From: Johannes Gäßler <redacted>
Date: Fri, 30 May 2025 19:22:03 +0000 (+0200)
Subject: CUDA: fix typo in FlashAttention code (#13926)
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=e562eece7cb476276bfc4cbb18deb7c0369b2233;p=pkg%2Fggml%2Fsources%2Fllama.cpp

CUDA: fix typo in FlashAttention code (#13926)
---

diff --git a/ggml/src/ggml-cuda/fattn-mma-f16.cuh b/ggml/src/ggml-cuda/fattn-mma-f16.cuh
index 7120053b..925f39e8 100644
--- a/ggml/src/ggml-cuda/fattn-mma-f16.cuh
+++ b/ggml/src/ggml-cuda/fattn-mma-f16.cuh
@@ -1246,7 +1246,7 @@ static __global__ void flash_attn_ext_f16(
         NO_DEVICE_CODE;
         return;
     }
-#endif __CUDA_ARCH__ == GGML_CUDA_CC_TURING
+#endif // __CUDA_ARCH__ == GGML_CUDA_CC_TURING
 
     static_assert(!mla || DKQ >= DV, "MLA needs DKQ >= DV");