On Jan 11, 2008 2:49 AM, Mladen Nikitovic <
mladen@xxxxxx> wrote:
Mike Marty wrote:
> it seems like a good approach to me. I can't comment on the completeness.
> There may be other steps involved...you may have to stall all the Simics
> processors, etc.
>
>
Yes, OpenSolaris goes through some steps that include pausing all
processors when disabling one of the processors.
Right now I'm just concerned with creating the side-effects of
disabling a processor, which is the lost context of local memory such as
L1 cache.
Regarding local memories in GEMS, I guess that flushing the cache will
take care of the directory indirectly, but what about the remaining
memories such as TBE's and write-buffers, do I need to "invalidate" them
as well?
I had an idea that the membar instruction would force flushing of the
write-buffer but I don't know whether GEMS will act this way.
Ruby doesn't model a write buffer.
If there are no outstanding requests, then you need not worry about the TBEs
--Mike
![[Computer Systems Lab]](http://www.cs.wisc.edu/pics/csl_logo.gif){kind=logo}