devtools/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2026-04-10 11:34:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

Misc improvements and optimizations

Browse Source
This commit is contained in:
Benoit Steiner
2015-07-01 13:59:11 -07:00
parent 4ed213f97b
commit 1e911b276c
3 changed files with 8 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
unsupported/Eigen/CXX11/src/Tensor/TensorConvolution.h
View File

@@ -510,7 +510,8 @@ struct TensorEvaluator<const TensorConvolutionOp<Indices, InputArgType, KernelAr
Scalar* local = (Scalar*)m_device.allocate(kernel_sz);
typedef TensorEvalToOp<const KernelArgType> EvalTo;
EvalTo evalToTmp(local, m_kernelArg);
internal::TensorExecutor<const EvalTo, Device>::run(evalToTmp, m_device);
const bool PacketAccess = internal::IsVectorizable<Device, KernelArgType>::value;
internal::TensorExecutor<const EvalTo, Device, PacketAccess>::run(evalToTmp, m_device);
m_kernel = local;
m_local_kernel = true;
@@ -815,7 +816,8 @@ struct TensorEvaluator<const TensorConvolutionOp<Indices, InputArgType, KernelAr
Scalar* local = (Scalar*)m_device.allocate(kernel_sz);
typedef TensorEvalToOp<const KernelArgType> EvalTo;
EvalTo evalToTmp(local, m_kernelArg);
internal::TensorExecutor<const EvalTo, GpuDevice>::run(evalToTmp, m_device);
const bool PacketAccess = internal::IsVectorizable<GpuDevice, KernelArgType>::value;
internal::TensorExecutor<const EvalTo, GpuDevice, PacketAccess>::run(evalToTmp, m_device);
m_kernel = local;
m_local_kernel = true;