Guoqiang,
Simulation will continue even with decode failures, since Opal checks
at retirement time that each instruction correctly updated architectural
state.
If these decode failures (e.g. deviations) are a signficant portion of
all dynamic instructions it will show up as a degradation to the "%
correct" statistic that Opal prints out at the end of simulations. Then
you have the option of implementing those instructions (see previous
posts) in Opal to drive up its functional correctness.
My hypothesis is that the slowdown for syncing up with Simics on
deviations is not that significant. I believe it is a small overhead on
top of the "checking with Simics" overhead that is always present for Opal
simulations.
Luke
On Fri, 18 Apr 2008, Guoqiang Yang wrote:
Hi Xuan Qi:
Thank you so much for your reply. But I did search the mailing list
before I post my question.
And I think currently my biggest concern is that: What do we have to do
with those decode failure? Will the simulation continue if we just
ignore those messages? It seems to me that I have met at least ten
instructions that can NOT be decoded, and it certainly takes lots of time.
Many thanks,
Guoqiang Yang
_______________________________________________
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.
|
![[Computer Systems Lab]](http://www.cs.wisc.edu/pics/csl_logo.gif){kind=logo}