Both of these things should be reasonably easy to implement. It's really just a matter of how often PrintStats() is called and how much summation you do inside of that code (and perhaps a flag somewhere that switches between per-epoch and overall only stats. It would be kind of a pain to implement both at the same time since you'd have to keep the overall totals while the per-epoch counters are reset.
I guess our use case has always been that we use the results output files (the CSV outputs) and then just do the post processing in some other application. If you want to try to take a look at it, I'd happily accept a patch -- I'm preoccupied with trying to wrap up some other work right now so I won't really have time to implement this myself. Also, I don't think this is quite the proper venue for discussing DRAMSim2 only issues as it's not our mailing list. Could you post an issue on our github issue tracker page and we can continue the discussion there? -Paul On Sat, Sep 15, 2012 at 12:57 AM, Alireza Haghdoost <[email protected]>wrote: > Hi Paul > > We are all users of full system simulator and only system level statistics > of the memory system interesting for us. However, DRAMSim provides > rank/bank based statistics. I think some body else also asked in the > mailing list the same issue about power consumption. I would suggest to > include system level statistics in addition to rank/bank based statistics > in final .log file. For example, "average power per simulation epoch" is > more interesting than "average power per rank per simulation epoch". If you > mind, I can take look at code and patch this change. > Moreover, I think if the simulation take more time than one epoch final > .log file includes the statistics for multiple simulation epochs. Again, it > would be more interesting to aggregate all of those "per epoch" stats in > one block. > > > > Thanks > Alireza >
_______________________________________________ http://www.marss86.org Marss86-Devel mailing list [email protected] https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
