From: snadampal <redacted>
Date: Fri, 26 Jan 2024 17:17:59 +0000 (-0600)
Subject: ggml : update softmax n_task calculation (#5126)
X-Git-Tag: upstream/0.0.4488~2508
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=7032f4f6349c17a8352f9f93f7d2122f45469e59;p=pkg%2Fggml%2Fsources%2Fllama.cpp

ggml : update softmax n_task calculation (#5126)

updated the n_task calculation to use max number of
threads possible. This has improved the prompt eval
performance by around 5% for DOT kernels and by
around 10% for MMLA kernels on AWS Graviton3.
---

diff --git a/ggml.c b/ggml.c
index ca98fde8..ef6fd8ca 100644
--- a/ggml.c
+++ b/ggml.c
@@ -16597,7 +16597,7 @@ static int ggml_get_n_tasks(struct ggml_tensor * node, int n_threads) {
             } break;
         case GGML_OP_SOFT_MAX:
             {
-                n_tasks = MIN(MIN(4, n_threads), ggml_nrows(node->src[0]));
+                n_tasks = MIN(n_threads, ggml_nrows(node->src[0]));
             } break;
         case GGML_OP_CONV_TRANSPOSE_1D:
             {