From 59d742b00fb48a3704f86c16afbb6b4ebcde8e68 Mon Sep 17 00:00:00 2001 From: Kawrakow Date: Wed, 18 Dec 2024 13:29:25 +0100 Subject: IQ5_K_R4 (#149) * iq5_k_r4: Zen4 Much slower than the others. * iq5_k_r5: WIP * Minor * iq5_k_r4: fix AVX2 nrc_y = 1 case * iq5_k_r4: better Zen4 But TG is still slower than iq5_k * iq5_k_r4: slightly better AVX2 * iq5_k_r4: NEON --------- Co-authored-by: Iwan Kawrakow --- ggml/include/ggml.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'ggml/include') diff --git a/ggml/include/ggml.h b/ggml/include/ggml.h index 77ee0fb9..07142692 100644 --- a/ggml/include/ggml.h +++ b/ggml/include/ggml.h @@ -426,6 +426,7 @@ extern "C" { GGML_TYPE_IQ2_K_R4 = 337, GGML_TYPE_IQ3_K_R4 = 338, GGML_TYPE_IQ4_K_R4 = 339, + GGML_TYPE_IQ5_K_R4 = 340, GGML_TYPE_Q8_K_R8 = 399, GGML_TYPE_COUNT, }; @@ -502,6 +503,7 @@ extern "C" { GGML_FTYPE_MOSTLY_IQ2_K_R4 = 330, // except 1d tensors GGML_FTYPE_MOSTLY_IQ3_K_R4 = 331, // except 1d tensors GGML_FTYPE_MOSTLY_IQ4_K_R4 = 332, // except 1d tensors + GGML_FTYPE_MOSTLY_IQ5_K_R4 = 333, // except 1d tensors GGML_FTYPE_MOSTLY_Q8_K_R8 = 399, // except 1d tensors }; -- cgit v1.2.3