public static long GetConvolutionForwardWorkspaceSize(IAllocator allocator, DNNConvolutionFwdAlgo algo, Cpu.ConvolutionDesc2d cd, TensorShape x, TensorShape w, TensorShape y)
{
if (!(allocator is CudaAllocator))
{
throw new InvalidOperationException("allocator must be a CUDA allocator");
}
var cudaAllocator = (CudaAllocator)allocator;
using (var dnn = cudaAllocator.Context.DNNForDevice(cudaAllocator.DeviceId))
{
var convDesc = GetConvDescriptor(cd, x.ElementType);
using (var xDesc = GetDescriptor(x))
using (var wDesc = GetFilterDescriptor(w))
using (var yDesc = GetDescriptor(y))
{
return(dnn.Value.GetConvolutionForwardWorkspaceSize(
xDesc,
wDesc,
convDesc,
yDesc,
(cudnnConvolutionFwdAlgo)algo));
}
}
}
public static void ConvForward(DNNConvolutionFwdAlgo algo, Cpu.ConvolutionDesc2d cd, CudaStorage workspace, Tensor x, Tensor w, Tensor y)
{
using (var dnn = CudaHelpers.TSContextForTensor(x).DNNForTensor(x))
{
var convDesc = GetConvDescriptor(cd, x.ElementType);
using (var workspacePtr = new CudaDeviceVariable <byte>(workspace.DevicePtrAtElement(0), false, workspace.ByteLength))
using (var xPtr = GetDeviceVar(x))
using (var wPtr = GetDeviceVar(w))
using (var yPtr = GetDeviceVar(y))
using (var xDesc = GetDescriptor(x))
using (var wDesc = GetFilterDescriptor(w))
using (var yDesc = GetDescriptor(y))
{
dnn.Value.ConvolutionForward(1,
xDesc, xPtr,
wDesc, wPtr,
convDesc,
(cudnnConvolutionFwdAlgo)algo,
workspacePtr,
0,
yDesc, yPtr);
}
}
}
