> On June 13, 2013, 7:53 a.m., Steve Reinhardt wrote:
> > Are we relying on pkt->bus{First,Last}WordDelay to be a multiple of the
> > cache clock period? Of course that would normally be the case if the cache
> > and bus are on the same clock, and I'm sure there are many things we get
> > wrong if they're not, but it seems to be it would be safer to have those
> > terms inside the call to clockEdge() as well. Unless there's a good reason
> > that they're not... if so, please enlighten me.
Very valid concerns indeed. The main reason they are not included is that they
are expressed in Ticks rather than cycles. We could definitely turn them into
cycles and them align them, but for now I'd rather keep it as is.
- Andreas
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://reviews.gem5.org/r/1895/#review4426
-----------------------------------------------------------
On June 4, 2013, 10:47 a.m., Andreas Hansson wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://reviews.gem5.org/r/1895/
> -----------------------------------------------------------
>
> (Updated June 4, 2013, 10:47 a.m.)
>
>
> Review request for Default.
>
>
> Description
> -------
>
> Changeset 9745:58e6081c3e49
> ---------------------------
> mem: Align cache timing to clock edges
>
> This patch changes the cache timing calculations such that the results
> are aligned to clock edges.
>
> Plenty stats change as a results of this patch.
>
>
> Diffs
> -----
>
> src/mem/cache/cache_impl.hh ea26ba576891
>
> Diff: http://reviews.gem5.org/r/1895/diff/
>
>
> Testing
> -------
>
> All regressions pass after stats updates
>
>
> Thanks,
>
> Andreas Hansson
>
>
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev
