fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-07 00:38:48 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

nir/algebraic: optimize exact f2u32(fmul(unpack_norm))

Browse source 
fossil-db (navi21):
Totals from 16 (0.01% of 202427) affected shaders:
Instrs: 17730 -> 17226 (-2.84%)
CodeSize: 97500 -> 95708 (-1.84%)
InvThroughput: 44437 -> 44419 (-0.04%)
Copies: 1502 -> 1446 (-3.73%)
VALU: 9973 -> 9525 (-4.49%)
SALU: 3509 -> 3453 (-1.60%)

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/40740>
This commit is contained in:
Rhys Perry 2026-03-31 17:45:23 +01:00 • committed by Marge Bot
parent f52dace6e8
commit b30c0d8264
1 changed files with 17 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

17
src/compiler/nir/nir_opt_algebraic.py
View file

@ -2708,6 +2708,23 @@ optimizations.extend([
127.0))),
'options->lower_unpack_snorm_4x8'),
# f2u32(fmul(unpack_*norm_*)). These are exact replacements.
# 0x37800080 = 1.0 / 65535.0
# 0x3b808081 = 1.0 / 255.0
# 0x38000100 = 1.0 / 32767.0
# 0x3c010204 = 1.0 / 127.0
(('f2u32', ('fmul', ('fmul', ('u2f32', ('extract_u16', 'a@32', b)), 0x37800080), 65535.0)),
('extract_u8', a, b),
'!nir_is_rounding_mode_rtz(info->float_controls_execution_mode, 32)'),
(('f2u32', ('fmul', ('fmul', ('u2f32', ('extract_u8', 'a@32', b)), 0x3b808081), 255.0)),
('extract_u8', a, b)),
(('f2u32', ('fmul', ('fmin', 1.0, ('fmax', -1.0, ('fmul', ('u2f32', ('extract_i16', 'a@32', b)), 0x38000100))), 32767.0)),
('imax', ('extract_i16', a, b), -32767),
'!nir_is_rounding_mode_rtz(info->float_controls_execution_mode, 32)'),
(('f2u32', ('fmul', ('fmin', 1.0, ('fmax', -1.0, ('fmul', ('u2f32', ('extract_i8', 'a@32', b)), 0x3c010204))), 127.0)),
('imax', ('extract_i8', a, b), -127),
'!nir_is_rounding_mode_rtz(info->float_controls_execution_mode, 32)'),
(('pack_half_2x16_split', 'a@32', 'b@32'),
('ior', ('ishl', ('u2u32', ('f2f16', b)), 16), ('u2u32', ('f2f16', a))),
'options->lower_pack_split'),