From a64a81a2946bffa8f108fd3476565fccb885820e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Johannes=20G=C3=A4=C3=9Fler?= <johannesg@5d6.de>
Date: Mon, 10 Jun 2024 09:17:58 +0200
Subject: [PATCH] fix writeback returning too early

---
 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 41fe2d814..62111f376 100644
--- a/ggml-cuda/mmq.cuh
+++ b/ggml-cuda/mmq.cuh
@@ -1344,7 +1344,7 @@ static __device__ __forceinline__ void mmq_write_back_mma(const float * __restri
             const int j = blockIdx.y*mmq_x + j0 + mma_C::get_j(l);
 
             if (j >= ne1) {
-                return;
+                continue;
             }
 
             const int i = blockIdx.x*mmq_y + i0 + mma_C::get_i(l);