IBM Drops 10 PFlop NCSA Supercomputer Project

  • Thread starter Thread starter Guest
  • Start date Start date
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
[citation][nom]JohnnyLucky[/nom]What did the NSCA plan to do with the supercomputer?[/citation]

I would suspect simulations and problem solving algorithms that would take days, if not weeks on ordinary computers. One of my electronic engineering teachers told us he did his thesis would've taken weeks to do in an ordinary computer, and he did it in 2 hours on his university's supercomputer.
 
They should contact NVidia for the computer.

And neverdyne is correct, things like Meteorological models would be calculated on this. A lot of math problems take forever to do out, even with a modern desktop computer, so they build supercomputers to grind the numbers.
 
[citation][nom]JohnnyLucky[/nom]What did the NSCA plan to do with the supercomputer?[/citation]
Find cures for male pattern baldness and erectile dysfunction.
 
[citation][nom]pasoleatis[/nom]Make your life easier through research.[/citation]
You know, checking emails, updating their Facebook status, messing up wikipedia entries.
 
[citation][nom]neverdyne[/nom]I would suspect simulations and problem solving algorithms that would take days, if not weeks on ordinary computers. One of my electronic engineering teachers told us he did his thesis would've taken weeks to do in an ordinary computer, and he did it in 2 hours on his university's supercomputer.[/citation]
Maybe simulations that would take years on normal computers. I myself I use more than 100 000 hours (about 10 years) of cpu time per year for a rather small project.
 
Bad blow for IBM. Whoever ends up finishing the project will have a lot of bragging rights, having done what IBM chickened out of.
 
[citation][nom]ronch79[/nom]Bad blow for IBM. Whoever ends up finishing the project will have a lot of bragging rights, having done what IBM chickened out of.[/citation]

I doubt it was chickening out. Sounds more like the NSCA didn't want to pay what it was worth, and IBM told them to ah heck off.
 
IBM backed out when they discovered that U of I intended to build the world's largest porn server. Big Blue didn't want its storied and respected brand sullied by such base schemes masquerading as noble scientific inquiry.
 
They realised ....Blue Waters is really the predecessor to Skynet. So they are TERMINATING it in advance....but rest assure, blue water WILL BE BACK again in THE RISE OF THE MACHINES....
 
[citation][nom]JohnnyLucky[/nom]What did the NSCA plan to do with the supercomputer?[/citation]

Same thing we do every night Pinky, "Try to take over the world!". SFX=Evil laugh.
 
[citation][nom]JasonAkkerman[/nom]I doubt it was chickening out. Sounds more like the NSCA didn't want to pay what it was worth, and IBM told them to ah heck off.[/citation]

If NCSA wants to pay less and ends up doing so to another supercomputer builder, it shows IBM couldn't deliver the technology at the same competitive cost. Either way, technological- or business-wise, IBM couldn't or didn't want to pull it off.
 
Blue Waters was originally planned to become NCSA's new flagship supercomputer with a peak performance of 10 PFlops that should have been delivered by at least 300,000 IBM Power7 cores. The original core architecture promised a quad-CPU module, which holds four 8-core Power7 processors. Each processor was promised to deliver a peak performance of 256 GFlops and each module about 1 TFlops. Other specs included more than 1 PB of memory, more than 25 PB of storage, 500 PB of archival storage and more than 100 Gbps of bandwidth.
Click to expand...

Impressive...
 
[citation][nom]JohnnyLucky[/nom]What did the NSCA plan to do with the supercomputer?[/citation]

Break AES-256 encryption??

If terrorist are using AES-256 someone needs to break it right?
 
JohnnyLucky wrote:
> What did the NSCA plan to do with the supercomputer?

Typical application areas are as follows (these refs are on SGI's site, but the
target customers apply to all HPC vendors):

http://www.sgi.com/solutions/research/
http://www.sgi.com/solutions/government/
http://www.sgi.com/solutions/energy/

Note that codes which can exploit more than 500 CPUs/cores are hard to write.
A lot of work is going on to extend parallelism in various fields to support
the execution of larger models. Cosmology inparticular is a strong research
area for this.

Ian.

 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom