• Joined on 2026-04-09
zhoulin pushed to main at zhoulin/zl_bench 2026-08-13 03:04:22 +08:00
da22885645 Add multi-op batch screening layer; fix cudagraph fallback regressions
zhoulin pushed to main at zhoulin/zl_bench 2026-08-12 18:55:41 +08:00
95895cea0e Adapt to upstream FlagGems repo split and record pre_hook configs for libtuner
zhoulin pushed to main at zhoulin/zl_bench 2026-07-29 20:14:29 +08:00
1e1d612031 Add opt-in multi-GPU sweep parallelism; fix GPU pinning and run dir races
zhoulin pushed to main at zhoulin/zl_bench 2026-07-29 08:38:27 +08:00
7e0e8648f1 Fix autotune record/replay silently skipping libtuner kernels
zhoulin pushed to main at zhoulin/zl_bench 2026-07-24 18:17:43 +08:00
f98ec10fd0 Renew scripts
zhoulin pushed to main at zhoulin/zl_bench 2026-07-21 00:54:32 +08:00
c8a42b66ee Move trace shape sets to top-level shape/ and frame ab_fold_test.sh as A/B reference example
00902e6c67 Default USE_FLAGTUNE to 0 and add fold_scale A/B script with trace shapes
Compare 2 commits »
zhoulin pushed to main at zhoulin/zl_bench 2026-07-20 03:21:25 +08:00
26b071c6e1 Stabilize first-run cudagraph timing and colorize terminal output
zhoulin pushed to main at zhoulin/zl_bench 2026-07-16 11:52:24 +08:00
8ace0d790c Add FlagGems single-op perf benchmark harness
zhoulin created repository zhoulin/zl_bench 2026-07-15 17:20:28 +08:00