fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-19 13:38:19 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 2
194463 commits 66 branches 935 tags 966 MiB
Commit graph

3278 commits

Author SHA1 Message Date
Rhys Perry
192ef947a9 aco: fix printing of d16 MIMG dmask 
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/30761>
 2024-09-03 21:14:27 +00:00
Rhys Perry
999362f895 aco: have get_subdword_definition_info update DefInfo 
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/30761>
 2024-09-03 21:14:27 +00:00
Rhys Perry
16ff51bb37 aco/ra: use DefInfo for get_reg_specified 
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/30761>
 2024-09-03 21:14:27 +00:00
Rhys Perry
1e6741bf6b aco/ra: fix sub-dword get_reg_specified in some cases 
For example, v6b MIMG can have sdw_def_info={4, 6}. This now has similar
behaviour as the DefInfo constructor.

fossil-db (navi31):
Totals from 5 (0.01% of 79395) affected shaders:
CodeSize: 29460 -> 29408 (-0.18%); split: -0.29%, +0.11%
Latency: 22133 -> 21934 (-0.90%); split: -2.51%, +1.61%
InvThroughput: 2953 -> 2963 (+0.34%); split: -0.03%, +0.37%
Copies: 410 -> 409 (-0.24%); split: -1.95%, +1.71%
VALU: 3242 -> 3241 (-0.03%); split: -0.25%, +0.22%

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fixes: 56345b8c61 ("aco: allow reading/writing upper halves/bytes when possible")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30761>
 2024-09-03 21:14:27 +00:00
Georg Lehmann
9bb10b58f3 aco: use v_cvt_pk_u8_f32 for f2u8 
Foz-DB Navi31:
Totals from 42 (0.05% of 79395) affected shaders:
Instrs: 3253747 -> 3248867 (-0.15%); split: -0.15%, +0.00%
CodeSize: 16690136 -> 16661772 (-0.17%); split: -0.17%, +0.00%
VGPRs: 4176 -> 4128 (-1.15%)
Latency: 18485157 -> 18479752 (-0.03%); split: -0.03%, +0.00%
InvThroughput: 3659404 -> 3658222 (-0.03%); split: -0.03%, +0.00%
Copies: 231891 -> 228145 (-1.62%)
VALU: 1785800 -> 1782054 (-0.21%)

Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29532>
 2024-09-02 19:49:20 +00:00
Daniel Schürmann
6ada0170c4 aco/ra: use arena allocator for hash maps 
Also change Key types to uint32_t instead of unsigned.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30898>
 2024-09-02 15:31:52 +00:00
Daniel Schürmann
709f60e71d aco/ra: use auto for std container iterators 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30898>
 2024-09-02 15:31:52 +00:00
Daniel Schürmann
534031761a aco/ra: avoid rename map lookups in most cases 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30898>
 2024-09-02 15:31:52 +00:00
Daniel Schürmann
72464e24b4 aco/ra: add add_rename() helper 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30898>
 2024-09-02 15:31:51 +00:00
Georg Lehmann
8f3bb1fb2e aco/ra: always reuse def register for literal copy 
Foz-DB Vega10:
Totals from 4056 (6.43% of 63053) affected shaders:
Instrs: 5540797 -> 5540760 (-0.00%); split: -0.00%, +0.00%
CodeSize: 29680912 -> 29680812 (-0.00%); split: -0.00%, +0.00%
SGPRs: 307808 -> 307680 (-0.04%)
Latency: 82483179 -> 82484437 (+0.00%); split: -0.00%, +0.00%
InvThroughput: 45546645 -> 45547091 (+0.00%); split: -0.00%, +0.00%
SClause: 185641 -> 185648 (+0.00%); split: -0.00%, +0.01%
Copies: 642229 -> 642214 (-0.00%); split: -0.01%, +0.01%
SALU: 643132 -> 643107 (-0.00%); split: -0.01%, +0.01%

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30932>
 2024-09-02 11:09:55 +00:00
Georg Lehmann
5fb54d1fde aco/ra: fix copying 64bit literal to sgprs 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30932>
 2024-09-02 11:09:55 +00:00
Georg Lehmann
364764356c aco/ra: use parallelcopy to copy literal instead of s_mov 
This can result in smaller code size.

Foz-DB Vega10:
Totals from 4745 (7.53% of 63053) affected shaders:
CodeSize: 33014176 -> 33007088 (-0.02%); split: -0.02%, +0.00%
Copies: 690055 -> 704738 (+2.13%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30932>
 2024-09-02 11:09:55 +00:00
Georg Lehmann
607cf5a8e9 aco/ra: unconditionally replace literal with sgpr when promoting to VOP3 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30932>
 2024-09-02 11:09:55 +00:00
Georg Lehmann
246e22ff4f aco/tests: do not use mul with constant to tests neg modifier 
The neg can be moved to the constant operand, which defeats the point
of the test.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30781>
 2024-08-27 20:41:10 +00:00
Georg Lehmann
bf67ac30fe aco/tests: allow literals with resolved swizzles in vop3p test 
My new optimizer code will resolve swizzles for constants.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30781>
 2024-08-27 20:41:09 +00:00
Georg Lehmann
6a18eb6afc aco/tests: parse neg(constant) in vop3p test 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30781>
 2024-08-27 20:41:09 +00:00
Georg Lehmann
52465956ca aco/print_ir: use neg() for constants 
Otherwise, it's not clear if -1 is 0xffffffff or 0x80000001.
LLVM uses a similar logic.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30781>
 2024-08-27 20:41:09 +00:00
Georg Lehmann
fb8e730d9b aco/tests: do not use add to tests neg modifer 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30781>
 2024-08-27 20:41:09 +00:00
Georg Lehmann
f71522e5cf aco/tests: don't test dpp constant propagation with row shift 
With bc=1, removing DPP for shifts is invalid.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30781>
 2024-08-27 20:41:09 +00:00
Rhys Perry
dea1fedf51 aco/tests: add more VALUMaskWriteHazard tests 
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/30818>
 2024-08-26 19:16:34 +00:00
Rhys Perry
11262a01ce aco: preserve bitsets after a lane mask is written 
fossil-db (navi31):
Totals from 4840 (6.10% of 79395) affected shaders:
Instrs: 13733449 -> 13761177 (+0.20%); split: -0.00%, +0.21%
CodeSize: 71997868 -> 72102520 (+0.15%); split: -0.00%, +0.15%
Latency: 128385177 -> 128408780 (+0.02%); split: -0.00%, +0.02%
InvThroughput: 21105847 -> 21109475 (+0.02%); split: -0.00%, +0.02%
VALU: 7741209 -> 7741210 (+0.00%)

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Backport-to: 24.1
Backport-to: 24.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30818>
 2024-08-26 19:16:34 +00:00
Rhys Perry
61e73c2323 aco: check SALU writing lanemask later for VALUMaskWriteHazard 
This should be done after reads are checked and
sgpr_read_by_valu_as_lanemask_then_wr_by_salu is reset. The old version
also skipped checking the reads if the write check passed.

fossil-db (navi31):
Totals from 193 (0.24% of 79395) affected shaders:
Instrs: 3212435 -> 3212735 (+0.01%)
CodeSize: 16462868 -> 16463848 (+0.01%); split: -0.00%, +0.01%
Latency: 19492377 -> 19492462 (+0.00%); split: -0.00%, +0.00%
InvThroughput: 4419705 -> 4419718 (+0.00%); split: -0.00%, +0.00%

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Backport-to: 24.1
Backport-to: 24.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30818>
 2024-08-26 19:16:34 +00:00
Rhys Perry
b1ba7d1b99 aco: don't consider sa_sdst=0 before SALU write to fix VALUMaskWriteHazard 
LLVM does but that's probably a bug.

fossil-db (navi31):
Totals from 311 (0.39% of 79395) affected shaders:
Instrs: 380453 -> 381075 (+0.16%)
CodeSize: 1961012 -> 1964744 (+0.19%)
Latency: 4799095 -> 4800313 (+0.03%)
InvThroughput: 958358 -> 958904 (+0.06%)
VALU: 242322 -> 242633 (+0.13%)

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Backport-to: 24.1
Backport-to: 24.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30818>
 2024-08-26 19:16:34 +00:00
Rhys Perry
8f5ee70d85 aco: also consider VALU reads for VALUMaskWriteHazard 
fossil-db (navi31):
Totals from 9776 (12.31% of 79395) affected shaders:
Instrs: 19348258 -> 19383680 (+0.18%); split: -0.00%, +0.19%
CodeSize: 101223460 -> 101366964 (+0.14%); split: -0.01%, +0.15%
Latency: 172853115 -> 172866070 (+0.01%); split: -0.01%, +0.01%
InvThroughput: 27590468 -> 27592390 (+0.01%); split: -0.00%, +0.01%

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11550
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11436
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11337
Gitlab: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11738
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11741
Backport-to: 24.1
Backport-to: 24.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30818>
 2024-08-26 19:16:34 +00:00
Rhys Perry
ee648326d9 aco: ignore exec and literals when mitigating VALUMaskWriteHazard 
LLVM ignores exec and literals don't seem to work in some cases.

fossil-db (navi31):
Totals from 2676 (3.37% of 79395) affected shaders:
Instrs: 10638979 -> 10646019 (+0.07%); split: -0.00%, +0.07%
CodeSize: 55929640 -> 55959416 (+0.05%); split: -0.00%, +0.06%
Latency: 107707408 -> 107712893 (+0.01%); split: -0.00%, +0.01%
InvThroughput: 18119843 -> 18120442 (+0.00%); split: -0.00%, +0.00%

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Backport-to: 24.1
Backport-to: 24.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30818>
 2024-08-26 19:16:34 +00:00
Daniel Schürmann
14de650d58 aco: call nir_copy_prop() and nir_opt_dce() before instruction selection 
Totals from 1037 (1.31% of 79395) affected shaders: (Navi21)

MaxWaves: 18760 -> 18960 (+1.07%)
Instrs: 4865258 -> 4860063 (-0.11%); split: -0.11%, +0.00%
CodeSize: 27094112 -> 27089224 (-0.02%); split: -0.06%, +0.04%
VGPRs: 68816 -> 68000 (-1.19%)
SpillVGPRs: 2140 -> 2105 (-1.64%)
Scratch: 4237312 -> 4234240 (-0.07%)
Latency: 55894512 -> 55748035 (-0.26%); split: -0.31%, +0.05%
InvThroughput: 11611286 -> 11372897 (-2.05%); split: -2.09%, +0.03%
VClause: 145331 -> 145285 (-0.03%); split: -0.04%, +0.01%
SClause: 150339 -> 150338 (-0.00%)
Copies: 472476 -> 468470 (-0.85%); split: -0.88%, +0.03%
Branches: 206562 -> 206067 (-0.24%); split: -0.24%, +0.00%
PreVGPRs: 61747 -> 61361 (-0.63%)
VALU: 3116434 -> 3112660 (-0.12%); split: -0.13%, +0.00%
SALU: 723154 -> 722887 (-0.04%); split: -0.04%, +0.01%
VMEM: 238656 -> 238586 (-0.03%)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30786>
 2024-08-26 12:59:00 +00:00
Rhys Perry
7b92e11e16 aco: forget valu delays after certain s_waitcnt_depctr/LDSDIR 
fossil-db (navi31):
Totals from 55242 (69.58% of 79395) affected shaders:
Instrs: 40507666 -> 40138006 (-0.91%); split: -0.91%, +0.00%
CodeSize: 212516104 -> 211025880 (-0.70%); split: -0.70%, +0.00%
Latency: 281643258 -> 281628053 (-0.01%); split: -0.01%, +0.00%
InvThroughput: 46370668 -> 46369637 (-0.00%); split: -0.00%, +0.00%

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Acked-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23337>
 2024-08-22 13:57:01 +00:00
Rhys Perry
30396ba604 aco: move insert_delay_alu to after insert_NOPs 
s_delay_alu doesn't affect any hazards, but hazard workarounds don't
update s_delay_alu and so can make the s_delay_alu affect the wrong
instructions.

fossil-db (navi31):
Totals from 55777 (70.25% of 79395) affected shaders:
Instrs: 40740011 -> 40765017 (+0.06%)
CodeSize: 213768484 -> 213870856 (+0.05%); split: -0.00%, +0.05%
Latency: 283713083 -> 283714959 (+0.00%); split: -0.00%, +0.00%
InvThroughput: 46551791 -> 46551835 (+0.00%); split: -0.00%, +0.00%

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Acked-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23337>
 2024-08-22 13:57:01 +00:00
Rhys Perry
807651561e aco: split insert_wait_states into two 
No fossil-db changes.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Acked-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23337>
 2024-08-22 13:57:00 +00:00
Rhys Perry
db39685e31 aco: validate temporary reachability 
These errors are nicer to read than the ones from live variable analysis.

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/30440>
 2024-08-21 10:47:20 +00:00
Rhys Perry
39270a8be3 aco: preserve SSA in try_eliminate_scc_copy 
Otherwise, there is no definition of this temporary. Fixes fail in future
validation.

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/30440>
 2024-08-21 10:47:20 +00:00
Rhys Perry
d81d5b7d00 aco: use dominance helpers 
This makes the passes slightly faster.

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/30440>
 2024-08-21 10:47:20 +00:00
Rhys Perry
d91e634c13 aco: calculate indices from dominance tree 
This lets us do fast dominance checks.

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/30440>
 2024-08-21 10:47:20 +00:00
Samuel Pitoiset
aad503ecfa aco: fix bogus assert in RT prolog on GFX11+ 
in_scratch_offset isn't defined on GFX11+ and only useful on < GFX9.

Fixes: bd525f4282 ("aco: Fix 1D->2D dispatch conversion on <gfx9")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30717>
 2024-08-19 21:07:31 +00:00
Friedrich Vock
bd525f4282 aco: Fix 1D->2D dispatch conversion on <gfx9 
out_args->scratch_offset and in_wg_id_x will alias on <gfx9.
To avoid the conversion code reading a garbage WG ID, move the
scratch/ring offset writing to the very end.

Fixes: 1e354172 ("radv,aco: Convert 1D ray launches to 2D")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30707>
 2024-08-17 17:01:10 +00:00
Rhys Perry
aafb49f56b aco: set prefer_remove for gfx9- too 
This is a hint that the branch is worth removing. Assume that's the case,
regardless of the gfx level.

fossil-db (vega10):
Totals from 22 (0.03% of 63053) affected shaders:
Instrs: 23927 -> 23856 (-0.30%)
CodeSize: 125096 -> 124812 (-0.23%)
Latency: 138258 -> 137765 (-0.36%)
InvThroughput: 55900 -> 55884 (-0.03%)
Branches: 391 -> 320 (-18.16%)

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/30321>
 2024-08-15 16:00:19 +00:00
Rhys Perry
9f1a5645cf aco: completely skip branches if they're never taken 
fossil-db (navi21):
Totals from 196 (0.25% of 79395) affected shaders:
Instrs: 101902 -> 101706 (-0.19%)
CodeSize: 576988 -> 576232 (-0.13%)
Latency: 750344 -> 750280 (-0.01%); split: -0.01%, +0.00%
InvThroughput: 119170 -> 119161 (-0.01%)
Branches: 3933 -> 3737 (-4.98%)

fossil-db (vega10):
Totals from 585 (0.93% of 63053) affected shaders:
Instrs: 346877 -> 346292 (-0.17%)
CodeSize: 1810600 -> 1808260 (-0.13%)
Latency: 1817743 -> 1814233 (-0.19%)
InvThroughput: 652142 -> 651944 (-0.03%)
Branches: 5087 -> 4502 (-11.50%)

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30321>
 2024-08-15 16:00:19 +00:00
Rhys Perry
c29d9f1184 aco: only remove branch jumping over SMEM/barrier if it's never taken 
SMEM might be an invalid access, and barriers are probably expensive.

fossil-db (navi21):
Totals from 126 (0.16% of 79395) affected shaders:
Instrs: 2764965 -> 2765377 (+0.01%)
CodeSize: 15155348 -> 15156788 (+0.01%)
Latency: 17604293 -> 17604296 (+0.00%)
Branches: 105211 -> 105623 (+0.39%)

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Backport-to: 24.1
Backport-to: 24.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30321>
 2024-08-15 16:00:19 +00:00
Rhys Perry
b934255510 aco: split selection_control_remove into rarely_taken and never_taken 
No fossil-db changes.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Backport-to: 24.1
Backport-to: 24.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30321>
 2024-08-15 16:00:18 +00:00
Daniel Schürmann
1d0a12438d aco/cssa: short-cut some trivial case 
If a phi-operand is not flagged as kill, it cannot be coalesced
because it interferes with the live-out variable.
Also do the regClass check earlier.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30182>
 2024-08-14 08:11:48 +00:00
Daniel Schürmann
d3e9aef5a2 aco/cssa: update RegisterDemand and validate live variable information 
instead of recomputing it.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30182>
 2024-08-14 08:11:48 +00:00
Daniel Schürmann
5a39cbdef6 aco: change signature of get_live_changes() and get_temp_registers() 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30182>
 2024-08-14 08:11:48 +00:00
Daniel Schürmann
d494c2a741 aco/cssa: fix kill flags during lowering to CSSA 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30182>
 2024-08-14 08:11:48 +00:00
Daniel Schürmann
541cfb21ba aco: don't attempt to spill dead phis 
These don't affect register pressure.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30182>
 2024-08-14 08:11:48 +00:00
Daniel Schürmann
e5d920e0b9 aco/scheduler: enable live variables validation when ACO_DEBUG=validate-livevars is set 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30182>
 2024-08-14 08:11:48 +00:00
Daniel Schürmann
b0c8c5e42e aco: implement aco::validate_live_vars() 
This is intended for passes which manually update live variables
and RegisterDemand, like e.g. the scheduler, and can be enabled
with ACO_DEBUG=validate-livevars.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30182>
 2024-08-14 08:11:48 +00:00
Daniel Schürmann
c1a3330ac7 aco/reindex_ssa: free memory of previous live variable sets 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30182>
 2024-08-14 08:11:47 +00:00
Georg Lehmann
dedfff9dbf aco: only set latekill in live_var_analysis 
Cleaner to have this all in one place, in my opinion.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30368>
 2024-08-12 10:31:09 +00:00
Georg Lehmann
510f5e55be aco/gfx10+: set lateKill for sgprs used by wave64 VALU writing a mask 
RDNA2 ISA doc, 6.2.4. Wave64 Destination Restrictions:
The first pass of a wave64 VALU instruction may not overwrite a scalar value
used by the second half.

Foz-DB Navi31:
Totals from 5221 (6.58% of 79395) affected shaders:
Instrs: 9751484 -> 9752179 (+0.01%); split: -0.01%, +0.01%
CodeSize: 50624072 -> 50626088 (+0.00%); split: -0.00%, +0.01%
Latency: 85646450 -> 85647419 (+0.00%); split: -0.00%, +0.00%
InvThroughput: 15039160 -> 15039277 (+0.00%); split: -0.00%, +0.00%
VClause: 200275 -> 200204 (-0.04%)
SClause: 248645 -> 248607 (-0.02%); split: -0.03%, +0.01%
Copies: 640802 -> 641413 (+0.10%); split: -0.01%, +0.11%
PreSGPRs: 236297 -> 236735 (+0.19%)
VALU: 5666449 -> 5666440 (-0.00%)
SALU: 967482 -> 968111 (+0.07%); split: -0.01%, +0.07%

Cc: mesa-stable

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30368>
 2024-08-12 10:31:09 +00:00
Georg Lehmann
df16f47036 aco: optimize dd[xy]_fine if it's only used by abs 
If we can ignore the sign of the derivative, we can swap the lanes
instead of broadcasting per direction. abs(a - b) = abs(b - a).
Shamelessly copied from bifrost.

Foz-DB Navi31:
Totals from 5 (0.01% of 79206) affected shaders:
Instrs: 6191 -> 6184 (-0.11%)
CodeSize: 31960 -> 31920 (-0.13%)
Latency: 111961 -> 111926 (-0.03%)
InvThroughput: 10390 -> 10372 (-0.17%)
VALU: 3286 -> 3279 (-0.21%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30557>
 2024-08-08 17:39:55 +00:00