Re: [Gems-users] CPU accesses to L2 cache

[Dan Gibson <degibson@xxxxxxxx>]

Are you sure that you're not merely observing that all other cpus 
(cpu1-cpu7) are in the solaris idle loop?
cpu0 is probably active only to respond to your keypresses and to fork 
your processes (ps -A, prstat, etc).
Also, did you issue instruction-fetch-mode instruction-fetch-trace, 
istc-disable, dstc-disable to Simics?
Try running a mulitthreaded benchmark, and be sure to use 
processor_bind() to get all the processors busy.
Regards,
Dan

Daniele Bordes wrote:

> Dear Gems Users, I tried to run some simple simulations using Simics + Ruby.
> The simulated machine is a 8-Sparc Processors with Solaris 9.
> L2-cache is a D-NUCA cache with 256 banks.
> I have noticed this fact (which seems to me strange) in ruby statistics:
>
>
> In Ruby statistics, CPU0 number of accesses on L2-cache is MUCH
> greater than other CPU's ones.
>
> The fact is that I have set Solaris fsflush mechanism frequency to be
> very very slow (by changing a parameter inside /etc/system and
> rebooting), and if I type "ps -A" from simulated machine, I see only
> this processes:
>
> sched
> init
> pageout
> fsflush
> sh
> ps
>
>
> And if I type "prstat" inside simulated machine, I see this:
>
> STATE     CPU    PROCESS
> cpu3       0.0%     prstat
> sleep      0.0%     sh
> sleep      0.0%     init
>
>
> So, in practice, the simulated machine seems to be "Idle", but there
> are many accesses of CPU0 to L2 cache.
>
> Since I am not a Solaris expert, does anybody know if this is correct?
>
> I am sorry to be a nuisance.
> Thank you very much.
> _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
>
>
>