Slashdot Mirror
White House Warns of Supercomputer Arms Race
dcblogs writes "The White House's science advisors, in a report last week, said a petaflop-by-petaflop race to achieve number one on the Top500 could prove costly and divert money from supercomputing research. 'While it would be imprudent to allow ourselves to fall significantly behind our peers with respect to scientific performance benchmarks that have demonstrable practical significance, a single-minded focus on maintaining clear superiority in terms of flops count is probably not in our national interest,' the report said (PDF). It is urging the supercomputing community to expand its benchmark measures beyond the Top500's Linpack. It says the Graph500, for data-intensive applications involving the rapid execution of graph operations, 'will be more relevant,' but also acknowledges that it will difficult to rely on any one measure."
Supercomputer Race. Unless supercomputers start blowing up or growing arms.
PETA doesn't want anymore arms used against defensless animals, or inevitably more animals will flop on the ground.
If another country starts to outshine you, try changing the rules.
America's strength used lie in an immense manufacturing culture, and that's given way to "intellectual property". Instead of dealing with tangibles, America is content to sit behind a desk and let the Chinese labour.
We cannot allow a Supercomputer gap!
At difference with War Games, playing is the only way that everyone wins.
If you really need to crunch a lot of numbers and are willing to spend a lot to do it, it often makes more sense to develop an ASIC or FPGA type solution. I know the EFF put together a key cracking system for $250,000 that would probably still blow modern supercomputers out of the water for that specific application.
I've had enough abrasive sigs. Kittens are cute and fuzzy.
"Mr. President, we must not allow a mineshaft gap!"
If you honestly think that the US can't cable together thousands of US GPU's in order to set yet another meaningless Linpack milestone, then you are not that bright.
Warning: PDF!
GENERATION 24: The first time you see this, copy it into your sig on any forum and add 1 to the generation. Social exper
while they are sitting there doing nothing waiting for cyber mutual assured destruction
warning pointless sig
When spending money on a supercomputer, wouldn't you do it to do something useful with it? I'm sure if it gets built, it's built and optimized for a certain purpose other than just being in spot 1 of the Top500 list. On the other hand, if someone does spend the money on a supercomputer purely to have it be in spot 1, well, it's their money and their choice...
so that people can sleep like morons, forgetting that the current government is practicing a worse type of censorship and repression on all freedoms in collaboration with private interests - worldwide too. popping up one censorship method after another, pressurizing foreign governments to implement censorship laws, trying to label journalists who go 'out of line' as terrorists ....
just like it was with 'terror' an external threat needs to be invented so that all kinds of practices violating freedoms can be justified.
censorship ? we are doing it for security on internet.
Read radical news here
The "race" is not about the hardware. All modern supercomputers are massively clustered, using various shared memory architectures. The technology is commodity level, and even a small sum like $10 million can buy a SHITLOAD of hardware. The challenge, and the point of competition, is the creation of software technologies and algorithms to effectively make use of clustered hardware. It's a question of who has the best minds working on the software. The hardware is a given. People have constructed impressive massively parallel processors using game consoles, after all.
It's the programmers, not the supercomputer makers, who will make the difference in this "race."
A lot of supercomputers are not used to their fullest extent; often, this is because scientists either do not know how to program a supercomputer, or do not have enough data, or have computations that are not easily parallelized. Some supercomputer centers have started renting their time to Wall St. firms, because there is not enough demand from scientists or engineers.
Palm trees and 8
'cuz we wouldn't want China to discover the eleventy-billionith prime number before we do.
'While it would be imprudent to allow ourselves to fall significantly behind our peers with respect to scientific performance benchmarks that have demonstrable practical significance, a single-minded focus on maintaining clear superiority in terms of flops count is probably not in our national interest,"
Now try to explain *that* to your TayPartists ...
1. Tianhe-1A (China)
2. Jaguar (USA, ORNL)
3. Nebulae (China)
4. TSUBAME (Japan)
5. Hopper (USA, LBNL)
6. Tera-100 (France)
7. Roadrunner (USA, LANL)
8. Kraken (USA, UT)
9. JUGENE (Germany)
10. Cielo (USA, LANL)
So, let's see -- half of the top ten are in the USA, two in China, two in Europe, and one in Japan. Granted, China is catching up (rapidly), but if you look beyond the #1 spot, the USA still pretty dominates the overall list. Expand this list out beyond the top ten, and SEVEN supercomputers from 11-20 are also in the USA (11-16 & 18), one is in Russia (17), and two in South Korea (19 & 20). So let's not all freak out here about China stealing the #1 and #3 slots on the list -- the USA still has quite a bit more computational resources than the Chinese,. . .
"The Republicans/Democrats/Whigs are dangerously naive about Soviet/Al Qaida/Chinese/Brazilian intentions regarding multicore MIMD instruction code. We must maintain supremacy in cyberspace to protect FREEDOM."
This would be almost as good as a bridge across the Pacific for keeping cold warriors busy.
Physics is like sex: sure, it may give some practical results, but that's not why we do it.
you're always at war -- might as well kill yourself
The advice to the president doesn't change rules for "fastest supercomputer". It tells the president not to be suckered into a supercomputer race measured on only the FLOPS, but rather on more useful performance measures. Because getting sidetracked into less useful metrics to see who's winning the race will waste US resources in winning the race, but not producing the most useful computer. And the US interest is in producing the most useful computer, not in nominally winning the race.
In fact, that report says "let China dominate the Top500, if the US still has the better computers". Which is exactly what I want the US doing, and what I prefer China to be doing rather than leaving the US behind in actual usefulness.
But if you want to get caught up in "the USA is dead" trip that leads into traps that actually would hurt the US if acted on, go ahead. You're not having any effect on the US supercomputer effort.
--
make install -not war
The race to shave yet another hundred of a thousandth of a second when launching your Minesweeper game.
OH NOES
I love the warm, comforting smell of a "national security problem" that can be solved by giving IBM some of my money, rather than one of those genuinely difficult ones like flushing the peasants with small arms out of some sandbox hellhole or doing something that actually improves airline security, rather than harassing pilots who point out problems...
This is laughable, but I can't tell what is funnier:
The fact that the government thinks they know shit about computers,
or the fact that they think they can do it better than any other technologically-advanced country in the world right now.
If the only way you can accept an assertion is by faith, then you are conceding that it can't be taken on its own merits
Craig Mundie Chief Research and Strategy Officer Microsoft Corporation I thing Microsoft is tired of not being in the top500. They will promote new benchmarks instead of the ones that make them look bad. And there are rumours on Windows for ARM on servers. I say Mr. Mundie, I can see what your strategy is all about...
Ok, not entirely worthless. Linear algebra is used in loads and loads HPC workloads, but Linpack as a benchmark is NOT prepresentative of a typical real-world HPC workload. It focuses on peak flops, leaving behind things like inter-node bandwidth and latency, which are crucial for many important, real scientific supercomputing tasks.
Our CSE department chair recently quotes an article he read. To paraphrase, we're heading to the point where computation is going to be basically free, and what costs all the energy will be moving the data around. This is true for several reasons. One is the recent trend towards near-threshold computing. Ultra-low voltage (i.e. 400mV, when 900mV is nominal Vdd) can save 100x on power. It costs us 10x on speed, but now we can pack in 100x as many nodes into the same power and cooling budget, allowing for a 10x increase in aggregate throughput. But this works best for highly parallel and communication-heavy workloads. Fortunately, for many important areas (bombs, climate simulations, astronomy, real-time raytracing), this is the case. And moreover, people are getting better at parallelizing work.
is coming.
A mine shaft gap!
Chas - The one, the only.
THANK GOD!!!
Let's suppose you can perform one AES128 decryption per millisecond. How many cores would you need to brute force a single key within a decade (keeping in mind that brute force is the only publicly known ciphertext-only attack on AES)? Now, how many cores does the world's fastest (known) supercomputer have?
There is a reason that only Dan Brown novels portray supercomputers breaking modern ciphers. It is true that the NSA is believed to have a very powerful supercomputer (or perhaps several) at its disposal, but I doubt that it is used to crack modern ciphers; more likely, they are using it to analyze non-encrypted data from all their signals intelligence work, or perhaps to derive as much information as possible from encrypted transmissions (where the transmission originated, when it was sent, how big it is was, etc.).
Palm trees and 8
China does the manufacturing--but the chips are designed in the US by US companies. Those companies chose to locate their manufacturing in China because labor is so cheap there.
None of this addresses the main point, that Linpack isn't a particularly useful metric.
isn't supercomputers, per se, but AI. The first country that develops scalable human-like AI wins. Period. End of story.
Naturally this isn't even on the *radar* of the leadership of the USA, whose government is dominated by lawyers and financiers, not engineers.
Please do not read this sig. Thank you.
It's about the same idea that some government in NL has to invest in fiber tech. Personally, I don't think that is the way to do it. The idea is to invest in internet content that *possibly* requires high speed internet. The high speed internet will follow. Of course, sometimes you have to push technology a bit as well. But the way to do that is investing in research and small scale try-outs. If the technology succeeds in the try-outs, leave it to the market to fill in demand. I think there are loads upon loads of specialized problems waiting for a supercomputer. Focus on those that need a specialized computer with specialized tech, and make initial investments to succeed in a non-commercial setup. The commercial applications will follow automatically if the tech becomes available.
Personally, I think that creating additional methods to measure speed are absolutely idiotic. They are only interesting to make yet another list of high speed computers. The only way to success is rather simply to solve problems, the computers themselves are only a means. Creating these lists is like comparing dicks, and that's a rather sad thing to do with scientific equipment.
If another country starts to outshine you, try changing the rules.
Is this Rule 2012?
I nominate your comment for an Oscar in the "2010 most insightful Slashdot comments" category.
It's so obvious how to blow the doors off the competition in this race... if everything is all hitting the limits of the Von Neuman wall... go around it with something like a BitGrid, which is a reconfigurable systolic array granular down to the bit level. Using the latest memristors with this idea it should be quite feasible to build Exaflop computers for the desktop.
Just commenting on the summary. Why does it propose to replace linpack with Graphpack exactly ? I understand that linpack catches only a subpart of the application workload patterns, but so does graph500. Both are important, and it is not a matter of changing the benchmark, but to complete the benchmark. Fortunately, the HPC Challenge have been around for quite a time, and does exactly that: measuring the machine in a variety of metric, and giving a summary of the capabilities regarding different relevant aspects.
Now having decided not to use plasma based power systems, Tokyo is again on the verge of another round of massive destruction. I have been studying this, and I am fairly sure that these new supercomputers will emit electromagnetic frequencies that will attract Godzilla, and perhaps other dangerous monsters currently at rest. We would have to ask the twin fairies to be sure, but I think this is a dire threat to Shibuya, and surrounding areas. Having lived there, I would hate to see it destroyed again.
Since when do we get tech advise from the likes of those occupying the White House? Just because you know how to lie and deceive the public does not make you an authority on anything technical.
"Computers are a lot like Air Conditioners" "They both work great until you start opening Windows"
http://it.slashdot.org/comments.pl?sid=1916240&cid=34612834
http://it.slashdot.org/comments.pl?sid=1916240&cid=34647708
http://slashdot.org/comments.pl?sid=1922942&cid=34665368
"3 strikes you're OUT!"
So this is basically Goodharts law applied to supercomputers.
http://en.wikipedia.org/wiki/Goodharts_law
Why don't they just rent the Amazon cloud??