operand
as otherwise, there might be no space left.

Totals from 5 (0.00% of 134572) affected shaders: (GFX10.3)
CodeSize: 21144 -> 21000 (-0.68%); split: -0.72%, +0.04%
Instrs: 3738 -> 3720 (-0.48%); split: -0.51%, +0.03%
Latency: 517229 -> 516319 (-0.18%); split: -0.18%, +0.00%
InvThroughput: 49068 -> 48902 (-0.34%); split: -0.38%, +0.04%
Copies: 501 -> 483 (-3.59%); split: -3.79%, +0.20%

Cc: mesa-stable
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14279>
 2021-12-21 17:15:45 +00:00
Rhys Perry
c5f02a1cd3 aco: swap multiplication operands if needed to create v_fmac_f32/etc 
For v_pk_fma_f32 and v_fma_f32 from nir_op_ffma, we don't try to put
scalars in the first operand.

No fossil-db changes.

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/9805>
 2021-12-13 11:22:33 +00:00
Daniel Schürmann
626aa7b648 aco: workaround GFX9 hardware bug for D16 image instructions 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13592>
 2021-11-15 18:28:20 +00:00
Daniel Schürmann
8f1483cd5c aco: add more D16 load/store instructions to RA and validator 
This enables correct handling for
buffer_load/store_format_d16_x and
D16 Image instructions.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13592>
 2021-11-15 18:28:20 +00:00
Daniel Schürmann
7b04c13a34 aco/ra: don't rewrite affinities for phi operands after register assignment 
The effect of doing so is random and not meaningful.

Totals from 52 (0.03% of 150170) affected shaders: (GFX10.3)
CodeSize: 538768 -> 538784 (+0.00%); split: -0.04%, +0.04%
Instrs: 100661 -> 100707 (+0.05%); split: -0.01%, +0.06%
Latency: 1205950 -> 1205768 (-0.02%); split: -0.07%, +0.05%
InvThroughput: 200106 -> 200040 (-0.03%); split: -0.31%, +0.28%
Copies: 5717 -> 5754 (+0.65%); split: -0.17%, +0.82%
Branches: 3153 -> 3162 (+0.29%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12836>
 2021-09-30 09:10:07 +00:00
Daniel Schürmann
b7af10449b aco/ra: create nested affinities for loop header phis 
Totals from 875 (0.58% of 150170) affected shaders: (GFX10.3)
CodeSize: 6084528 -> 6066628 (-0.29%); split: -0.32%, +0.02%
Instrs: 1136497 -> 1133565 (-0.26%); split: -0.28%, +0.02%
Latency: 23355051 -> 22952592 (-1.72%); split: -1.83%, +0.10%
InvThroughput: 13028151 -> 12859628 (-1.29%); split: -1.38%, +0.09%
Copies: 85673 -> 82790 (-3.37%); split: -3.62%, +0.26%
Branches: 25049 -> 25098 (+0.20%); split: -0.08%, +0.28%

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12836>
 2021-09-30 09:10:07 +00:00
Daniel Schürmann
d6bfc95732 aco/ra: create affinities between nested phis 
Totals from 17143 (11.42% of 150170) affected shaders: (GFX10.3)
VGPRs: 1138112 -> 1138440 (+0.03%); split: -0.00%, +0.03%
CodeSize: 131235532 -> 131147080 (-0.07%); split: -0.14%, +0.07%
Instrs: 24848044 -> 24775419 (-0.29%); split: -0.32%, +0.02%
Latency: 599031816 -> 596005601 (-0.51%); split: -0.52%, +0.01%
InvThroughput: 152059329 -> 151054105 (-0.66%); split: -0.66%, +0.00%
VClause: 410951 -> 410958 (+0.00%); split: -0.01%, +0.01%
Copies: 1696885 -> 1621908 (-4.42%); split: -4.64%, +0.22%
Branches: 846710 -> 851052 (+0.51%); split: -0.29%, +0.80%

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12836>
 2021-09-30 09:10:07 +00:00
Daniel Schürmann
a2ffdca26a aco/ra: don't set affinities for ssa-repair phis 
These have no effect anymore.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12836>
 2021-09-30 09:10:07 +00:00
Daniel Schürmann
ae5cbd8f3a aco/ra: for phis try to find an operand-matching register earlier 
Totals from 3557 (2.37% of 150170) affected shaders: (GFX10.3)
VGPRs: 257976 -> 257984 (+0.00%)
CodeSize: 34296232 -> 34270552 (-0.07%); split: -0.09%, +0.01%
Instrs: 6512289 -> 6506900 (-0.08%); split: -0.10%, +0.01%
Latency: 136376181 -> 136262553 (-0.08%); split: -0.10%, +0.02%
InvThroughput: 33042816 -> 32992849 (-0.15%); split: -0.18%, +0.03%
VClause: 104687 -> 104686 (-0.00%)
SClause: 238657 -> 238663 (+0.00%); split: -0.00%, +0.00%
Copies: 477690 -> 471058 (-1.39%); split: -1.52%, +0.13%
Branches: 223058 -> 224326 (+0.57%); split: -0.02%, +0.59%

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12836>
 2021-09-30 09:10:07 +00:00