fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-24 13:10:10 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

nir/algebraic: optimize 64-bit comparisons with zero'd halves to 32-bit

Browse source 
These expect nir_lower_int64 to replace u2u64 to pack_64_2x32_split(, 0).

fossil-db (navi31):
Totals from 149 (0.19% of 79242) affected shaders:
Instrs: 433095 -> 431830 (-0.29%); split: -0.29%, +0.00%
CodeSize: 2165980 -> 2160284 (-0.26%); split: -0.27%, +0.00%
SpillSGPRs: 689 -> 688 (-0.15%)
Latency: 3801497 -> 3799901 (-0.04%); split: -0.05%, +0.01%
InvThroughput: 1547916 -> 1546567 (-0.09%); split: -0.09%, +0.01%
VClause: 4698 -> 4693 (-0.11%)
SClause: 9981 -> 9977 (-0.04%); split: -0.05%, +0.01%
Copies: 66148 -> 65431 (-1.08%); split: -1.09%, +0.01%
PreSGPRs: 6732 -> 6729 (-0.04%)
PreVGPRs: 7976 -> 7945 (-0.39%)
VALU: 252936 -> 252336 (-0.24%)
SALU: 51794 -> 51274 (-1.00%); split: -1.03%, +0.02%

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27335>
This commit is contained in:
Rhys Perry 2024-01-26 20:48:32 +00:00 • committed by Marge Bot
parent 417eb390c6
commit b37804c8de
1 changed files with 13 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

13
src/compiler/nir/nir_opt_algebraic.py
View file

@ -1811,6 +1811,19 @@ for op in ('extract_u8', 'extract_i8'):
optimizations.extend([((op, ('ishl', 'a@32', 24 - 8 * i), 3), (op, a, i)) for i in range(2, -1, -1)])
optimizations.extend([((op, ('ishl', 'a@64', 56 - 8 * i), 7), (op, a, i)) for i in range(6, -1, -1)])
for op, repl in [('ieq', 'ieq'), ('ine', 'ine'),
('ult', 'ult'), ('ilt', 'ult'),
('uge', 'uge'), ('ige', 'uge')]:
optimizations.extend([
((op, ('pack_64_2x32_split', a, 0), ('pack_64_2x32_split', b, 0)), (repl, a, b)),
((op, ('pack_64_2x32_split', a, 0), '#b(is_upper_half_zero)'), (repl, a, ('unpack_64_2x32_split_x', b))),
((op, '#a(is_upper_half_zero)', ('pack_64_2x32_split', b, 0)), (repl, ('unpack_64_2x32_split_x', a), b)),
((op, ('pack_64_2x32_split', 0, a), ('pack_64_2x32_split', 0, b)), (op, a, b)),
((op, ('pack_64_2x32_split', 0, a), '#b(is_lower_half_zero)'), (op, a, ('unpack_64_2x32_split_y', b))),
((op, '#a(is_lower_half_zero)', ('pack_64_2x32_split', 0, b)), (op, ('unpack_64_2x32_split_y', a), b)),
])
optimizations.extend([
# Subtracts
(('ussub_4x8_vc4', a, 0), a),