* fix cdn issue, test=develop

* fix cdn issue, test=develop

2. refine trt code
test=develop

* Support Sync Batch Norm.
* Note, do not enable it in one device.

Usage:

build_strategy = fluid.BuildStrategy()
build_strategy.sync_batch_norm = True
binary = fluid.compiler.CompiledProgram(tp).with_data_parallel(
        loss_name=loss_mean.name,
        build_strategy=build_strategy)

* Upgrade MKLDNN to v0.18-rc and fix issue caused by lib/lib64

Upgrade MKLDNN to v0.18-rc
Also fix the issue during upgrade

test=develop

* Rebase MKLDNN to rls-v0.18 branch

Some issues in v0.18-rc which caused INT8 conv op unit test failure was fixed
in rls-v0.18 branch

test=develop

* Upgrade MKLDNN from v0.18rc to formal v0.18 tag

test=develop

* Fix the windows compile issue.

test=develop

* staged.

* polish code

* polish code. test=develop

* polish code. test=develop

* api change. test=develop

* fix default value. test=develop

* fix default value. test=develop

test=develop

* staged.

* polish code

* polish code. test=develop

* polish code. test=develop

* api change. test=develop

* fix default value. test=develop

* fix default value. test=develop

test=develop

test=develop

* Revert "Optimze Gelu with MKL Erf function (#15770)"

This reverts commit 676995c8.

* test=develop

* Optimize for gelu operator

* Set up the low accuracy mode of MKL ERF function.

test=develop

* Only enable MKLML ERF when OS is linux

* Use the speical mklml version included vmsErf function to verify gelu mkl kernel.

test=develop

* Add the CUDA macro to avoid NVCC's compile issue.

test=develop

* Add the TODO comments for mklml library modification.

test=develop

* Clean Code

test=develop

* Add the comment of marco for NVCC compiler.

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop

test=develop