direct_convolution.cuh 444 Bytes
Newer Older
Chao Liu's avatar
Chao Liu committed
1
2
3
4
5
6
7
8
9
10
11
12
#pragma once
#include "device_tensor.cuh"

template <class TFloat, int NBlockDim>
__global__ void direct_convolution(DeviceTensorDescriptor in_desc,
                                   TFloat* const in,
                                   DeviceTensorDescriptor wei_desc,
                                   TFloat* const wei,
                                   DeviceTensorDescriptor out_desc,
                                   TFloat* out)
{
}