Re: [Gems-users] Reg. RUBY randomization

[niranjan soundararajan <niranjan83@xxxxxxxxx>]

Hello Arka,

Thank you for letting me know.

-Niranjan

On Mon, Aug 3, 2009 at 11:44 PM, Arkaprava Basu <basu@xxxxxxxxxxx> wrote:

Hello Niranjan,

               There is a dynamically configurable parameter called
"g_RANDOM_SEED" . For different values of this parameter ruby will
produce slightly different timing for the accesses. For same values of
"g_RANDOM_SEED" ruby will generate exactly same timings for all the
accesses ( i.e. you will see same Ruby_cycles etc) . So to get an effect
of randomization of access latencies across different runs, set
g_RANDOM_SEED to different integral values.

By un-setting MEM_FIXED_DELAY you just tell ruby to properly model the
DRAM latencies instead of returning exact same delay  for each DRAM
accesses ( it models things like bank contention etc , when
MEM_FIXED_DELAY is not set ) .  But this does NOT buy you any
randomization across different runs of the simulation.


I hope this helps,

Thanks
Arka

niranjan soundararajan wrote:
> Hello,
>
> On GEMS 2.1, does setting MEM_FIXED_DELAY = 0 alone take care of randomizing
> the memory hierarchy. Is there additional variables that need to be set?
>
> Thank you,
> Niranjan
>
>
>

> ------------------------------------------------------------------------
>
> _______________________________________________
> 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.
>
_______________________________________________
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.