From ccec00939a30aa7762a232ac4dcadba985ef9ee4 Mon Sep 17 00:00:00 2001 From: Kawrakow Date: Tue, 3 Dec 2024 06:15:29 +0100 Subject: Q8_0_R4 (#120) * Adding q8_0_r4 We get PP-512(LLaMA-3.1-8B) = 268 t/s on a Ryzen-7950X compared to 175.6 t/s for Q8_0. * q8_0_r4: NEON We get PP-512(LLaMA-3.1-8B) = 112.6 t/s on M2-Max. * q8_0_r4: Zen4 matrix-vector specialization --------- Co-authored-by: Iwan Kawrakow --- include/llama.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include') diff --git a/include/llama.h b/include/llama.h index fc034b3a..5e935533 100644 --- a/include/llama.h +++ b/include/llama.h @@ -181,6 +181,7 @@ extern "C" { LLAMA_FTYPE_MOSTLY_IQ4_KSS = 148, // except 1d tensors // LLAMA_FTYPE_MOSTLY_Q4_0_R4 = 202, // except 1d tensors + LLAMA_FTYPE_MOSTLY_Q8_0_R4 = 207, // except 1d tensors LLAMA_FTYPE_MOSTLY_IQ4_NL_X4 = 225, // except 1d tensors LLAMA_FTYPE_GUESSED = 1024, // not specified in the model file -- cgit v1.2.3