fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-06-15 15:38:22 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

nir/opt_algebraic_tests: Fix fuzzing levels for multi-component inputs.

Browse source 
We were enumerating enough for a single component, but not all the
combinations.  This helps show that our fdots fail pretty consistently.
And triggers more skipping from the fany_equal16s thanks to varied inputs.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39369>
This commit is contained in:
Emma Anholt 2026-01-16 23:31:27 -08:00 • committed by Marge Bot
parent 7fd0287a89
commit e5a9eae2b5
2 changed files with 10 additions and 10 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
src/compiler/nir/nir_opt_algebraic.py
View file

@ -541,19 +541,19 @@ optimizations.extend([
(('fdot4', ('vec4', a, 0.0, 0.0, 0.0), b), ('fmul', a, 'b.x'), 'true', TestStatus.XFAIL), # XFAIL is that fdot(vec(-1.0, 0.0...), 0.0) produces -0.0 instead of 0.0.
(('fdot4', ('vec4', a, b, 0.0, 0.0), c), ('fdot2', ('vec2', a, b), c), 'true', TestStatus.XFAIL),
(('fdot4', ('vec4', a, b, c, 0.0), d), ('fdot3', ('vec3', a, b, c), d), 'true', TestStatus.XFAIL),
(('fdot4', 'a(w_is_zero)', b), ('fdot3', 'a.xyz', 'b.xyz')),
(('fdot4', 'a(z_is_zero)', b), ('fdot3', 'a.xyw', 'b.xyw')),
(('fdot4', 'a(y_is_zero)', b), ('fdot3', 'a.xzw', 'b.xzw')),
(('fdot4', 'a(x_is_zero)', b), ('fdot3', 'a.yzw', 'b.yzw')),
(('fdot4', 'a(w_is_zero)', b), ('fdot3', 'a.xyz', 'b.xyz'), 'true', TestStatus.XFAIL), # XFAIL is that fdot((1,-1,0,0), (0,.12345,1,inf) produces -.12345 instead of -nan)
(('fdot4', 'a(z_is_zero)', b), ('fdot3', 'a.xyw', 'b.xyw'), 'true', TestStatus.XFAIL),
(('fdot4', 'a(y_is_zero)', b), ('fdot3', 'a.xzw', 'b.xzw'), 'true', TestStatus.XFAIL),
(('fdot4', 'a(x_is_zero)', b), ('fdot3', 'a.yzw', 'b.yzw'), 'true', TestStatus.XFAIL),
(('fdot3', ('vec3', a, 0.0, 0.0), b), ('fmul', a, 'b.x'), 'true', TestStatus.XFAIL),
(('fdot3', ('vec3', a, b, 0.0), c), ('fdot2', ('vec2', a, b), c), 'true', TestStatus.XFAIL),
(('fdot3', 'a(x_is_zero)', b), ('fdot2', 'a.yz', 'b.yz')),
(('fdot3', 'a(y_is_zero)', b), ('fdot2', 'a.xz', 'b.xz')),
(('fdot3', 'a(x_is_zero)', b), ('fdot2', 'a.yz', 'b.yz'), 'true', TestStatus.XFAIL),
(('fdot3', 'a(y_is_zero)', b), ('fdot2', 'a.xz', 'b.xz'), 'true', TestStatus.XFAIL),
(('fdot3', 'a(z_is_zero)', b), ('fdot2', 'a.xy', 'b.xy'), 'true', TestStatus.XFAIL),
(('fdot2', ('vec2', a, 0.0), b), ('fmul', a, 'b.x'), 'true', TestStatus.XFAIL),
(('fdot2', 'a(x_is_zero)', b), ('fmul', 'a.y', 'b.y')),
(('fdot2', 'a(x_is_zero)', b), ('fmul', 'a.y', 'b.y'), 'true', TestStatus.XFAIL),
(('fdot2', 'a(y_is_zero)', b), ('fmul', 'a.x', 'b.x'), 'true', TestStatus.XFAIL),
(('fdot2', a, 1.0), ('fadd', 'a.x', 'a.y')),
@ -1536,12 +1536,12 @@ optimizations.extend([
(('fall_equal3', a, b), ('seq', ('fany_nequal3', a, b), 0.0), 'options->lower_vector_cmp'),
(('fall_equal4', a, b), ('seq', ('fany_nequal4', a, b), 0.0), 'options->lower_vector_cmp'),
(('fall_equal8', a, b), ('seq', ('fany_nequal8', a, b), 0.0), 'options->lower_vector_cmp'),
(('fall_equal16', a, b), ('seq', ('fany_nequal16', a, b), 0.0), 'options->lower_vector_cmp'),
(('fall_equal16', a, b), ('seq', ('fany_nequal16', a, b), 0.0), 'options->lower_vector_cmp', TestStatus.UNSUPPORTED), # all test inputs skipped
(('fany_nequal2', a, b), ('fmax', ('sne', 'a.x', 'b.x'), ('sne', 'a.y', 'b.y')), 'options->lower_vector_cmp'),
(('fany_nequal3', a, b), ('fsat', ('fdot3', ('sne', a, b), ('sne', a, b))), 'options->lower_vector_cmp'),
(('fany_nequal4', a, b), ('fsat', ('fdot4', ('sne', a, b), ('sne', a, b))), 'options->lower_vector_cmp'),
(('fany_nequal8', a, b), ('fsat', ('fdot8', ('sne', a, b), ('sne', a, b))), 'options->lower_vector_cmp'),
(('fany_nequal16', a, b), ('fsat', ('fdot16', ('sne', a, b), ('sne', a, b))), 'options->lower_vector_cmp'),
(('fany_nequal16', a, b), ('fsat', ('fdot16', ('sne', a, b), ('sne', a, b))), 'options->lower_vector_cmp', TestStatus.UNSUPPORTED), # all test inputs skipped
# Vulkan allows us to use any rounding mode, so choose rtz because it's simple.
# Avoid some NaNs being converted to Inf if the lsb are cut off.

2
src/compiler/nir/tests/nir_algebraic_pattern_test.cpp
View file

@ -159,7 +159,7 @@ map_input(nir_builder *b, nir_intrinsic_instr *intr, void *data)
ty = INT;
test->inputs.push_back(nir_algebraic_pattern_test_input(nir_instr_as_load_const(nir_def_instr(load)),
ty, test->fuzzing_bits));
test->fuzzing_bits += get_seed_bit_size(ty);
test->fuzzing_bits += get_seed_bit_size(ty) * intr->def.num_components;
nir_def_replace(&intr->def, load);