Files
ANSCORE/modules/ANSFR
nghianguyenandClaude Opus 4.8 da9d922498 ANSFR: gate recognizer batch size by GPU VRAM
Batch-4 was tuned for 8 GB GPUs; its TRT workspace (~54 MiB per batch
unit) crowds out the other engines on smaller cards. VramCappedBatch
keeps the configured batch on 8 GB-class cards and up (threshold
7.5 GiB — 8 GB cards report ~8188 MiB) and drops to single-face batches
below that: identical results, multi-face frames just run sequentially,
~160 MiB freed on a 4 GB RTX A400. Also caps the legacy 8/32 LoadEngine
profile, whose .b32 engine (~1.7 GB workspace) would be fatal on small
cards. Initialize() and OptimizeModel() share the same call so their
profiles cannot drift.

Small cards rebuild the recognizer engine once (filename loses the .b4
suffix); 8 GB+ machines keep their cached engines unchanged.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-08-21 10:44:22 +10:00
..
2026-03-28 19:56:39 +11:00
2026-04-24 17:10:29 +10:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-04-24 17:10:29 +10:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00
2026-03-28 19:56:39 +11:00