From: Eve <redacted>
Date: Sat, 7 Sep 2024 19:02:26 +0000 (+0000)
Subject: llamafile : disable sgemm for batch-size 1 (#9330)
X-Git-Tag: upstream/0.0.4488~803
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=e536426ded3fb4a8cd13626e53508cd92928d6c2;p=pkg%2Fggml%2Fsources%2Fllama.cpp

llamafile : disable sgemm for batch-size 1 (#9330)
---

diff --git a/ggml/src/llamafile/sgemm.cpp b/ggml/src/llamafile/sgemm.cpp
index f0988ba7..d0c2bb28 100644
--- a/ggml/src/llamafile/sgemm.cpp
+++ b/ggml/src/llamafile/sgemm.cpp
@@ -1006,6 +1006,10 @@ bool llamafile_sgemm(int64_t m, int64_t n, int64_t k, const void *A, int64_t lda
     assert(nth > 0);
     assert(ith < nth);
 
+    // only enable sgemm for prompt processing
+    if (n < 2)
+        return false;
+
     if (Ctype != GGML_TYPE_F32)
         return false;