Re: [Gems-users] Obtaining total execution time of runs

[Mike Marty <mikem@xxxxxxxxxxx>]

> hey mike, thanks for your reply,
> the reason i ask is because i'm getting quite different number for
> consecutive runs between the same program - a difference of 3million
> ruby_cycles. any idea of the top of your head why that might be the case?
> thanks
> sid
>

Is the g_RANDOM_SEED different for each run (see the stats)?  We use it to
inject randomness into the system.  See the "Addressing Workload
Variability in Architectural Simulations" paper by Alaa Alameldeen et. al.

You can view this paper by visiting
http://www.cs.wisc.edu/multifacet/papers/

If you don't want this, just ensure you use the same RANDOM_SEED between
runs.

--Mike