...
A detailed design document about this will be created and is worked on separately (Thread-safety in MXNet).
Informative errors
Documentation and versioning of the API
...
// MultiplyOperation.h
class Tensor;
class MultiplyOperation {
public:
void setA(const Tensor &a);
void setB(const Tensor &b);
Tensor getResult() const;
private:
class MultiplyOperationPrivate;
std::unique_ptr<MultiplyOperationPrivate> p;
};
// MultiplyOperation.cpp
#include <MultiplyOperation.h>
void MultiplyOperation::setA(const Tensor &a) {
p->a = a;
}
void MultiplyOperation::setB(const Tensor &b) {
p->b = b;
}
Tensor MultiplyOperation::getResult() const {
return p->getResultImpl();
}
class MultiplyOperationPrivate {
public:
Tensor a, b;
Tensor getResultImpl() const;
};
// MultiplyOperation_CUDA.cpp
Tensor MultiplyOperationPrivate::getResultImpl() const {
return CUDA_CALL(multiply(a, b));
}
// MultiplyOperation_MKL.cpp
Tensor MultiplyOperationPrivate::getResultImpl() const {
return mkl_multiply(a, b);
} |
Semantic versioned API with no additional dependencies
...
