fix the build issue on gpu mode for win

* Added accuracy ngraph op test=develop

* fixed name type test=develop

* add skip send.recv test=develop

* enhanced print message. test=develop

* rerun ci. test=develop

test=develop

size_t is an unsigned integer, with a conversion rank
 larger than int, therefore in the following expression
 the int value was promoted to size_t, making it a
 subtraction of unsigned values. The result of such
 a subtraction is also an unsigned value.

Revert "Async double buffered py reader"

test=develop

Async double buffered py reader

* chang the timeout of test_pe_resnet
test=develop

* follow comment
test=develop

Fix unittest failure in debug mode

Ir inplace pass

optimize density_prior_box_op.h in detect model for cpu

test=develop, fast_install shell for linux and mac

* update_readme

* test=develop

To make CUDA_LAUNCH_KERNEL_HELPER support large size.

test=develop

* expose peak gpu memory API to python test=develop

* add unittest for peak gpu memory monitoring test=develop

* add pybind change test=develop

* add mutex to gpu mem usage monitor test=develop

* update benchmark flag definition file test=develop

* tweak unittest for memory monitoring test=develop

test=develop

Add box clip op

test=develop

refine bert