From: Jeff Bolz <redacted>
Date: Sat, 13 Dec 2025 15:19:51 +0000 (-0600)
Subject: llama_context: synchronize before reallocating output buffer (#17974)
X-Git-Tag: upstream/0.0.7446~59
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=5266379bcae74214af397f36aa81b2a08b15d545;p=pkg%2Fggml%2Fsources%2Fllama.cpp

llama_context: synchronize before reallocating output buffer (#17974)
---

diff --git a/src/llama-context.cpp b/src/llama-context.cpp
index 2692297d..9914b327 100644
--- a/src/llama-context.cpp
+++ b/src/llama-context.cpp
@@ -1318,6 +1318,7 @@ uint32_t llama_context::output_reserve(int32_t n_outputs) {
             // This doesn't happen often, but may be annoying in some cases (like the HellaSwag benchmark)
             LLAMA_LOG_INFO("%s: reallocating output buffer from size %.02f MiB to %.02f MiB\n", __func__, prev_size / 1024.0 / 1024.0, new_size / 1024.0 / 1024.0);
 #endif
+            synchronize();
             buf_output = nullptr;
             logits = nullptr;
             embd = nullptr;