Re: [Gems-users] Stall time


Date: Fri, 04 Apr 2008 16:10:15 -0500
From: Dan Gibson <degibson@xxxxxxxx>
Subject: Re: [Gems-users] Stall time
After a stall is completed -- either counts down to zero or is cancelled via SIMICS_unstall_proc(), Simics always re-issues the request to Ruby. So yes, after each stalled transaction completes, Simics re-issues it to ruby_operate() again.

Regards,
Dan

pathscale@xxxxxxx wrote:
Thanks, Dan
One other thing I'm interested is that
I've observed that when you return a stall time with SIMICS_RETRY_TIME,
the same transactions will come before "ruby_operate" next cyle.
I'm wonderign about that:
if ruby use SIMICS_unstall_proc to unstall the processor, will be last stalled transaction come to "ruby_operate" again.
or it just simply graduate?
Thanks.

在2008-04-05,"Dan Gibson" <degibson@xxxxxxxx> 写道:

That is exactly how ruby implements it. I don't know if its the only way, but it works well.

    Regards,
    Dan

    pathscale@xxxxxxx wrote:
    > Thanks, Dan.
    > But I wonder how ruby implement it.
    > Is it by using API SIM_time_post_cycle in function "runRubyEventQueue" ?
> Is it the only way to wake up ruby every SIMICS_RUBY_MULTIPLIER cycles > and run it indepently?
    > Thanks
    >
    > 在2008-04-05,"Dan Gibson" <degibson@xxxxxxxx> 写道:
    >
> No, but Ruby runs every SIMICS_RUBY_MULTIPLIER cycles independently of > whether any new memory requests are presented to Ruby.
    >
    >     Regards,
    >     Dan
    >
    >     pathscale@xxxxxxx wrote:
    >     > Thanks.
> > Does it mean that even you stall a memory transaction with stall time > > 200000000000
    >     > It will come to timing-model again next cycle, anyway?
    >     >
    >     > 在2008-04-04,"Dan Gibson" <degibson@xxxxxxxx> 写道:
    >     >
    >     >     Often, a memory stall time is not known at request-time. In this
    >     >     case, Ruby stalls Simics for a long long long time, but un-stalls
    >     >     it when Ruby determines its coherence actions have been completed.
    >     >
    >     >     Regards,
    >     >     Dan
    >     >
    >     >     pathscale wrote:
    >     >>     Hi,
    >     >>     I'm studying ruby's core and I'm curious about how a memory
    >     >>     transaction can be returned with stall time like this:
    >     >>     { Miss , SIMICS_STALL_TIME},
    >     >>     which is to say an infinite stall time.
    >     >>     I don't know in what condition this is supposed to happen.
    >     >>     Thanks
    >     >>
    >     >>
    >     >>     ------------------------------------------------------------------------
    >     >>     中国 最 强 网 游 --- 网 易 梦幻 西 游 ,166 万 玩 家 同 时 在 线
    >     >>     <http://event.mail.163.com/chanel/click.htm?from=NO_26&domain=126>
    >     >>     ------------------------------------------------------------------------
    >     >>
    >     >>     _______________________________________________
    >     >>     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! > > > >
    >     >
    >     >
    >     > ------------------------------------------------------------------------
> > 中国 最 强 网 游 --- 网 易 梦 幻 西 游 ,166 万 玩 家 同 时 在 线 > > <http://event.mail.163.com/chanel/click.htm?from=NO_26&domain=126>
    >     > ------------------------------------------------------------------------
    >     >
    >     > _______________________________________________
    >     > 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!
    >
    >     _______________________________________________
    >     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.
    >
> >
    >
    >
    > ------------------------------------------------------------------------
> 中国 最 强 网 游 --- 网 易 梦 幻 西 游 ,166 万 玩 家 同 时 在 线 > <http://event.mail.163.com/chanel/click.htm?from=NO_26&domain=126>
    > ------------------------------------------------------------------------
    >
    > _______________________________________________
    > 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!

    _______________________________________________
    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.



------------------------------------------------------------------------
中国 最 强 网 游 --- 网 易 梦 幻 西 游 ,166 万 玩 家 同 时 在 线 <http://event.mail.163.com/chanel/click.htm?from=NO_26&domain=126>
------------------------------------------------------------------------

_______________________________________________
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→]