fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-08 02:38:04 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

nir/opt_algebraic: Optimize mediump ops done in highp

Browse source 
When we do f2fmp(fadd(f2f32(a), f2f32(b))) we can always optimize it to
fadd(a, b) and obtain the same bit-precise result, I verified this using
a custom script with Berkley SoftFloat.

This commit enables these optimizations for fadd, fmul, fdiv and frem,
which have all been proven to be bit-by-bit identical.
This commit is contained in:
Lorenzo Rossi 2026-05-06 18:18:51 +02:00
parent e5e375593b
commit 99a636e434
1 changed files with 8 additions and 5 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

@ -1963,11 +1963,6 @@ optimizations.extend([
(('f2u32', ('f2fmp', 'a@32')), ('f2u32', a), 'true', TestStatus.UNSUPPORTED),
(('i2f32', ('i2imp', 'a@32')), ('i2f32', a), 'true', TestStatus.UNSUPPORTED),
# f16 -> f2f32 -> fmul32 -> f2fmp when the second operand is a constant
# The optimization only works when the constant can be safely represented with 16 bits
(('f2fmp', ('fmul(is_used_once,contract)', ('f2f32', 'a@16'), '#b(is_representable_as_f16)')), ('fmul', a, ('f2fmp', b)), 'true', TestStatus.UNSUPPORTED),
(('f2fmp', ('fadd(is_used_once,contract)', ('f2f32', 'a@16'), '#b(is_representable_as_f16)')), ('fadd', a, ('f2fmp', b)), 'true', TestStatus.UNSUPPORTED),
(('ffloor', 'a(is_integral)'), a),
(('fceil', 'a(is_integral)'), a),
(('ftrunc', 'a(is_integral)'), a),
@ -2011,6 +2006,14 @@ optimizations.extend([
(('ult', a, 0), False),
])
# f2fmp(op(f2f16(a), f2f16(b))) -> op(a, b)
# The result is bit-per-bit identical except for NaN representations.
for op in ['fadd', 'fmul', 'fdiv', 'frem']:
optimizations += [
(('f2fmp', (f'{op}(is_used_once)', ('f2f32', 'a@16'), '#b(is_representable_as_f16)')), (op, a, ('f2fmp', b)), 'true', TestStatus.UNSUPPORTED),
(('f2fmp', (f'{op}(is_used_once)', ('f2f32', 'a@16'), ('f2f32', 'b@16'))), (op, a, b), 'true', TestStatus.UNSUPPORTED),
]
for bits in [16, 32, 64]:
cond = '!(options->lower_doubles_options & nir_lower_fp64_full_software)' if bits == 64 else 'true'
bcsel = 'bcsel@{}'.format(bits)