From b19af3643f68d2d1b9b27dd7cfd829f9dd33928e Mon Sep 17 00:00:00 2001
From: Pierrick HYMBERT <pierrick.hymbert@gmail.com>
Date: Fri, 22 Mar 2024 07:03:14 +0100
Subject: [PATCH] llama_model_loader: be sure the model mappings has enough
 capacity before allocating backend buffer

---
 llama.cpp | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/llama.cpp b/llama.cpp
index 891892f25..3e0aec8f6 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -5135,9 +5135,11 @@ static bool llm_load_tensors(
     ml.done_getting_tensors();
 
     ml.init_mappings(true, &model.mlock_mmaps);
+    model.mappings.reserve(ml.mappings.size());
 
     // create the backend buffers
     std::vector<std::pair<ggml_context *, std::unordered_map<uint32_t, ggml_backend_buffer_t>>> ctx_bufs;
+    ctx_bufs.reserve(ctx_map.size());
 
     // Ensure we have enough capacity for the maximum backend buffer we will potentially create
     size_t n_max_backend_buffer = ctx_map.size() * ml.files.size();