diff options
| author | Kawrakow <48489457+ikawrakow@users.noreply.github.com> | 2023-09-01 11:15:57 +0300 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2023-09-01 11:15:57 +0300 |
| commit | e8d91589258f9204397a7ac5f9b3c857835c98f8 (patch) | |
| tree | 5909f71a59fc0822fd4310c8208655b43022e575 /ggml-metal.m | |
| parent | bce1fef328941499dc0acb76cc7fd7ac90449c2f (diff) | |
metal: somewhat faster f16 x f32 matrix multiply kernel (#2951)
* Somewhat faster f16 x f32 matrix multiply kernel
* Better use 32 thread groups for f16 x f32
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'ggml-metal.m')
| -rw-r--r-- | ggml-metal.m | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/ggml-metal.m b/ggml-metal.m index e929c4b0..8c3c64f5 100644 --- a/ggml-metal.m +++ b/ggml-metal.m @@ -840,7 +840,7 @@ void ggml_metal_graph_compute( switch (src0t) { case GGML_TYPE_F16: { - nth0 = 64; + nth0 = 32; nth1 = 1; [encoder setComputePipelineState:ctx->pipeline_mul_mat_f16_f32]; } break; |