Closed jialeiwang closed 10 years ago
Couple of super broad comments:
On Cornell server, it is 149.73s for testing with GPU enabled(including EI/gradEI consistency check and optimization unit test), 92.61s for testing with GPU enabled(without optimization unit test), and 89.85s without GPU.
~1 minute for the optimization test? Is there any way to make this run faster? Like fewer MC iterations, fewer multistarts or GD steps, etc?
lmk when the other changes are up
couple more cleanup things
few more comments + a test for the analytic case as we discussed in person
one more thing, please update the changelog, thanks!
move the changelog line and shipit!
***** PEOPLE ***** Primary reviewer: @suntzu86
Reviewers: @sc932
***** DESCRIPTION ** Branch Name: jialei_gh270_connect_cudaEI_to_optimization_algorithm Ticket(s)/Issue(s): Closes #270
***** TESTING DONE ***** make test cpplint.py