Update to Cutlass 3.5

cutlass @ 7d49e6c7

-Subproject commit bbe579a9e3beb6ea6626d9227ec32d0dae119a49
+Subproject commit 7d49e6c7e2f8896c47f586706e67e1fb215529dc

csrc/flash_attn/src/flash_bwd_kernel.h

@@ -4,7 +4,7 @@
 
 #pragma once
 
-#include <cute/algorithm/copy.hpp>
+#include <cute/tensor.hpp>
 
 #include <cutlass/cutlass.h>
 #include <cutlass/array.h>

csrc/flash_attn/src/flash_bwd_preprocess_kernel.h

@@ -4,7 +4,7 @@
 
 #pragma once
 
-#include <cute/algorithm/copy.hpp>
+#include <cute/tensor.hpp>
 
 #include <cutlass/cutlass.h>
 #include <cutlass/array.h>

csrc/flash_attn/src/flash_fwd_kernel.h

@@ -4,7 +4,7 @@
 
 #pragma once
 
-#include <cute/algorithm/copy.hpp>
+#include <cute/tensor.hpp>
 
 #include <cutlass/cutlass.h>
 #include <cutlass/array.h>

csrc/flash_attn/src/kernel_traits.h

@@ -4,7 +4,7 @@
 
 #pragma once
 
-#include "cute/algorithm/copy.hpp"
+#include "cute/tensor.hpp"
 
 #include "cutlass/cutlass.h"
 #include "cutlass/layout/layout.h"

csrc/flash_attn/src/rotary.h

@@ -4,7 +4,7 @@
 
 #pragma once
 
-#include <cute/algorithm/copy.hpp>
+#include <cute/tensor.hpp>
 
 #include "utils.h"
 

csrc/flash_attn/src/utils.h

@@ -14,8 +14,7 @@
 #include <cuda_bf16.h>
 #endif
 
-#include <cute/algorithm/copy.hpp>
-#include <cute/algorithm/gemm.hpp>
+#include <cute/tensor.hpp>
 
 #include <cutlass/array.h>
 #include <cutlass/cutlass.h>