Your results make sense... only stats from the CPUs that are used in your simulation show up, so only when you use both CPUs (by switching in the middle) do you see both sets of stats.
Steve On Fri, Aug 13, 2010 at 12:37 AM, Iordan Alexandru <[email protected]> wrote: > > Small error on my part. I wrongfully stated in my previous e-mail, that M5's > statistics are recorded in the stats.txt file. Because I run my simulations > with the "--stats-file" parameter, that statement is not true. But do not > think that I checked the wrong file and that is my problem. I always checked > the correct file and all the problems I described in my previous e-mail are > still true. > > I still hope somebody will be able to tell me how I can get all statistics > outputted at once. > > > > --- On Wed, 8/11/10, Iordan Alexandru <[email protected]> wrote: > >> From: Iordan Alexandru <[email protected]> >> Subject: [m5-users] M5 - McPAT integration...another problem with insts >> counters and stats >> To: [email protected] >> Date: Wednesday, August 11, 2010, 11:37 PM >> >> Hello >> >> After reading a post by Richard Strong >> (http://www.mail-archive.com/[email protected]/msg02833.html) >> I added the respective statistics to my M5. However, I >> encountered some problems in using them. >> >> As you can see from Richard's post, some new stats are >> added to the BaseSimpleCPU class and some to the FullO3CPU >> class. My problem is that if I run M5 with: >> >> ./build/ALPHA_FS/m5.opt --stats-file=canneal_4cores_stats2 >> fs.py -n 4 --caches --l2cache >> >> I only get the BaseSimpleCPU stats in the stats.txt file, >> and if I run it with: >> >> ./build/ALPHA_FS/m5.opt --stats-file=canneal_4cores_stats2 >> fs.py -n 4 --detailed --caches --l2cache >> >> I only get the FullO3CPU ones. >> >> The real funny thing is what happens when I run M5 with: >> >> ./build/ALPHA_FS/m5.opt --stats-file=canneal_4cores_stats2 >> fs.py -n 4 --detailed --caches --l2cache -F 5000000000 >> >> and I use a "switchcpu" command from the terminal, before >> launching my benchmark. In this case, all the stats are >> included in stats.txt, but the BaseSimpleCPU ones are all >> recorded with the value 0. >> >> I would really like to find a way to have all stats >> correctly included in the output of M5. Has anyone an idea >> on how to do that given my above mentioned situation? >> >> 10x in advance >> >> >> >> _______________________________________________ >> m5-users mailing list >> [email protected] >> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >> > > > > _______________________________________________ > m5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
