Removed uncessary sync.

TimDettmers · TimDettmers · commit 4decb3cc6878 · 2023-05-02T09:38:14.000-07:00
diff --git a/csrc/kernels.cu b/csrc/kernels.cu
@@ -3145,7 +3145,6 @@ template <typename T, int BITS, int THREADS> __global__ void gemm_device(int M,
     }
     ticktock = ticktock == 0 ? 1 : 0;
 
-    __syncthreads();
     if(warp_id == (WARPS-1))
       for(int k = 0; k < batch_size_warps; k++)
       {

Original file line number	Diff line number	Diff line change
`@@ -3145,7 +3145,6 @@ template <typename T, int BITS, int THREADS> __global__ void gemm_device(int M,`
`3145`	`3145`	`}`
`3146`	`3146`	`ticktock = ticktock == 0 ? 1 : 0;`
`3147`	`3147`
`3148`		`- __syncthreads();`
`3149`	`3148`	`if(warp_id == (WARPS-1))`
`3150`	`3149`	`for(int k = 0; k < batch_size_warps; k++)`
`3151`	`3150`	`{`