Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Thursday May 04 2017, @10:43PM   Printer-friendly
from the good-fast-cheap dept.

NASA wants scientific computer experts to take a look at one of its oldest software suites in the hope they can speed it up.

The code in question is called "FUN3D" and was first developed in the 1980s. It's still an important part of the agency's computational fluid dynamics (CFD) capability, and had its most recent release in September 2016.

The agency is now sponsoring a competition with the aim of getting it to go at least 10 times faster. If you can crank it up to ten thousand times faster – without any loss of accuracy – all the better.

Michael Hetle, program executive at NASA's Transformative Aeronautics Concepts Program (TACP) explains that "some concepts are just so complex, it's difficult for even the fastest supercomputers to analyse these models in real time. Achieving a speed-up in this software by orders of magnitude hones the edge we need to advance our technology to the next level".

[Update: Original story title was taken directly from the referenced article; updated to remove condescension. --martyb]


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 3, Insightful) by looorg on Friday May 05 2017, @08:58AM (2 children)

    by looorg (578) on Friday May 05 2017, @08:58AM (#504739)

    There is somewhat more relevant information over at HeroX with all the rules and hoops you have to jump thru beyond just being a US citizen.
    https://herox.com/HPFCC [herox.com]

    Anyway ...

    FUN3D design software so it runs ten to 10,000 times faster on the Pleiades supercomputer without any decrease in accuracy.

    That answered my first question, otherwise they could just have built faster hardware and the code would have "run" faster without even looking at it.

    “This is the ultimate ‘geek’ dream assignment,” said Doug Rohn

    Sure cause nothing says fun and geek like writing code in Fortran ...

    Examples of modifications would be simplifying a single subroutine so that it runs a few milliseconds faster. If this subroutine is called millions of times, this one change could dramatically speed up the entire program’s runtime.

    I wonder if this is really going to be code optimization or if it will be a maths optimization problem. One would think by now that the code is fairly optimized as it is, even tho I will never know why my foreign spying eyes.

    If one looks at the rule page over at HeroX it seems more like this is going to be about module and math optimization then finding little coding errors, those should have been few to non-existing already. Which then becomes even harder as the deadline is just a month and a half or so away as I write this. Good luck getting into the code, replace sections of it with something shiny and new and then hope it runs faster then before. All without access to the supercomputer ... cause you are not allowed to use those unless you own one yourself that you have in your basement.

    Starting Score:    1  point
    Moderation   +1  
       Insightful=1, Total=1
    Extra 'Insightful' Modifier   0  
    Karma-Bonus Modifier   +1  

    Total Score:   3  
  • (Score: 1, Interesting) by Anonymous Coward on Friday May 05 2017, @12:08PM

    by Anonymous Coward on Friday May 05 2017, @12:08PM (#504793)

    tfa is another great candidate for the "tickler file" -- this topic should be automatically added to the submissions list around the time that the "contest" ends...with a note to Eds or others that followup could be interesting. Then we can discuss how it worked out for NASA and for the FORTRAN programming elite.

    My guess? Something from HAKMEM will save the day http://www.catb.org/jargon/html/H/HAKMEM.html [catb.org]

  • (Score: 1, Informative) by Anonymous Coward on Friday May 05 2017, @01:44PM

    by Anonymous Coward on Friday May 05 2017, @01:44PM (#504837)

    I am a FORTRAN expert, or was 20 years ago, and I saw this and thought that it would be fun to dig into the code and see if I could identify any low hanging fruit. However, this is Fortran (not FORTRAN), which I have not ever attempted to code in, so maybe not as fun, but I thought I'd take a peek at it anyway. Then, it turns out to be a big PITA to just access to the code, so now I'm not so interested anymore.