From: Georgi Gerganov <redacted>
Date: Thu, 27 Jul 2023 08:00:54 +0000 (+0300)
Subject: metal : disable graph concurrency optimization due to bug (#2413)
X-Git-Tag: gguf-v0.4.0~389
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=1a941869cbef8e9cc351a6c6987e4ae3b0f021f7;p=pkg%2Fggml%2Fsources%2Fllama.cpp

metal : disable graph concurrency optimization due to bug (#2413)
---

diff --git a/llama.cpp b/llama.cpp
index 024af99a..9a8ecdcf 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1722,9 +1722,10 @@ static bool llama_eval_internal(
 
 #ifdef GGML_USE_METAL
     if (lctx.ctx_metal && N == 1) {
-        if (!ggml_metal_if_optimized(lctx.ctx_metal)) {
-            ggml_metal_graph_find_concurrency(lctx.ctx_metal, gf);
-        }
+        // TODO: disabled until #2413 is resolved
+        //if (!ggml_metal_if_optimized(lctx.ctx_metal)) {
+        //    ggml_metal_graph_find_concurrency(lctx.ctx_metal, gf);
+        //}
         ggml_metal_set_n_cb     (lctx.ctx_metal, n_threads);
         ggml_metal_graph_compute(lctx.ctx_metal, gf);
         ggml_metal_get_tensor   (lctx.ctx_metal, cur);