Suppose you had an opportunity to get some applications (open source and otherwise) onto CCS. Which apps would you like to see? I have my own list, but I would like to hear yours. I ask as I know most apps run great on the current mainstream clusters. You know, those myriad of Linux units.

What are your major concerns in getting apps on there? Right now there is a defined lack of apps on CCS. This is not for competitive G2, or for bashing purposes. This is to help us figure out if there really is an interest, or is it just marketing hype.
We have been asked to get Scalable HMMer and MPI-HMMer onto OSX. We tried getting it onto Solaris 10, but the build there was 20% slower than under Linux on the same hardware, with optimal build config/options on Solaris 10, and non-optimal under Linux. x64 hardware, and yes, I took great pains to build an x64 binary under Solaris (not easy/default by the way, despite the performance advantages of doing so).
Unfortunately, we don’t see OSX or Solaris 10 as long term growth markets, so unless someone wants to fund this work, we are not likely to continue it.
CCS could grow, if there is interest (buying installations is not a long term viable solution). So my question is, what apps would generate interest. And more importantly, is there enough interest from you to consider CCS with apps ported from Linux. Not just app ports I presume, but integration.
Put another way, how do you want to work on your clusters? Ignore the marketing aspects and what people are telling you in terms of how you want to work. How do you really want to work?
We have been doing the web based interface for a while, and it works well. Is this the approach you want?
Part of this is desktop direction. Are you going to stick with windows? Are you moving to Linux to reduce costs, and increase security? Solaris? OSX? How do you want to interact with your computing systems?
We ask our customers this all the time, but as it seems we have .. a few … readers, it might be nice to see some feedback on this.

3 thoughts on “Inversion”

  1. A lot of computational chemists are used to Windows. I have a feeling that they might appreciate running apps on CCS (docking, some QM programs, search algos, clustering algos etc). Engineers also might be interested. I still feel that for biology-oriented apps Windows will never have enough of a market.

  2. There is some interest in the HPC-biology side from users. Not enough IMO. Lots are interested in acceleration, of their own pet applications.
    I want to see whether or not there is real interest in CCS vs Linux, enough for someone to say “yeah our next cluster will be a CCS rather than a Linux one”. So if we have the chance to do a port of a few apps (CCS appears, for all its mainstream claims, to be missing the vast majority of the mainstream apps which happily run today, out of the box, on linux clusters), my question is which are most important. Also, how would end users want to use them.
    Most users don’t care whats on the back-end as long as it works, is fast, is up, and doesn’t cost them more money. I am trying to understand, if we actually try to (artificially) help level the playing field by removing a portion of the application barrier which exists for CCS, is this sufficient to get interest. I have my suspicions, but I want to see what people think.

  3. My guess is that many users won’t care. In a prior life we had algorithms running on a large Linux cluster, but the front end, which is what the scientists used, was web-based. The question then becomes one for algorithm developers and IT managers. No developer I know is going to develop for Windows (YMMV). Not so sure about the IT manager

Comments are closed.