diff options
| author | Kawrakow <iwankawrakow@gmail.com> | 2024-12-10 12:26:40 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-12-10 12:26:40 +0100 |
| commit | 361174ee6aee8792b4fbb227b9bc328bf9bd6eb9 (patch) | |
| tree | 887c0cb982664604ef977cdac27657d7c61c88a4 /ggml/include/ggml.h | |
| parent | 3ec193b4856df8e5827b83a8c7686e8498c5e5b8 (diff) | |
Q6_K_R4 (#130)
* Adding q6_k_r4
* q6_k_r4: 1st functional AVX2 version
* q6_k_r4: AVX2 and simple Zen4
"Simple" as in processing 4 instead of 8 rows at once.
On Zen4 we get PP-512(LLaMA-3.1-8B) = 238.3 t/s vs
195.2 t/s for Q6_K. TG-128 @ 1 thread is 7.94 t/s
vs 5.38 t/s for Q6_K.
* q6_k_r4: 1st NEON version
PP-512(LLaMA-3.1-8B) = 78 t/s vs 57.6 t/s for q6_K.
TG-128 is slightly lower rthan q6_K for low number of threads,
becomes very slightly better at 8 threads.
* q6_k_r4: slightly faster NEON
PP-512(LLaMA-3.1-8B) = 83.25 t/s
* q6_k_r4: slightly faster Zen4
238.3 t/s -> 243.2 t/s
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'ggml/include/ggml.h')
| -rw-r--r-- | ggml/include/ggml.h | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/ggml/include/ggml.h b/ggml/include/ggml.h index 38e01dd5..6486407f 100644 --- a/ggml/include/ggml.h +++ b/ggml/include/ggml.h @@ -413,6 +413,7 @@ extern "C" { GGML_TYPE_Q5_0_R4 = 206, GGML_TYPE_Q8_0_R4 = 208, GGML_TYPE_Q4_K_R4 = 212, + GGML_TYPE_Q6_K_R4 = 214, GGML_TYPE_IQ4_NL_R4 = 220, GGML_TYPE_IQ4_XS_R4 = 223, GGML_TYPE_Q6_0_R4 = 233, @@ -480,6 +481,7 @@ extern "C" { GGML_FTYPE_MOSTLY_Q8_0_R4 = 207, // except 1d tensors GGML_FTYPE_MOSTLY_Q5_0_R4 = 208, // except 1d tensors GGML_FTYPE_MOSTLY_Q4_K_R4 = 212, // except 1d tensors + GGML_FTYPE_MOSTLY_Q6_K_R4 = 214, // except 1d tensors GGML_FTYPE_MOSTLY_IQ4_NL_R4 = 219, // except 1d tensors GGML_FTYPE_MOSTLY_IQ4_XS_R4 = 222, // except 1d tensors GGML_FTYPE_MOSTLY_Q6_0_R4 = 227, // except 1d tensors |