From bd8422dbe7aeb87490d7d8de29f146f6e7468b9b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Johannes=20G=C3=A4=C3=9Fler?= <johannesg@5d6.de>
Date: Mon, 3 Jun 2024 22:10:28 +0200
Subject: [PATCH] fix out-of-bounds write

---
 ggml-cuda/mmq.cuh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ggml-cuda/mmq.cuh b/ggml-cuda/mmq.cuh
index 62df7c45f..dda1e341e 100644
--- a/ggml-cuda/mmq.cuh
+++ b/ggml-cuda/mmq.cuh
@@ -1124,7 +1124,7 @@ static __global__ void mul_mat_q(
         for (int i0 = 0; i0 < mmq_y; i0 += WARP_SIZE) {
             const int i = blockIdx.x*mmq_y + i0 + threadIdx.x;
 
-            if (need_check && i > ne0) {
+            if (need_check && i >= ne0) {
                 continue;
             }