[Bioclusters] About the parallel Blast on PVM

Akpodigha Filatei bioclusters@bioinformatics.org
Mon, 19 Aug 2002 19:08:38 -0400


Hi, I am interested in this as well.  I have NCBI blast running twice =
faster on a 2-processor Intel Linux box than 4-processor UltraSparc =
server(SUN E3500, Solaris 8).  It is not using more than 1 processor on =
the solaris server even when we specify 4 or more for the -a parameter, =
as it does on the Linux server.  Is there any known implementation that =
takes advantange of the 64-bit architecture of Sun servers?  Please let =
me know if you have any useful info on this.

Filatei=20


> -----Original Message-----
> From: Jimmy Wu [mailto:jimwu@ms25.url.com.tw]
> Sent: Monday, August 19, 2002 8:24 AM
> To: bioclusters@bioinformatics.org
> Subject: [Bioclusters] About the parallel Blast on PVM
>=20
>=20
> Dear Ognen:
>=20
> I just start to study the parallel Blast and get some=20
> information about this
> topic.  I am interested in the parallel version of Blast on=20
> PVM, and I would
> like to compare the result with what I have in SMP machine.  Could you
> please let me know how to contact with the author or obtain=20
> the code to have
> a test run?  Thank you for the help!
>=20
> With Best Regards,
>=20
> Jim
>=20
> On Fri, 2002-06-07 at 06:14, Chris Dwan (CCGB) wrote:
> >
> > > Is there or is there not a parallel version of blast available
> > > somewhere?
> >
> > (addressing a different part of the question than Chris D.)
> >
> > I'm not aware of an MPI version of BLAST, nor would I use=20
> one if it was
> > available.  My problem is throughput, not response time on=20
> BLAST jobs.  In
> > this situation, anything less than a parallel effficiency of one is
> > wasting resources.
> >
> > NCBI's BLAST has the "-a <NUM_CPUS>" option, which enables=20
> threading.  If
> > your operating system is intelligent about SMP and you have=20
> more than one
> > CPU on the board, you can use it to run in parallel.  I=20
> haven't studied it
> > with any rigor, but thumbnail tests indicate that the parallel
> > efficiency is fairly high for NUM_CPU <=3D 8.
> >
> > This is great for decreasing wait time for web users, but it doesn't
> > address my interests at all.  Not to beat it into the ground, but:
> > It's throughput that we need.  That's why we're all so fond=20
> of queuing
> > systems and processing farms, rather than high performance parallel
> > machines.
>=20
> There is a parallel version of Blast based on PVM written by a former
> colleague of mine. It was written/tested on our 32-node=20
> beowulf cluster.
> Instead of posting his email address online, if interested people can
> email me and I will make sure they get in touch with him for sharing
> experiences / results and possibly obtaining the code (I dont=20
> know what
> licensing agreements he and our former employer have in place).
>=20
> Ognen
>=20
>=20
>=20
> _______________________________________________
> Bioclusters maillist  -  Bioclusters@bioinformatics.org
> https://bioinformatics.org/mailman/listinfo/bioclusters
>=20