On Mon, 2005-01-17 at 09:14 -0600, Chris Friesen wrote:
> Benjamin Herrenschmidt wrote:
>
> > Well.. the cache flush part requires some not-really-documentd stuff on
> > the 970, but I'll try to come up with something.
>
> Details? We've got a cache-flush routine put together based on the
> doc
Benjamin Herrenschmidt wrote:
Well.. the cache flush part requires some not-really-documentd stuff on
the 970, but I'll try to come up with something.
Details? We've got a cache-flush routine put together based on the
documentation that seems to be working, but if there's something else
that has
On Mon, 17 Jan 2005, Benjamin Herrenschmidt wrote:
> On Sun, 2005-01-16 at 21:37 -0700, Zwane Mwaikambo wrote:
> > Hello Ben,
> >
> > On Sun, 16 Jan 2005, Benjamin Herrenschmidt wrote:
> >
> > > Looks good, but you could do even better :) I still want to look at the
> > > proper mecanism to flus
On Sun, 2005-01-16 at 21:37 -0700, Zwane Mwaikambo wrote:
> Hello Ben,
>
> On Sun, 16 Jan 2005, Benjamin Herrenschmidt wrote:
>
> > Looks good, but you could do even better :) I still want to look at the
> > proper mecanism to flush the CPU cache on 970, but the idea here is to
> > flush it, and
Hello Ben,
On Sun, 16 Jan 2005, Benjamin Herrenschmidt wrote:
> Looks good, but you could do even better :) I still want to look at the
> proper mecanism to flush the CPU cache on 970, but the idea here is to
> flush it, and put the CPU into a NAP loop (the 970 has no SLEEP mode)
> with the cache
On Thu, 2005-01-13 at 17:43 -0700, Zwane Mwaikambo wrote:
> I found the following very handy for use as a reference platform when
> working on i386 hotplug cpu recently.
>
> It's been tested on a G5 system with a cpu going on/offline every second
> and make -j. I've also tried a number of config
6 matches
Mail list logo