This bug occasionally causes dead lock in test_RecurrentGradientMachine
In general, conditional_variable::notify should be used together with mutex for changing condition.

* findLastSet function: size_t could be  uint, ulong, ulonglong
* add default cuda 7.0 in flags.cmake

Change-Id: Ica9d09520e6a1468e5fc97e2c29c1ea7c73dafbb

Change-Id: Ib7f7d3c65c9850b8420f9a4799835a7ea03572bd

ISSUE=4586495

git-svn-id: https://svn.baidu.com/idl/trunk/paddle@1408 1ad973e4-5ce8-4261-8a94-b56d1f490c56