Look for notIdleFraction is the source code an try to identify where it’s >1.
Ali On Apr 21, 2014, at 11:17 AM, Yuchen Hou <[email protected]> wrote: > Thank you Ali. I am not sure where to find the variable and the operations > executed upon it. Can you give me some pointer on how to effectively pin down > this bug in the large amount of source files? > > ---------- Forwarded message ---------- > From: Ali Saidi <[email protected]> > To: gem5 users mailing list <[email protected]> > Cc: > Date: Sun, 20 Apr 2014 12:05:18 -0500 > Subject: Re: [gem5-users] Negative CPU not_idle_fraction > It’s very possible there is a bug here and a place where the idle vs. > not-idle isn’t handled. When the cpu goes idle it subtracts 1 from a variable > and when it resumes it adds one. If one subtraction is missing it will end up > being >=2 which is the cause of the issue. You should be able to put some > asserts around every increment/decrement of the variable and see when it is > happening. > > Ali > > On Apr 19, 2014, at 10:48 PM, Yuchen Hou <[email protected]> wrote: > > > Hi folks, > > I need some help here. Although it is hard to believe, I do get negative > > CPU not_idle_fraction in my stats.txt once in a while (with the > > corresponding CPU idle_fraction > 1! ). > > I am using alpha full system simulation, running a 12-thread application I > > made with various number of CPUs (from 1 to 16). Both atomic and timing CPU > > type have this problem, detailed CPU does not. No matter how many CPUs I > > have in the simulation, at most 1 CPU has this problem. Have you > > experienced this problem? Do you have any idea about these strange stats? > > Thanks for your help! > > Yuchen > > ______________________________ > _________________ > > gem5-users mailing list > > [email protected] > > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users > > _______________________________________________ > gem5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
_______________________________________________ gem5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
