[Beowulf] Acceptable rad limits for cluster rooms?

Greg Lindahl greg.lindahl at qlogic.com
Mon Jun 19 14:52:59 PDT 2006


On Mon, Jun 19, 2006 at 02:45:40PM -0700, Jim Lux wrote:

> >Processor *caches* do have ECC, and since they're more susceptible
> >than the logic, this gets most of the benefit.
> 
> Some caches do, some don't.

You will be hard-pressed to find one you'd use in a Linux cluster that
doesn't have ECC, other than L1 icache which tends to just have parity
and invalidation if there's a single-bit error.

> And it's tough to find out exactly what they mean by ECC.

That's very true, that's usually a secret right up there with the
cache replacement algorithm.

-- greg




More information about the Beowulf mailing list