Print 32 comment(s) - last by Sunner.. on Oct 5 at 4:08 AM

Fermi supercomptuer will be ten times more powerful than today's fastest supercomptuer

NVIDIA was at the forefront of the push to move high-performance computing from CPUs to GPUs in scientific and other areas of research. As it turns out, the GPU is a very effective tool for running calculations historically run on the CPU.

NVIDIA announced its new Fermi architecture at its GPU Technology Conference recently. The new architecture was designed from the ground up to enable a new level of supercomputing using GPUs rather than CPUs. At the conference, Oak Ridge National Laboratory (ORNL) associate lab director for Computing and Computational Sciences, Jeff Nichols, announced that ORNL would be building a next generation supercomputer using the Fermi architecture.

The new supercomputer is expected to be ten times faster than today's fastest supercomputer. Nichols said that Fermi would enable substantial scientific breakthroughs that would have been impossible without the technology.

Nichols said, "This would be the first co-processing architecture that Oak Ridge has deployed for open science, and we are extremely excited about the opportunities it creates to solve huge scientific challenges. With the help of NVIDIA technology, Oak Ridge proposes to create a computing platform that will deliver exascale computing within ten years."

ORNL also announced at the conference that it would create a Hybrid Multicore Consortium with the goal of working with developers of major scientific codes to prepare the applications for the next generation of supercomputers using GPUs.

“The first two generations of the CUDA GPU architecture enabled NVIDIA to make real in-roads into the scientific computing space, delivering dramatic performance increases across a broad spectrum of applications,” said Bill Dally, chief scientist at NVIDIA. “The ‘Fermi’ architecture is a true engine of science and with the support of national research facilities such as ORNL, the possibilities are endless.”

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

It does depend on one factor though...
By Saist on 10/3/2009 1:42:30 AM , Rating: 2
The proposed supercomputer does depend on one important factor... that Fermi actually is real. The big problem Nvidia has at this point... is that it's not. As far as I'm aware, Nvidia wasn't even able to show Fermi silicon working on IKOS boxes... much less actually showing off fabbed versions of even beta silicon. White papers and theories are one thing... real-life performance is another.

Remember the P4 architecture? Northwood and Prescott? Scaling to 8ghz and beyond? Remember Hyperthreading? Remember the Radeon 8500's 3 textures per pass? Just because those technologies worked on paper, didn't mean they actually worked when put in Silicon... and didn't mean that developers actually coded to use the features.

Nvidia's got even larger issues than non-working silicon. Their stock has been taking repetitive beatings each time a vendor issues a recall on Nvidia hardware. It's no secret that Nvidia's got a huge problem with runaway thermal envelopes. The promise to offer a certain amount of computing power in a certain thermal envelope... should send most investors into fits of hysteria.

From an outside viewpoint, having a vendor interested in Fermi this early is no doubt good news for Nvidia and it's stock. However, if Fermi isn't all that Nvidia's hyped it as... it could just be the nail that sends Nvidia into bankruptcy.

By habibo on 10/4/2009 1:46:20 AM , Rating: 2
What silicon do you propose that they were demoing this past week at their GPU Technology Conference? Are you suggesting a "faked moon landing" conspiracy theory?

"Can anyone tell me what MobileMe is supposed to do?... So why the f*** doesn't it do that?" -- Steve Jobs
Related Articles

Most Popular ArticlesFree Windows 10 offer ends July 29th, 2016: 10 Reasons to Upgrade Immediately
July 22, 2016, 9:19 PM
Top 5 Smart Watches
July 21, 2016, 11:48 PM

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki