From f64de08203aaee95ca755336de3e1db85d990198 Mon Sep 17 00:00:00 2001 From: Kawrakow Date: Wed, 4 Dec 2024 15:20:07 +0100 Subject: IQ4_XS_R4 (#123) * Adding iq4_xs_r4 This is a 1st working version on Zen4. We get PP-512(LLaMA-3.1-8B) = 226 t/s, so 16% slower than iq4_nl_x4. * iq4_xs_r4: WIP * iq4_xs_r4: Use AVX2 version for matrix x vector on Zen4 * iq4_xs_r4: NEON We get PP-512(LLaMA-3.1-8B) = 115.6 t/s on M2-Max, up from 68.2 t/s for iq4_xs! * DRY --------- Co-authored-by: Iwan Kawrakow --- ggml/src/ggml-quants.c | 1 + 1 file changed, 1 insertion(+) (limited to 'ggml/src/ggml-quants.c') diff --git a/ggml/src/ggml-quants.c b/ggml/src/ggml-quants.c index 94950a36..4fdd2c36 100644 --- a/ggml/src/ggml-quants.c +++ b/ggml/src/ggml-quants.c @@ -15197,6 +15197,7 @@ bool ggml_validate_row_data(enum ggml_type type, const void * data, size_t nbyte case GGML_TYPE_IQ4_KS: break; case GGML_TYPE_IQ4_KSS: break; case GGML_TYPE_IQ4_NL_X4: break; + case GGML_TYPE_IQ4_XS_R4: break; case GGML_TYPE_Q4_0_R4: break; case GGML_TYPE_Q5_0_R4: break; case GGML_TYPE_Q6_0_R4: break; -- cgit v1.2.3