diff options
| author | Kawrakow <iwankawrakow@gmail.com> | 2025-07-02 09:27:47 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-07-02 09:27:47 +0200 |
| commit | 3248a3599275c3d2a98bc41d512e1d3ba66e32ba (patch) | |
| tree | 14bdaae22a6db922aa9eb0a7131141da81dfd251 /include/llama.h | |
| parent | 46f2e5d249568b6395d17a80feb2ba9ab48156e5 (diff) | |
Adding IQ3_KS quants (#566)
* iq3_ks: basics
* iq3_ks: CUDA dequantize
* iq3_ks: CUDA mmvq
* iq3_ks: mmq
* iq3_ks: faster mmq
* iq3_ks: Zen4
* iq3_ks: AVX2 convert to q8_k_r8
This gives usPP-512 = 360 t/s.
* iq3_ks: AVX2 GEMM/GEMV
* iq3_ks: NEON GEMM/GEMV
* iq3_ks: NEON convert to q8_k_r8
This gives us PP-512 = 164 t/s.
* iq3_ks: Metal dequantize
* iq3_ks: Metal gemv - pathetic performance
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'include/llama.h')
| -rw-r--r-- | include/llama.h | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/include/llama.h b/include/llama.h index 53adeb94..6c8bff95 100644 --- a/include/llama.h +++ b/include/llama.h @@ -201,6 +201,7 @@ extern "C" { LLAMA_FTYPE_MOSTLY_IQ2_KT = 151, // except 1d tensors LLAMA_FTYPE_MOSTLY_IQ3_KT = 152, // except 1d tensors LLAMA_FTYPE_MOSTLY_IQ4_KT = 153, // except 1d tensors + LLAMA_FTYPE_MOSTLY_IQ3_KS = 154, // except 1d tensors // LLAMA_FTYPE_MOSTLY_Q4_0_R8 = 202, // except 1d tensors LLAMA_FTYPE_MOSTLY_Q8_0_R8 = 207, // except 1d tensors |