[Bioclusters] question: what is your rate limiting computational step?

[Joe Landman]

Hi folks:

(purely non-commercial purpose question)

We are looking to get feedback from informatics users and researchers
where your computational bottlenecks are. Basically, we are trying to
understand which programs or algorithms are the most time/resource
consuming. The groups we speak to frequently often have their own
particular needs, so I would like to make sure we are addressing the
larger community of informatics users and researchers.

The goal of this effort is to explore whether or not acceleration of
some of these codes is possible or beneficial on GPU based systems.

Part of my reason for asking is that I see a number of dynamic
programming algorithms have been accelerated on FPGAs and other
platforms, but I am not seeing large usage of these accelerators. This
suggests that the barriers to obtaining/using these accelerators may be
larger than the benefit of using them ... either because the costs are
too high, or the wrong thing is being accelerated, and bottlenecks are
elsewhere. The GPU-HMMer work recently announced does seem to have good
uptake, and we want to see what else we should be looking at.

We would like to know where the computational rate limiting steps are
today, and what you need to run faster, and what the benefit of
additional speed would be (as well as how much additional speed is
needed to positively impact your research).

Please feel free to email me offline, if there is interest, we will
summarize. Again, strictly non-commercial in nature.

Regards,

Joe
--
Joseph Landman, Ph.D
Founder and CEO
Scalable Informatics LLC,
email: landman at scalableinformatics.com
web : http://www.scalableinformatics.com
http://jackrabbit.scalableinformatics.com
phone: [phone] x121
fax : [phone]
cell : [phone]