Re: Debian Beowulf (and intent to package xmpi and scalapack)
On 15 Dec, Camm Maguire wrote:
>
> Greetings! We have our 16 node Debian Beowulf system up and running.
> Nodes: PII 350, 128M RAM, 3G IDE disk, Intel EtherExpressPro NIC.
> Server: Dual PII 350, 128M RAM, 2 9G SCSI Seagate Barracuda, RAID1, 2
> NICS (1 eepro). Baystack 350-T Switch. ~$17k.
>
> We've just completed our benchmarks last week. I would like to use
> the work that has been done to help the Debian project in this area.
> The ways that come to mind are:
>
> 1) contributing an xmpi package (working version already installed
> here)
>
> 1.5) contributing a Scalapack package (not done yet)
>
> 2) helping out with the lam package (currently consulting with
> maintainer.) (LAM is much faster than MPICH on our code where latency
> is an issue.)
>
> 3) setting up a web site with benchmarks, comparing with
> our SGI Power Challenge, and discussing our choice of Debian. (2-9
> times faster, more than 10 times cheaper)
>
> 4) Perhaps writing and contributing a Beowulf administration package.
> I was thinking about something like procmeter, running in gtk over the
> whole cluster. Maybe even some transparent way to do package
> management over the cluster.
>
> 5) Perhaps contributing a kernel module package which enables
> temperature sensing (lm-sensors).
>
> There is so much which we need to do anyway, and which can be usefully
> IMHO shared with the Debian Project, that its hard to know where to
> begin. I would greatly appreciate ideas about what others would like
> to see in Debian for Beowulves; it seems that this is an area in which
> the distribution could excel.
>
> My problem: I've been a developer of a single package for some time,
> but can no longer subscribe to the mailing lists, because the volume
> overwhelms my ISP's fragile mail system, and/or I get periodically
> unsubscribed when my ISP temporarily refuses to accept mail. Please
> send any replies to camm@enhanced.com.
>
Put me on your list too, I am also interested in this subject.
Xavier Calbet
Reply to: