Two records in one day for SGI supercomputer

MOFFETT FIELD, Calif.--Even as Silicon Graphics Inc. trumpeted on Tuesday a new speed record with the Columbia supercomputer it built for NASA, CNET News.com has learned, it quietly submitted another, faster result: 51.9 trillion calculations per second.

During the unveiling of the Columbia supercomputer, SGI touted a speed of 42.7 trillion calculations per second, or 42.7 teraflops. That handily beat the machine at the top of a list of the world's 500 fastest machines, NEC's Earth Simulator at 35.9 teraflops, as well as a top challenger, IBM's Blue Gene/L at 36.0 teraflops.

The 42.7 teraflops speed used only 16 of Columbia's 20 servers. That means that 2,048 of the 10,240 Itanium processors in the supercomputer weren't being used--and the unused chips are the newest generation of Itaniums, each with 9MB of high-speed cache memory, SGI Chief Executive Bob Bishop said.

SGI clocked the full 20-server system at a sustained speed of 51.9 teraflops, according to a source familiar with the test. On a secondary but still scrutinized measurement, peak speed, Columbia ran at 61.0 teraflops, a smidgen ahead of the 60-teraflop speed Intel President Paul Otellini predicted in September.

Full results in the closely watched competition are released every June and November at supercomputing conferences; the newest Top500 list will be released Nov. 8, organizer Jack Dongarra said. Despite the interest in the list, its organizers and others recognize that the speed test, called Linpack, is a convenient but incomplete performance measurement.

/topic/silicon-graphics-inc.html">Silicon Graphics Inc., teraflop, supercomputer, Intel Itanium, IBM Corp.

Add a Comment (Log in or register) (3 Comments)

• prev
• 1
• next

linpack is not incomplete, just silly

by markhahn October 27, 2004 5:30 AM PDT

linpack is not incomplete, just silly

it's common knowlege that you can calculate your top500 score by multiplying your peak theoretical flops rate by about .65. your peak, of course, is merely a product of the number of cpus and the flops/cycle for that cpu (2 for most, 4 for cpus that have fused-mul-add like it2 and power chips).

in other words, it's not a benchmark of real code, just a slightly obfuscated theoretical rating.

Reply to this comment

Nanoelectronics breakthrough

by October 27, 2004 11:38 AM PDT

Nanoelectronics breakthrough

Vulvox Nano/Biotechnology Corporation has announced a research program leading to development of 60 terabit storage devices and FPGA
type logic devices containing 60 trillion transistors on a one cm square chip. Neil Farbstein the President of Vulvox Nano/Biotechnology Corporation was interviewed by
Arya Kumar of Nano Investor News. You can read the interview by clicking A link on the front page of the Vulvox Inc. website at http://www.vulvoxnanobio.tripod.com
The record breaking Colmbia teraflop computer
might be duplicated on a single molecular electronics chip using DNA transistor and nanowire components. Vulvox is also developing molecular size Q-bit quantum processor elements that have been theorized to arrive at answers a million times faster than current computers.
Corporations or other research laboratories are encouraged to contact Neil Farbstein, President of Vulvox Nano/Biotechnology at [e-mail]

Reply to this comment

How much did it cost SGI vs BigMac or Colossus?

by Jill_Gates October 31, 2004 12:44 AM PDT

How much did it cost SGI vs BigMac or Colossus?

How much did it cost SGI to build this beastie? How about to
cool / power and run it?

How much did it cost Virginia Tech for Big Mac?

Reply to this comment

(3 Comments)

• prev
• 1
• next

Add a comment

Comment SUBMIT

Click here to add another comment.

Popular discussions on CNET:

1. Crave giveaway of the week: HP MediaSmart Server LX195

   July 10, 2009 5:00 AM PDT

   (1947 recent comments)

2. What Chrome OS has on Windows that Linux doesn't

   July 9, 2009 4:00 AM PDT

   (180 recent comments)

3. Google plans Chrome-based Web operating system

   July 7, 2009 10:16 PM PDT

   (135 recent comments)

4. Chasing the Toyota Prius' 50-mpg nirvana

   July 10, 2009 9:19 AM PDT

   (122 recent comments)

5. Is Sony on the right track with its PS3 strategy?

   July 9, 2009 6:00 AM PDT

   (112 recent comments)

The posting of advertisements, profanity, or personal attacks is prohibited. Click here to review our Terms of Use.

Comment reply

Submit Cancel

The posting of advertisements, profanity, or personal attacks is prohibited. Click here to review our Terms of Use.

Report offensive content:

If you believe this comment is offensive or violates the CNET's Site Terms of Use, you can report it below (this will not automatically remove the comment). Once reported, our staff will be notified and the comment will be reviewed.

Select type of offense:

Offensive: Sexually explicit or offensive language

Spam: Advertisements or commercial links

Disruptive posting: Flaming or offending other users

Illegal activities: Promote cracked software, or other illegal content

Comments (optional):

Report Cancel

E-mail this comment to a friend.

E-mail this to: (Separate multiple e-mail addresses with commas. Limited to 10 addresses.)

Your e-mail address:

Send me a copy of this message

Note: Your e-mail address is used only to let the recipient know who sent the e-mail and in case of transmission error. Neither your address nor the recipients's address will be used for any other purpose.

Add your own personal message: (Optional)

Hi, I found this user's comment on CNET and thought you might be interested in reading it. Send e-mail Cancel

Warning! You will be deleting this comment and all its replies (if applicable).

Click to delete FOREVER Cancel

[markhahn]

linpack is not incomplete, just silly

it's common knowlege that you can calculate your top500 score by multiplying your peak theoretical flops rate by about .65. your peak, of course, is merely a product of the number of cpus and the flops/cycle for that cpu (2 for most, 4 for cpus that have fused-mul-add like it2 and power chips).

in other words, it's not a benchmark of real code, just a slightly obfuscated theoretical rating.

Nanoelectronics breakthrough

Vulvox Nano/Biotechnology Corporation has announced a research program leading to development of 60 terabit storage devices and FPGA
type logic devices containing 60 trillion transistors on a one cm square chip. Neil Farbstein the President of Vulvox Nano/Biotechnology Corporation was interviewed by
Arya Kumar of Nano Investor News. You can read the interview by clicking A link on the front page of the Vulvox Inc. website at http://www.vulvoxnanobio.tripod.com
The record breaking Colmbia teraflop computer
might be duplicated on a single molecular electronics chip using DNA transistor and nanowire components. Vulvox is also developing molecular size Q-bit quantum processor elements that have been theorized to arrive at answers a million times faster than current computers.
Corporations or other research laboratories are encouraged to contact Neil Farbstein, President of Vulvox Nano/Biotechnology at [e-mail]

[Jill_Gates]

How much did it cost SGI vs BigMac or Colossus?

How much did it cost SGI to build this beastie? How about to
cool / power and run it?

How much did it cost Virginia Tech for Big Mac?