Re: [Gems-users] GEMS simulation results

[Dan Gibson <degibson@xxxxxxxx>]

Are caches warm? Are caches large? 8m instructions might not be sufficient, even for warmup. Are the workloads warmed up on disk and in the TLBs? Are you simulating within the execution of the binary, or does the simulation begin in the OS when the binary is launched?

Mike Marty wrote:

Something is wrong.  I the ruby cycles was 21 million?

How much memory does your host P4 machine have?

CPI is very high.  What is the target memory latency set to?

--Mike


  

We simulated spec 2000 benchmark (mgrid) on sarek solaris 9 op. system with 8 processor(ultra sparc III).
our simulation environment is as below:
simics 2.2.19 and gems 1.3.1

We used both ruby and opal in our simulation and we got the following results which seems there are some mistakes.
First of all, our simulation took more than 10 hours on a P4  host machine for 8 million instruction.
When we dumped ruby stats, we took these results:
ruby cycles: 21.613.976
instructions executed :8.440.784
average cycles per inst: 20.5
We cant realize where we have done mistake.
Thanks.



__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
    

_______________________________________________
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.