Created by: wulipc

目前使用的是 paddlepaddle-gpu (1.5.0.post87)， 在厂内开发机会出现以下错误：

C++ Callstacks: 
CUDNN_STATUS_INTERNAL_ERROR at [/paddle/paddle/fluid/platform/device_context.cc:217]
PaddlePaddle Call Stacks: 
0       0x7fe813e73e10p void paddle::platform::EnforceNotMet::Init<char const*>(char const*, char const*, int) + 352
1       0x7fe813e74189p paddle::platform::EnforceNotMet::EnforceNotMet(std::__exception_ptr::exception_ptr, char const*, int) + 137
2       0x7fe815f9a1a4p paddle::platform::CudnnHolder::CudnnHolder(CUstream_st* const*, paddle::platform::CUDAPlace const&) + 996
3       0x7fe815f9a470p
4         0x38c040cbe0p pthread_once + 80
5       0x7fe815f996ebp paddle::platform::CUDADeviceContext::cudnn_holder() const + 91
6       0x7fe815f99719p paddle::platform::CUDADeviceContext::cudnn_handle() const + 9
7       0x7fe814344b2ap paddle::platform::CanCUDNNBeUsed(paddle::framework::ExecutionContext const&) + 250
8       0x7fe814aa048bp paddle::operators::ConvOp::GetExpectedKernelType(paddle::framework::ExecutionContext const&) const + 219
9       0x7fe815ee819bp paddle::framework::OperatorWithKernel::ChooseKernel(paddle::framework::RuntimeContext const&, paddle::framework::Scope const&, boost::variant<paddle::platform::CUDAPlace, paddle::platform::CPUPlace, paddle::platform::CUDAPinnedPlace, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_, boost::detail::variant::void_> const&) const + 235

之前paddlecloud上也出现相同的错误， 经过系统部排查说是内核版本太低导致的，所以建议官方发布 paddle 1.5+ 的post85 版本 ， 毕竟cudnn5是可以在老内核上运行的。