New I.B.M. Supercomputer to Begin Its Weather Work

The New York Times ^ | June 6, 2003 | JOHN MARKOFF

[Ernest_at_the_Beach]

---

June 6, 2003

New I.B.M. Supercomputer to Begin Its Weather Work

By JOHN MARKOFF

he nation's most powerful supercomputer for weather forecasting is scheduled to go online today, I.B.M. said yesterday, a machine that may eventually rival the Japanese Earth Simulator as the world's fastest supercomputer.

The new computer, with a theoretical peak computing power of 7.3 trillion operations a second, is expected to be enhanced over the next few years, and it may reach speeds up to 100 trillion operations a second by 2009, I.B.M. said.

It ranks third in the United States in speed, behind two Hewlett-Packard machines at Los Alamos National Laboratories in New Mexico.

The federal government will use the new computer to improve the accuracy of weather forecasts, particularly to help predict the path of hurricanes three to five days in advance, providing additional time to prepare for the storms.

The Earth Simulator, financed by the Japanese government, was installed in Yokohama, west of Tokyo, at a cost of $350 million to $400 million. It is capable of 35.8 trillion operations a second.

The National Weather Service plans to spend more than $200 million on the I.B.M. system, which is now composed of 1,408 processors arranged in a cluster of 44 interconnected systems. The system, which is housed in an I.B.M. computing center in Gaithersburg, Md., will be constantly expanding.

It will be used to produce weather forecasts for the National Weather Services National Centers for Environmental Prediction. Those forecasts are used as the basis for television and newspaper forecasts.

The machine will also perform a smaller amount of climate research. It is capable of roughly the same performance as a machine I.B.M. built for European weather forecasters.

Supercomputers have become increasingly crucial for science, technology and military development. Since the completion of the Japanese Earth Simulator, some policy makers have expressed renewed concern about United States technology leadership. The debate is centered on how much money the government should invest in supercomputing, as well as what the design and focus of the new machines should be.

[Ernest_at_the_Beach]

OFFICIAL BUMP(TOPIC)LIST

[CPOSharky]

...operations a second...

Yo, Markoff, tell the person you cribbed this report from that it is "operations PER second."

[kellynla]

$200M? Want to know what the weather is? I wonder if anyone ever tried sticking their head out the window? "Yep, sun's out and it looks like rain?" Can I have my $200M back? These morons in DC make a shipload of drunken sailors on a weekend pass look like choirboys!

[q_an_a]

The national weather service provides at ____ cost to radio and TV, the News which the TV people who use to provide on the peoples airwaves, at little benefit to a civil society, weather data that is so useless and so often wrong as to be a joke. This is another example of the pooer use of government spending.

Would 200 million dollars help them know more about Andrews path to Miami or that at the last minute it would gain strength? People knew it was coming and they knew it was strong, they could not prepare against the force. So how does this improve life? How do the weather people at the local NBC or ABC TV station give money to the government for this investment?

What this will mean is that the weather people will have more reasons to interrupt everything to give advice, cut to a commerical and return to normal programming. /rant off

[Ernest_at_the_Beach]

I think the corrections department in california is wasting more than 200 million every year!

[OrioleFan]

And, if Control Data Corporation was still around, this would still be second best.

[oldcomputerguy]

To the people who can't see the enormous economic requirements for accurate and extended weather forecasting, talking to them will do no good as they are so far out into space as to be incapable of understanding.

With IBM's switching sooooooooooo slow in their clustered systems, I just hope we get the forecast before it is out of date. Their processor speeds sound fast but they can't come close to matching it in real production.

[Uncle George]

The movement of a butterflys wings in Brazil can effect the weather in Omaha next week. Can the supercomputer count the butterfly wing beats in Brazil???

[konaice]

"$200M? Want to know what the weather is? I wonder if anyone ever tried sticking their head out the window? "Yep, sun's out and it looks like rain?" Can I have my $200M back? These morons in DC make a shipload of drunken sailors on a weekend pass look like choirboys!"

---
Ask Tommy Franks what it would have been worth to know the
precise arrival time of those sand storms in Iraq.

Ask any huricane/tornado victom if 5 or 10 hours advanced ACCURATE notice would have helped.

Then crawl back into your cave, roll the stone in front of the entrance, and keep your luddite opinions to yourself where the reveal your ignorance to no one.

[Servant of the Nine]

A Complete Fraud and Waste of Taxpayers Money.

For 40 years computational weather wizards have been promising results if they just get a bigger computer.
We have wasted billions suplying them and yet prediction is no better today than it was when they started.

Since the development of Chaos Theory it has been obvious that no amount of computational power can predict the weather.

Supercomputer manufacturers and Computational Meteorologists have been conspiring to keep this fraud going for a good 20 years after it was proven worthless, because their jobs depend on it.

So9

[freedumb2003]

>>Can the supercomputer count the butterfly wing beats in Brazil??? <<

Sure:

"um dois três estada de quatro... esperas! ainda!! ... um dois três quatro..."

[konaice]

"And, if Control Data Corporation was still around, this would still be second best."

Third. Cray beat CDC every time.

But the massive arrays of cheap Intel chips still give the big iron a run for the money every time. (Running Linux of course).

[OrioleFan]

Seymour Cray came from Control Data where he designed the 6000 and 7000 series mainframes and Control Data bankrolled his startup in Chippewa Falls.

[konaice]

"We have wasted billions suplying them and yet prediction is no better today than it was when they started."

Prediction is VASTLY better today than when they "started".

(as if you or I are in a position to determine when they started).

Define "Wasted".

This sounds like the old anti-Space argument against "Wasteing billions in space" which suggests that $100 doller bills are shoveled aboard a rocket and put in orbit. The money is spent here on earth, in the
US, putting folks to work and paying for groceries and rent and cars.

And in the process weather prediction gets marginally better over time, crops can be better planned, disasters better predicted, etc.

You would rather spend the money on wellfare where the first derivitive is lost?

There are some things, like research and defense that are just better if done and OWNED collectivly than not done at all or owned by another Bill Gates.

[konaice]

"Seymour Cray came from Control Data where he designed the 6000 and 7000 series mainframes and Control Data bankrolled his startup in Chippewa Falls."

Yup.
Seymour also once said "Pairity checking is for farmers".

But he eventually had to put that back into the design.

[oldcomputerguy]

But the massive arrays of cheap Intel chips still give the big iron a run for the money every time

Not really except in some specialized applications.

Cray was lost for some years while SGI was trying to bury it and missed a complete product cycle. That is now past and nothing will keep Cray off the top list anymore. The X1 is out now and Cray is back on rhythm for new equipment and new generations. IBM will recede back into the back waters of supercomputing again. They really do not have any serious committment to it anyway.

[Servant of the Nine]

Prediction is VASTLY better today than when they "started".

Do to satelite observation, not computation.

This sounds like the old anti-Space argument against "Wasteing billions in space" which suggests that $100 doller bills are shoveled aboard a rocket and put in orbit. The money is spent here on earth, in the US, putting folks to work and paying for groceries and rent and cars.

I am all for Government funding cutting edge research. Lets buy all the supercomputers we can for Physics, genetics, the CIA and NSA, but buying one for the weather whackos is no better than giving one to a special design team for dowsing rods.

I don't object to the money, I object to its waste, and I don't care if it is spent here, it is spent out of my pocket.

So9

[NormsRevenge]

I think the corrections department in california is wasting more than 200 million every year!

That's a conservative estimate if I ever heard one! ;-)

[OrioleFan]

"Contrasting this modest effort [of Seymour Cray in his laboratory to build the CDC 6600] with 34 people including the janitor with our vast development activities, I fail to understand why we have lost our industry leadership position by letting someone else offer the world's most powerful computer."
Thomas J. Watson, IBM President, 1965

"It seems Mr. Watson has answered his own question."
Seymour Cray