From bf64c0d9c28f2759960a007bbb5edd40f8f7deba Mon Sep 17 00:00:00 2001 From: Frank Date: Tue, 11 Aug 2026 16:36:39 -0700 Subject: [PATCH] perf: frank-default-large batch 12288 / ubatch 4096 --- scripts/frank-default-large.env | 4 ++-- scripts/serve-llama-frank-large.sh | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/scripts/frank-default-large.env b/scripts/frank-default-large.env index 9ca3eea..59b936a 100644 --- a/scripts/frank-default-large.env +++ b/scripts/frank-default-large.env @@ -23,7 +23,7 @@ N_GPU_LAYERS=all # KV on GPU, quantized (K q8_0, V f16). CACHE_TYPE_K=q8_0 CACHE_TYPE_V=f16 -BATCH_SIZE=2048 -UBATCH_SIZE=512 +BATCH_SIZE=12288 +UBATCH_SIZE=4096 THREADS=8 THREADS_HTTP=8 diff --git a/scripts/serve-llama-frank-large.sh b/scripts/serve-llama-frank-large.sh index a008861..48a3bf2 100755 --- a/scripts/serve-llama-frank-large.sh +++ b/scripts/serve-llama-frank-large.sh @@ -35,8 +35,8 @@ fi : "${N_GPU_LAYERS:=all}" : "${CACHE_TYPE_K:=q8_0}" : "${CACHE_TYPE_V:=f16}" -: "${BATCH_SIZE:=2048}" -: "${UBATCH_SIZE:=512}" +: "${BATCH_SIZE:=12288}" +: "${UBATCH_SIZE:=4096}" : "${THREADS:=8}" : "${THREADS_HTTP:=8}"