Here’s a collection of highlights, selected totally subjectively, from this week’s HPC news stream as reported at insideHPC.com and HPCwire.
>>10 words and a link
Gates outlines Microsoft cloud with “many millions” of servers;
http://insidehpc.com/2008/06/04/gates-and-the-microsoft-cloud-many-millions-of-servers/
AMD gears up to make Budapest generally available;
http://insidehpc.com/2008/06/04/budapest-out-the-door-at-amd-finally/
Why the utility computing business makes sense for Amazon;
http://insidehpc.com/2008/06/02/why-utility-compute-works-for-amazon-as-a-business/
Google Fellow sheds some light on infrastructure, robustness to failures;
http://insidehpc.com/2008/06/02/google-fellow-sheds-some-light-on-infrastructure-robustness-in-face-of-failure/
Clemson buys new 31 TFOPS Dell;
http://insidehpc.com/2008/06/04/clemson-selects-dell-for-next-machine/
Utility computing prices compared;
http://insidehpc.com/2008/06/04/utility-computing-prices-compared/
T2K “open source” supers enter production in Japan;
http://insidehpc.com/2008/06/03/t2k-supers-enter-production/
Blender Foundation’s new animated short rendered on Sun’s Network.com;
http://insidehpc.com/2008/06/03/big-buck-bunny-an-open-source-success-story/
The National Biomedical Computation Resource plans 3rd Summer Institute;
http://insidehpc.com/2008/06/01/nbcr-summer-institute-2008/
Appro and XtremeData demo in-socket FPGA accelerators on Intel platform;
http://insidehpc.com/2008/06/02/appro-set-to-demonstrate-xtremedata-integrated-fpgas/
>>Where’s My Supercomputer!?
That’s what the folks of Cheyenne, Wyoming, have been asking the National Center for Atmospheric Research (NCAR) lately. NCAR has long been planning and promising to build a new supercomputing facility in Cheyenne. So what’s the holdup!?
“NCAR is coming,” Business Council CEO Bob Jensen told Cheyenne business leaders at the Greater Cheyenne Chamber of Commerce’s annual Economic Reports Lunch. “It’s been caught up in a bureaucratic morass at the National Science Foundation but we’ve had continuous conversations with both the staff at NCAR and the staff at the National Science Foundation, and it is coming.”
According the the officials at NCAR, their $530 million project has been caught up in an internal “restructuring” effort within the Office of Cyberinfrastructure at the National Science Foundation. This is shorthand for “Fun with Washington Bureaucracy.”
Why are the folks in Wyoming getting their feathers so ruffled? Well, they’ve committed themselves to the tune of $47.4 million or 9 percent of the 20-year project cost of $530 million. This includes a good chunk of land donated west of Cheyenne to construct the facility. Moreover:
UW President Tom Buchanan said NCAR’s choice to partner with UW “will transform the University of Wyoming and we believe the state.”
NCAR is going to do this via a proposed system that is “two to five times larger than any existing computer in the world,” according to NCAR Director Tim Killeen. This equates to a system running at between 978 Tflops and 2.39 Pflops [based on today’s Top500].
For 2.3 Pflops, I would be a bit nervous too…
You can read more about the situation here.
>>SGI and Bull: a good summer spat
When the temperature rises tempers flare, and we’re kicking off the summer season of cat fighting with a spat between Bull and SGI.
This week Bull announced the installation of a 20 TFLOPS super based on Intel quad Xeons, stating:
The computer is one of the first major systems in a UK university to use Intel Xeon Quad-core processors, with four cores to each chip.
Note the phrase “one of the first.” This is not the original press release, which, thanks to the Google cache (linked here), we know says:
The computer is the first major system in a UK university to use Intel Xeon Quad-core processors, with four cores to each chip.
The original release is incorrect (unless you want to go all Clinton on the definintion of “major”). SGI’s system at Exeter, announced in Sept. of last year, was certainly earlier and may have been the first Intel quad-core Xeon in Europe. That system looks to have been a 128 core system, which I think would clock it in around 1.5 TFLOPS.
Lots of news outlets ran with the first version of the press release (c.f. the Google cache of the original story on Cardiff’s own Web site and the current version, which has been fixed), including your very own insideHPC.com, which sourced yet another site on its story.
Of course I found out about this because Emily Gallagher of Portfolio Communications emailed me to let me know (it was a nice note, very professional). Yes, SGI is on Portfolio’s client list. Emily’s note has some added info, which you can read on the full story at insideHPC.
—–
John West is part of the team that summarizes the headlines in HPC news every day at insideHPC.com. You can contact him at [email protected].