[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[ale] I want one!
- Subject: [ale] I want one!
- From: jdp at algoloma.com (JD)
- Date: Fri, 01 Aug 2014 12:49:54 -0400
- In-reply-to: <CAEo=5Pz-mQBhJXSXuKhw-K=MfbwDbfs29-cwynW0qfMfVW_uqw@mail.gmail.com>
- References: <CAEo=5Pz-mQBhJXSXuKhw-K=MfbwDbfs29-cwynW0qfMfVW_uqw@mail.gmail.com>
On 08/01/2014 11:33 AM, Jim Kinney wrote:
> Unboxing a Cray supercomputer.
>
> http://www.itnews.com.au/Gallery/390430,photos-unboxing-the-magnus-supercomputer.aspx/1#pic
>
> OK. I also want the facility and funding to actually have mine running
> crunching numbers for solving world problems, like, how to maximize the
> beer produced from a dwindling barley supply as the changing climate ruins
> the crop. Or how to genetically modify barley to be grown in arid areas for
> beer production. Or how to reduce the water consumed during beer
> production.
>
> Hmm. I see a pattern here.
Back in college, I wrote some really inefficient CFD code (FORTRAN) and blew
through the classroom account funny-money on the Cybers over a weekend before a
program was due. The prof saw that I'd done this and got me a login on the Cray
Y-MP ... which solved the exact same problem in about 20 minutes. As an
engineer, unfamilar with other computers, I preferred the dual cybers - the MNF
compiler wasn't as picky, so I never used the cray again.
BTW, I attempted to run the same program on MS-Dos - it compiled fine, but kept
running into out-of-memory issues. Who knew that a compute space of
100Kx100Kx100K with 8 values for every point would tax any computer? ;)
The following week, we were taught to use non-Cartesian space.