Claude 9bd8089cbb test(agc): compare llvmpipe shader clocks modulo 2^32 ...

llvmpipe builds the high dword of the shader clock with a left shift by
32 (emit_clock in Mesa's lp_bld_nir_soa.c), so it is always 0 and
s_memtime and s_memrealtime wrap every 2^32 ns. A pass of
agc_driver_shader_clock whose two reads straddle a wrap failed with
"memtime: lane N went back from 4294921902 to 3823"; dispatching right
before the boundary fails every lane of that pass.

On llvmpipe compare the low dwords modulo 2^32, which stays correct if
Mesa fixes the shift. Other devices keep the 64-bit comparison.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ha6Tsy62AFUyp9a5Mf7vjA

2026-10-09 10:30:28 +00:00

..

ARCHITECTURE.md

feat(shader): implement s_swappc_b64 fetch-shader call

2026-10-08 20:05:24 +02:00

BUILD.md

build(relinker): add a dependency-free conversion build mode

2026-10-08 18:21:30 +11:00

CONVENTIONS.md

refactor(docs): separated user and dev in docs + compatibility file

2026-09-28 22:58:17 +03:00

HW_ORACLE.md

fix(oracle)!: require explicit float modes

2026-10-07 14:58:28 +05:00

PROGRESS.md

fix(progress): exclude library test fixtures from function counts

2026-10-08 17:30:43 +05:30

TechnicalDebt.md

test(agc): compare llvmpipe shader clocks modulo 2^32

2026-10-09 10:30:28 +00:00