Hi all,

Just wondering if there has been any developments regarding Phi cards?

We have just installed a small 10-node cluster with two MICs per node, and are
wondering how best to use the cards.

I intend to try the cross-compilation to install slurm on the cards, and then
have a separate queue, similar to that described in the discussion on linkedin
below.

But I think the users would also like to be able to run offload as well, so
that's a bit of an issue.

I have an additional question: the users have asked if a portion of the host
memory can be ``reserved'' for comms with the MICs, with the rest available for
whoever runs on the host. Is that possible??

Thanks,
Paddy


On Tue, Feb 18, 2014 at 01:39:44AM -0800, Olli-Pekka Lehto wrote:

> 
> Getting the daemon running on the Phi is certainly possible and we tried this 
> a year or
> so ago. The real challenge lies in being able to run offload, host-native, 
> symmetric and 
> Phi-native mode programs nicely on the same set of nodes. It is something 
> that would 
> really be needed in order to maximize utilization of a Phi-cluster. 
> 
> Furthermore, having a simple way to maintain affinity of Phi cards with their 
> associated 
> hosts when doing symmetric runs would be very useful. It's sort of possible 
> with the topology 
> plugin but a bit clunky. 
> 
> Also it would be nice to have a more lightweight daemon in order to conserve 
> the precious
> resources of the Phi as was presented in the Slurm User Group presentation. I 
> expect that
> this would be a more significant undertaking, however(?)
> 
> I'm wondering if people have been working on these kind of things?
> 
> Olli-Pekka
> 
> On Feb 18, 2014, at 1:05 PM, Ralph Castain <[email protected]> wrote:
> 
> > 
> > I know others have direct-launched processes onto the Phi before, both with 
> > Slurm and just using rsh/ssh. The OpenMPI user mailing list archive talks 
> > about the ssh method (search for "phi" and you'll see the chatter)
> > 
> > http://www.open-mpi.org/community/lists/users/
> > 
> > and the folks at Bright talk about how they did it with Slurm here:
> > 
> > https://www.linkedin.com/groups/Yes-we-run-SLURM-inside-4501392.S.5792769036550955008
> > 
> > Ralph
> > 
> > On Feb 17, 2014, at 5:46 PM, Christopher Samuel <[email protected]> 
> > wrote:
> > 
> >> 
> >> -----BEGIN PGP SIGNED MESSAGE-----
> >> Hash: SHA1
> >> 
> >> Hi all,
> >> 
> >> At the Slurm User Group in Oakland last year it was mentioned that
> >> there was intended to be support for a lightweight Slurm daemon on
> >> Xeon Phi (MIC) cards.
> >> 
> >> I had a quick look in the git master last night but couldn't spot
> >> anything related, is this still the intention?
> >> 
> >> Olli-Pekka Lehto from CSC is running a Xeon Phi workshop at VLSCI at
> >> the moment and it's of interest to a number of us.
> >> 
> >> We're going to run a hack day on Wednesday and we'll see if we can
> >> build an LDAP enabled Xeon Phi stack, if we can then we we'll see if
> >> we can get standard Slurm going too. Nothing like having lofty goals!
> >> 
> >> All the best!
> >> Chris
> >> - -- 
> >> Christopher Samuel        Senior Systems Administrator
> >> VLSCI - Victorian Life Sciences Computation Initiative
> >> Email: [email protected] Phone: +61 (0)3 903 55545
> >> http://www.vlsci.org.au/      http://twitter.com/vlsci
> >> 
> >> -----BEGIN PGP SIGNATURE-----
> >> Version: GnuPG v1.4.14 (GNU/Linux)
> >> Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
> >> 
> >> iEYEARECAAYFAlMCuvIACgkQO2KABBYQAh+pwgCcCLPvoUJamArfmpxY5igcJm3I
> >> 0p0AnjF51qUgZfoZtIsKTDLCK+pJe+bf
> >> =7HO3
> >> -----END PGP SIGNATURE-----
> 

-- 
Paddy Doyle
Trinity Centre for High Performance Computing,
Lloyd Building, Trinity College Dublin, Dublin 2, Ireland.
Phone: +353-1-896-3725
http://www.tchpc.tcd.ie/

Reply via email to