I have been reading here about problems with Dual Athlon systems and
Linux. I understand some of these are apparently chipset or DMA
related.
Can anyone point me to more recent information on newer boards and such
where I can find out if this is still a problem now or not?
Trever Adams
> I have been reading here about problems with Dual Athlon systems and
> Linux. I understand some of these are apparently chipset or DMA
> related.
>
> Can anyone point me to more recent information on newer boards and such
> where I can find out if this is still a problem now or not?
Pick up the 760MP errata document from AMD if you are curious. The major
problems in the 760MP about DMA are fixed in the production steppings. From
memory the production ones have a couple of IDE errata (performance and
can't enable prefetching) and an APIC one
Alan
Alan Cox wrote:
>
> memory the production ones have a couple of IDE errata (performance and
> can't enable prefetching) and an APIC one
Is there workaround for these in recent -ac kernels? So is it safe to buy
Tyan Tiger MP for example?
- Jussi Laako
--
PGP key fingerprint: 161D 6FED 6A92 39E2 EB5B 39DD A4DE 63EB C216 1E4B
Available at PGP keyservers
> Alan Cox wrote:
> > memory the production ones have a couple of IDE errata (performance and
> > can't enable prefetching) and an APIC one
>
> Is there workaround for these in recent -ac kernels? So is it safe to buy
> Tyan Tiger MP for example?
Nothing should be needed. If it is then running "noapic" is going to cure
it.
Alan Cox wrote:
>>Alan Cox wrote:
>>
>>>memory the production ones have a couple of IDE errata (performance and
>>>can't enable prefetching) and an APIC one
>>>
>>Is there workaround for these in recent -ac kernels? So is it safe to buy
>>Tyan Tiger MP for example?
>>
>
>Nothing should be needed. If it is then running "noapic" is going to cure
>it.
>-
>
I've got my Tyan Tiger MP up now for about a week, running a 2.4.9-ac18
kernel with
some of the patches from Red Hat's Rawhide kernel rpm applied, and I've
not needed
"noapic" and it's been perfectly stable. I'm not through testing or
optimizing performance
though, but so far so good.
Willem Riede.