From 94c1ae86689bd8861382b17deacaa54cfdaf09d5 Mon Sep 17 00:00:00 2001
From: slaren <slarengh@gmail.com>
Date: Wed, 24 Jan 2024 12:48:14 +0100
Subject: [PATCH] llama : pre-allocate input tensors in a separate buffer
 (llama/5100)

---
 ggml-alloc.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/ggml-alloc.c b/ggml-alloc.c
index 89b85d3..60141a3 100644
--- a/ggml-alloc.c
+++ b/ggml-alloc.c
@@ -109,8 +109,8 @@ void ggml_tallocr_alloc(ggml_tallocr_t alloc, struct ggml_tensor * tensor) {
         if (block->size >= size) {
             best_fit_block = alloc->n_free_blocks - 1;
         } else {
-            fprintf(stderr, "%s: not enough space in the buffer (needed %zu, largest block available %zu)\n",
-                    __func__, size, max_avail);
+            fprintf(stderr, "%s: not enough space in the buffer to allocate %s (needed %zu, largest block available %zu)\n",
+                    __func__, tensor->name, size, max_avail);
             GGML_ASSERT(!"not enough space in the buffer");
             return;
         }