#1585
Tests: skip ps_vcmpx_forms on narrow fragment subgroups (TEST-205)
Summary
- TEST-205: skip
ps_vcmpx_formson devices whose subgroup range has min under 32 (Intel UHD 770 class), wherevk_harnesscannot pin fragment and the 32-lanev_cmpxEXEC packing is not comparable — same root cause #1318 handled forps_bitmask_perm. - Shared
vcmpx_forms_narrow_skipincheck.py/run_gpu_tests.py; unit tests cover skip vs judge on 8..32 / 32..32 / missing width. - Claimed in #1548.
Test plan
-
python -I -m unittest discover -s tests -p test_run_gpu_pixel_checks.py -v -
python -I -m unittest discover -s tests -p test_shader_checks.py -v - Negative control: without the skip, Intel 8..32 + bad words is a 4096/4096
differ(false failure class); with skip it returnsSkip - Optional: on an Intel GPU with subgroup 8..32, confirm
ps_vcmpx_formsis SKIP rather than FAIL; on NVIDIA 32..32 it still judges
AI-assisted. No game files. Dual-tested (author + reviewer/negative control). GPU for optional confirm: Intel UHD 770 / any 8..32 fragment device.
Activity
4You're right — thanks for the bit-level decode. Word 3 matching 0x15550000 rules out the #1318 lane-mask story; every bad bit has a NaN operand, so this is the fast-math / NaN-preserve case (#664 / #1512), not a narrow fragment subgroup.
Closing this skip: it would hide real failures (including on MoltenVK's 4..32 range). #1512 is already on main; I'll leave judging to #664 for the translator side.
AI-assisted.
Closing: misdiagnosed as #1318-class skip; root cause is NaN compares (see @Jolymmiles). Yield to #664 / #1512.
I don't think a subgroup-width skip is the right fix here. The words show that the cause is NaN comparisons, not lane masks.
Decoded with the bit layout in the
.asmheader (bit 4k+i of word w = op k on pair i):Intel UHD 770 (the numbers posted on #1374):
ngt(NaN, 1.0)got 0, want 1nle(NaN, 1.0)got 0, want 1eq(NaN, NaN)got 1, want 0le(NaN, NaN)got 1, want 0nle(NaN, NaN)got 0, want 1Apple M3 Max, MoltenVK 1.4.2 (macOS 27.3,
mainat 99f99890),got 0xf05e8e16, want 0xf05e8a12in word 1:eq(NaN, NaN)got 1, want 0le(NaN, NaN)got 1, want 0Every mismatching bit has a NaN operand. All non-NaN bits match, and so does word 3, the EXEC/VCC word (
0x15550000is the expected value, not a half lane mask). Both drivers treat the floats as never NaN:x == xfolds to true, and!(a > b)becomesa <= b.On the Mac I checked it both ways:
MVK_CONFIG_FAST_MATH_ENABLED=0:ps_vcmpx_formspasses, 4096/4096 exact. With=1or=2it fails as above. This is what #1512 sets for the test tools.SignedZeroInfNanPreserve 32, applied on 99f99890) with MoltenVK's default fast math: it passes, 4096/4096 exact.The guard here also skips on MoltenVK, which reports a 4..32 subgroup range. On a Mac it would turn a real translator/driver NaN problem into a SKIP. I'd keep the test judging and let #664 (and #1512 for MoltenVK) fix it. Could you check
ps_vcmpx_formson the UHD 770 with #664 applied?Tested on Apple M3 Max, macOS 27.3, MoltenVK 1.4.2 (Homebrew), no game files. AI-assisted.