amirjamez / cobayn Goto Github PK

Hi, I've been trying to use more compiler passes to produce a dataset for training. However, when I run cbench by myself, each application won't last for more than 30 seconds on a dataset when default loop wrap parameter is used. Besides, the difference between the wall time or the cpu time is not as obvious as in cBench_onPandaboard_24app_5ds.csv when different configuration combinations are applied. Since we run cbench in a similar hardware environment, I guess this is because we run cbench in a different way. So, would you tell what the loop wrap was and how you obtained the dataset? Did you use __compile and __run in cbench to compile and run those 24 applications?
Thanks a lot!

Hello, I want to get 99 features of the program. How can I configure the mica.conf file