CUTLASS 2.1 (#83)

CUTLASS 2.1 contributes:
- BLAS-style host-side API added to CUTLASS Library
- Planar Complex GEMM kernels targeting Volta and Turing Tensor Cores
- Minor enhancements and bug fixes

include/cutlass/arch/mma.h

@@ -30,7 +30,9 @@
 
 #include "cutlass/array.h"
 #include "cutlass/numeric_types.h"
+
 #include "cutlass/gemm/gemm.h"
+#include "cutlass/arch/arch.h"
 
 /////////////////////////////////////////////////////////////////////////////////////////////////
 
@@ -49,6 +51,11 @@ struct OpMultiplyAddSaturate;
 
 /////////////////////////////////////////////////////////////////////////////////////////////////
 
+/// Tag indicating the complex multiply-add operation
+struct OpMultiplyAddComplex;
+
+/////////////////////////////////////////////////////////////////////////////////////////////////
+
 /// Tag indicating the inner product is defined by (XOR, POPC)
 struct OpXorPopc;