From 504fb890d90ec27e5f4822b7bd772fa94d4d6aac Mon Sep 17 00:00:00 2001
From: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Date: Sun, 11 May 2025 12:22:19 +0300
Subject: Revert "Fix race in the CUDA DeepSeek FA kernel (#406)"

This reverts commit 36e6e888b75ae93fb5aac212bb0e147d8379ae23.
I should have tested. We get NaNs.
---
 ggml/src/ggml-cuda/fattn-new-mma.cu | 2 --
 1 file changed, 2 deletions(-)

diff --git a/ggml/src/ggml-cuda/fattn-new-mma.cu b/ggml/src/ggml-cuda/fattn-new-mma.cu
index 8da96370..d1484451 100644
--- a/ggml/src/ggml-cuda/fattn-new-mma.cu
+++ b/ggml/src/ggml-cuda/fattn-new-mma.cu
@@ -898,8 +898,6 @@ static __device__ __forceinline__ void flash_attn_ext_f16_process_tile(
             KQ_crs += __shfl_xor_sync(0xFFFFFFFF, KQ_crs, offset, WARP_SIZE);
         }
 
-        __syncthreads();
-
         // Write back combined meta data:
 #pragma unroll
         for (int imeta = 0; imeta < nmeta; ++imeta) {
-- 
cgit v1.2.3