46   typename ElementAccumulator_,
    88     return op(accumulator, element);
 CUTLASS_HOST_DEVICE ReduceAdd(Params params_=Params())
Constructor. 
Definition: reduction_operators.h:78
ElementAccumulator_ ElementAccumulator
Definition: reduction_operators.h:56
Definition: aligned_buffer.h:35
Defines a structure containing strides, bounds, and a pointer to tensor data. 
Mixed-precision reduction. 
Definition: reduction_operators.h:50
Definition: functional.h:46
cutlass::Array< Element, kCount > FragmentElement
Definition: reduction_operators.h:61
Params params
Parameters object. 
Definition: reduction_operators.h:70
Statically sized array of elements that accommodates all CUTLASS-supported numeric types and is safe ...
static int const kCount
Definition: reduction_operators.h:58
Boost-like numeric conversion operator for CUTLASS numeric types. 
#define CUTLASS_HOST_DEVICE
Definition: cutlass.h:89
Top-level include for all CUTLASS numeric types. 
Element_ Element
Definition: reduction_operators.h:57
Definition: reduction_operators.h:63
CUTLASS_HOST_DEVICE FragmentAccumulator operator()(FragmentAccumulator accumulator, FragmentElement element) const 
Operator. 
Definition: reduction_operators.h:82
cutlass::Array< ElementAccumulator, kCount > FragmentAccumulator
Definition: reduction_operators.h:60
Basic include for CUTLASS. 
Define basic numeric operators with specializations for Array<T, N>. SIMD-ize where possible...