Re: [Gems-users] ruby configuration and FFT results


Date: Fri, 14 Dec 2007 11:54:40 -0600
From: Dan Gibson <degibson@xxxxxxxx>
Subject: Re: [Gems-users] ruby configuration and FFT results
How is your memory system configured? (look in $GEMS/ruby/config/rubyconfig.defaults)

In Simics+Ruby configurations, the scalability is determined by:
1) The workload (SPLASH-2 scales OK for small configurations)
2) The ruby setup (e.g., cache sizes, coherence protocol, etc)

The commands you mentioned below aren't performance options: they are *necessary* for proper, accurate functionality of ruby. Have a look at the Simics reference manual to find out what they do -- to summarize: they allow ruby to realistically model caches.


Regards,
Dan

nikiforg@xxxxxxxxxx wrote:
Hi all,
In the wiki documentation, it is declared as important to follow some steps as:

  simics> instruction-fetch-mode instruction-fetch-trace
  simics> istc-disable
  simics> dstc-disable
  simics> cpu-switch-time 1

configuring ruby as above, the results of the SPLASH2 FFT running on
MOSI_SMP_Bcast are not good! There is no scalability! System with 2 CPUs is
faster than any aother system! If I do not follow the steps above, everything
seems OK(scalable up to 8 CPUs).

Something else I want to ask is about L1 ang L2 caches and their size! Where can
I configure them!

Nikiforos.


_______________________________________________
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.


--
http://www.cs.wisc.edu/~gibson [esc]:wq!

[← Prev in Thread] Current Thread [Next in Thread→]