From cad50d19717f2d0b9d0115791cf8ec5c0cfd8183 Mon Sep 17 00:00:00 2001
From: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Date: Fri, 1 Sep 2023 10:44:13 +0300
Subject: [PATCH] Better use 32 thread groups for f16 x f32

---
 ggml-metal.m | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ggml-metal.m b/ggml-metal.m
index e929c4b07..8c3c64f53 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;