fix clang_format error on windows

* refine copy actor

* using unix path separator

* separate Init for CopyHdActor and CopyCommNetActor

* refine CopyHdActor with different stages

* refine CopyCommNetActor with different stages

* remove useless code

* remove useless code

* using cuda stream in thread context

* using OF_SET_MSG_HANDLE

* remove useless code

* remote regst warpper

* fix blob ptr in RemoteRegstWarpper

* fix memory leak

* add model diff accumulate actor

* implement model diff accumulate actor

* fix style

* alignment

* update innerproduct

* add test of inner_product

* update inner_product_test

* mod inner_product_kernel by KernelUtil

* fix issues

* rename BnInOp2BlobPtr

* mod BuildKernel function

* fix issue

* fix issue

* merge Create Blob and Create BnInOp2BlobPtr

* add blank line and rename variable

* update

* add memcpy kinds:(h2d, d2d, d2h) to Memcpy function

* copy hd kernel

* update copy hd kernel

* add memcpy kinds:(h2d, d2d, d2h) to Memcpy function

* fix typo

* fix typo

cuda util

* add cblas_template

* update cblas_gemm

* add CBLAS_ORDER

* update blas

* update dot product

* add cublas_template

* mod gemv and gemm for column major in cublas

export LD_LIBRARY_PATH=/opt/intel/lib/intel64_lin:/opt/intel/mkl/lib/intel64:/opt/oneflow/caffe/lib:/opt/intel/lib/intel64_lin:/opt/intel/mkl/lib/intel64:/usr/local/cuda-8.0/lib64/

* update innerproduct

* add test of inner_product

* update inner_product_test

* mod inner_product_kernel by KernelUtil

* fix issues

* rename BnInOp2BlobPtr

* mod BuildKernel function

* fix issue

* fix issue

* merge Create Blob and Create BnInOp2BlobPtr

* add blank line and rename variable

* update

PersistentCircularLineReader

* add cblas_template

* update cblas_gemm

* add CBLAS_ORDER

* update blas

* update dot product

* add cublas_template

* mod gemv and gemm for column major in cublas

* Added a demo math class

* Change inherit-mode to Singleton

* update master

* fixed cblas template

* fix kernelUtil cu issues

* fixed issues, added other blas templates

* fixed issues

* some parameters should be const

* add cblas_template

* update cblas_gemm

* add CBLAS_ORDER

* update blas

* update dot product

* add cublas_template

* add cblas_template

* update cblas_gemm

* add CBLAS_ORDER

* update blas

* update dot product

remote register warpper

* total_piece_num

* fw

* fix for review

* code style

* fix bug

* fix bug for data loader

* kStart in FW for data loader

* fix bug