ggml-org/whisper.cpp

Should we use process time for benchmarks/timings?

オープン

#382 opened on 2023/01/07

 (2 件のコメント) (0 件のリアクション) (0 人の担当者)C++ (6,074 件のフォーク)batch import
enhancementgood first issue

Repository metrics

Stars
 (52,968 個のスター)
PR merge metrics
 (平均マージ 5d 20h) (30d で 36 merged PRs)

説明

Currently, the bench tool simply spits out the timings from whisper_print_timings. These timings are not process specific and as such are influenced by other processes on the system.

Should we use something like CLOCK_PROCESS_CPUTIME_ID instead of CLOCK_MONOTONIC in ggml_time_*? This would make the benchmarks in #89 outdated, but would make future benchmarks more realistic.

コントリビューターガイド