cray spends 4-5 years of time to build a machine, just to sell a very small few of them, throw almost all of the technology away and start over again.
That's exactly the problem Cray has had competing with microprocessors. It's not that microprocessors are technically superior to custom vectors for that market. They're not! It's that Cray's market segment is too small to support the fast development cycles and huge research budgets of chip companies like AMD and Intel. By latching onto AMD's products for some of the work, Cray can avoid spending so much on a development cycle. I'm not trying to claim that the marketing has been perfect, but that they're at least considering the problem unlike kneejerk clusterheads. For the size of some Crays sold, the MTTI has been quite respectable.
The parent poster posts well for one ignorant of the simplest precepts of marketing. The first things a marketer learns is he must segment a market and only compete in the segments or niches in which competition is profitable. Cray isn't competing directly against clusters because clusters don't have the bandwidth necessary for the sorts of problems Crays are aimed at and Crays tend to be overkill for the problems clusters are aimed at. Cray doesn't seek out customers $.5M for that reason. Anyone who actually uses the supercomputers to solve problems knows that a 50% difference in interconnect speed per single link could mean a 90+% slowdown on a large system using a large program with high overhead.
Plain old clusters aren't targetted against Crays, except by some communities that don't buy supercomputers for supercomputer problems anyway, like most Slashdot users. In the supercomputer world, MTTI is everything! That means mean time to interrupt. A bad memory module or a CPU fan blowing out on your single CPU might happen every 3 years on average, but multiply these sorts of problems by 10,000 CPUs on a supercomputer and your cluster will never get any useful work done before something goes out and it crashes.
Disclaimer: I worked on the X1/X1e, which is still faster than any other chip on select problems which vectorize well. I agree that the AMD partnership was and continues to be an excellent decision, but it only says that AMD does SCALAR performance better, not everything!
The parent poster needs to be reminded that a large chip manufacturer like Intel, IBM, and AMD makes much more than CPUs! They play a fundamental role in the design and system architecture of the machines built out of their chips. Interfaces like Hypertransport, PCI Express, and DDR are the work of these chip giants. To claim that changing the fundamental design of the CPUs has anything to do with the interaction of a supercomputer company and AMD is naive. Far more likely are changes in Hypertransport, interfaces to memory, or other bus-level projects that are more useful to a supercomputer vendor looking for the best possible overall system bandwidth anyway.
There is no solid definition of what differentiates a cluster, but it's a bit misleading and confusing to refer to an architecture that doesn't have any of the characteristics of a traditional cluster as a cluster. Most clusters have a star or bus network topology of Infiniband, Myrinet, or Ethernet. An X1 uses a 3D torus of redundant, high-bandwidth links. Each node is a board in a backplane, not just a rackmount system. While defining a cluster as a machine where each node has its own OS is a convienient definition, that also includes things which clearly aren't true clusters like networks of workstations and many MPPs.
It's very nice to see someone who gets it. One of the more interesting links off the article is Sandia's reevaluation of Amdahl's Law. If you read Sandia's other research involving the Red Storm project, they point out that Amdahl's Law doesn't have to be as big a killer as previously thought on truly massive systems when paired up with a system that has good balance in memory, network, and processor bandwidth. Most clusters have drastically worse balance than custom MPPs and it causes many algorithms to suffer.
The problem has to match the computer or your cluster will never solve it. For example, a problem that requires excessive message passing won't run well on Blue Gene, which has 1/22nd the network bandwidth of Red Storm. The real advantage of Blue Gene is in low power and tight packaging. There's a common misunderstanding that Linpack numbers = real performance. Unless your company gets paid everytime Linpack ends, you can't judge the value of an HPC system by the top 500. For many problems, the Earth Simulator is still winning.
if only it had compatibility for the/pizza command from Everquest 2. Until they add the ability to bring up a pizza ordering screen with only 7 keystrokes, I'll stick with browsing the internet using my copy of Everquest 2, which is f'ing cool according to other slashdot users. Why would anyone ever want to use anything other than EQ2? Happy birthday Opera and thanks, but no thanks.
An underling at a used computer store I worked at in college was building a PC for the city fire chief. He chose a macintosh monitor off of the used monitor shelf and attached it to the PC's serial port, which matches the connector and carries a charge. Ironically, this lit the monitor on fire. The fire chief was a good sport about it.
A college roommate was impressed by how easily I built my own computers and got too anxious to install his DDR memory in his new PC when he got home that he didn't wait for an expert to arrive. He managed to jam the memory in backwards against the key as hard as he could until he heard a snap. Both the motherboard and memory started on fire.
Wow. Only on Slashdot could someone quote a post, miss its point entirely, post one's misinterpretation, and score +5 insightful.
Prior art is not affected by first-to-file. You can still invalidate any patent with prior art under a first-to-file system. The bill (at least in the version I read) also allows unrelated third party prior art challenges, which weren't allowed before.
http://democracyrising.us/content/view/196/164/
This is an excellent article and it's the best I've read on the draft rumors. Summary: The rumors of secret draft plans are only rumors, but the military can't keep using mercenaries to account for their poor recruiting numbers. Sooner or later, the merc recruiters are hiring the same people at higher cost who could've served in the military and bodies are needed on the ground in the field. Like with police, you need support people for every person on duty. You can retrain some troops for infantry combat and ship them, but you can't cannibalize all corps and lose the support people to win the war. Nevermind that we went to war without any clear victory conditions set.
I suggest to Lucas that he target the Slashdot audience directly by making Natalie Portman naked on screen. That is if they can still stomach her after she turns evil.
But maybe it had 64 bit applications? Nope, guess again. No 64 bit OpenOffice, no 64 bit Eclipse, not one goddamn app I needed to use was ported yet. Just for a lark I tried emerging Pingus.
Knowing how many working 64-bit Opteron Linux supercomputers have been sold over the last two years for real business purposes, I concur with you that Linux will still not be 64-bit ready(tm) until YOU can play pingus.
Re:it's not just about the user experience
on
Opera 8 Released
·
· Score: 1
You're implying that if it doesn't take place on top of or underneath a typical hardcore Slashdot user's desktop, it matters. If a tree falls in the forest and it doesn't run Debian...
Re:Here is a MUCH easier way to get Opera for Free
on
Opera 8 Released
·
· Score: 1
Opera 8 is a free upgrade to Opera 7 so the point is moot.
I'm feeding the troll here, but #1 why not? and #2 damn right. Opera does one thing and does it well. Firefox is mostly buzz these days as most of the features aren't user-friendly, swift, or light on memory. So long as IE continues to be ignored or held back, the market is wide open for innovators.
Why does Open Source have to beat competitors who fill a niche and really try to do their jobs well? Do we have to release a faster, lighter, Minimo (Moperilla?) browser simply because commercial competitors are evil or is it because of the spirit of advancing the art? It's fine to admit that the competition is better at something and in fact, it's required in order to improve. Why not admit that the Firefox interface is still pretty ugly?
Who cares if they aren't open source? I just need something fast with real sessions and MDI that doesn't hog my memory or invite spyware in. Web standards only matter when things actually don't work.
Is anyone reading this tripe an engineer who actually cares about best tool for the job anymore? Some people use browsers and operating systems to do paying work. So what's wrong with using Linux and Opera?
If you look at the Nullstone website, they have an excellent list of optimizations to expect from a compiler and a few to hope for. It must be remembered, however, that some of these can produce wrong answers such as illegal hoisting if the compiler isn't tested thoroughly enough. You must remember that you can make code very, very fast if you don't care about correct answers and useful results. Remember your priorities!
Whether you support Badnarak (as so many meta-moderators), Nader, Cobb, or even Kerry, I think we can all agree that the wrong issues are being discussed in the media and that this election is a sham. As Ralph Nader said in Harlem a couple weeks ago, there are thirty times as many serious birdwatchers in America as Congress-watchers. Don't be a fool and let yourself be sucked into minor differences between the two candidates when both support unjust war and the continuing destruction of our democracy. Counterterrorism starts at home.
Sandia's data sheet on Red Storm
on
Cray XT-3 Ships
·
· Score: 1
Here's all the theory and architecture information you might want on Red Storm including a discussion of how Amdahl's Law is wrong. They add overhead due to communications to the equation and make some very interesting proposals about scalability.
http://www.lanl.gov/orgs/ccn/salishan2003/pdf/camp.pdf
Their bottom line to the research: "A well-balanced architecture is nearly insensitive to communications overhead. By contrast a system with weak communications can lose half its power for applications where communications are important."
Conclusion: "For most large scientific and engineering applications the performance is more determined by parallel scalability and less by the speed of individual CPUs. There must be balance between processor, interconnect, and I/ O
performance to achieve overall performance.
To date, only a few tightly-coupled, parallel computer systems have been able to demonstrate a high level of scalability on a broad set of scientific and engineering applications."
I'll leave your knee-jerk reaction of free trade libertarianism alone since it's off-topic, but your criticism of the effort is open season. I wouldn't say this bill represents "the Democrats." The bills are sponsored by the Congressional Black Caucus, the most oft-ignored group in the House. These are genuine attempts at reform by folks who are very cynical of the system and are not merely griping that Al Gore didn't get in. Why don't you RTFA before taking such a jaded view?
If the Condorcet advocates who whine here ran our govermment, we'd still be using 56-bit DES encryption for our most sensitive applications. The complaints offered on the Condorcet website everyone quotes are academic breaks. That means there has yet to be a proven exploit for IRV in closed voting. In the battle for the AES (advanced encyption standard), the advocates of Twofish pointed out that Rijndael (the winner) had academic breaks. Luckily Bruce Schneier, an author of Twofish, didn't have any delusions. Both cyphers were near equals in terms of progress and real world security.
Similarly, IRV isn't perfect and known academic breaks exist, but for all intents and purposes, people know that voting your conscience is still the safest way to make your candidate win in any large IRV election. After all, Australians have experience running IRV since the 20's and they know there's no perfect cheat vote.
Also, there would be several major benefits:
The duopoly would become optional for once.
Americans would become open to further voting reforms.
Voting equipment would support optical scanners and exclude vendors like Diebold who can hardly certify their equipment for duopoly voting, let alone IRV.
People would feel like their vote counted.
These rabid Condorcet advocates would rather see no progress at all in spite of all the hard work and direct lobbying it took for people to get this far in spite of real world opposition from the two parties. If you actually gave a damn, you'd push for an amendment to the bill, not spam slashdot to write armchair letters opposing our hard work.
No, the Founding Fathers weren't stupid, but your response is a common myth. I recommend that article highly for its insight into why the arguments for the electoral college fail. They were making a concession so the southern slaveholding states would receive more votes and support the constitution. I assume you don't intend that we revive slavery, so that there's still a purpose for the electoral college? Since when do State's rights trump fair federal elections anyway?
IBM has already proved that American technology is, at least, as good as Japanese technology despite all the moans and groans about how we have fallen behind Japan upon the introduction of the Earth Simulator.
And how many birthdays has the Earth simulator celebrated before IBM finally beat only its Linpack numbers with a low memory per CPU specialized solution? I think BlueGene still loses the HPC Challenge benchmarks. That isn't what I call winning.
The American mainstream clearly isn't what you think it is. Before you invoke the mainstream, be sure you side with them.
The mainstream supports debates that include third parties, the duopoly doesn't.
The mainstream wants corporations to pay their share of taxes, so middle class taxes can be lowered. The duopoly wants to give them more special interest subsidies and tax breaks.
The mainstream is tired of this war. Neither party wants to take a definitive stance against the war, even the same John Kerry who said in fighting against the war in Vietnam, "This is not the fight of one day or one war, but the fight of our entire lives."
The mainstream isn't voting in many states. A lot of people are frustrated with politics for many good reasons.
Slashdot Mirror
The parent poster posts well for one ignorant of the simplest precepts of marketing. The first things a marketer learns is he must segment a market and only compete in the segments or niches in which competition is profitable. Cray isn't competing directly against clusters because clusters don't have the bandwidth necessary for the sorts of problems Crays are aimed at and Crays tend to be overkill for the problems clusters are aimed at. Cray doesn't seek out customers $.5M for that reason. Anyone who actually uses the supercomputers to solve problems knows that a 50% difference in interconnect speed per single link could mean a 90+% slowdown on a large system using a large program with high overhead. Plain old clusters aren't targetted against Crays, except by some communities that don't buy supercomputers for supercomputer problems anyway, like most Slashdot users. In the supercomputer world, MTTI is everything! That means mean time to interrupt. A bad memory module or a CPU fan blowing out on your single CPU might happen every 3 years on average, but multiply these sorts of problems by 10,000 CPUs on a supercomputer and your cluster will never get any useful work done before something goes out and it crashes. Disclaimer: I worked on the X1/X1e, which is still faster than any other chip on select problems which vectorize well. I agree that the AMD partnership was and continues to be an excellent decision, but it only says that AMD does SCALAR performance better, not everything!
The parent poster needs to be reminded that a large chip manufacturer like Intel, IBM, and AMD makes much more than CPUs! They play a fundamental role in the design and system architecture of the machines built out of their chips. Interfaces like Hypertransport, PCI Express, and DDR are the work of these chip giants. To claim that changing the fundamental design of the CPUs has anything to do with the interaction of a supercomputer company and AMD is naive. Far more likely are changes in Hypertransport, interfaces to memory, or other bus-level projects that are more useful to a supercomputer vendor looking for the best possible overall system bandwidth anyway.
There is no solid definition of what differentiates a cluster, but it's a bit misleading and confusing to refer to an architecture that doesn't have any of the characteristics of a traditional cluster as a cluster. Most clusters have a star or bus network topology of Infiniband, Myrinet, or Ethernet. An X1 uses a 3D torus of redundant, high-bandwidth links. Each node is a board in a backplane, not just a rackmount system. While defining a cluster as a machine where each node has its own OS is a convienient definition, that also includes things which clearly aren't true clusters like networks of workstations and many MPPs.
It's very nice to see someone who gets it. One of the more interesting links off the article is Sandia's reevaluation of Amdahl's Law. If you read Sandia's other research involving the Red Storm project, they point out that Amdahl's Law doesn't have to be as big a killer as previously thought on truly massive systems when paired up with a system that has good balance in memory, network, and processor bandwidth. Most clusters have drastically worse balance than custom MPPs and it causes many algorithms to suffer. The problem has to match the computer or your cluster will never solve it. For example, a problem that requires excessive message passing won't run well on Blue Gene, which has 1/22nd the network bandwidth of Red Storm. The real advantage of Blue Gene is in low power and tight packaging. There's a common misunderstanding that Linpack numbers = real performance. Unless your company gets paid everytime Linpack ends, you can't judge the value of an HPC system by the top 500. For many problems, the Earth Simulator is still winning.
if only it had compatibility for the /pizza command from Everquest 2. Until they add the ability to bring up a pizza ordering screen with only 7 keystrokes, I'll stick with browsing the internet using my copy of Everquest 2, which is f'ing cool according to other slashdot users. Why would anyone ever want to use anything other than EQ2? Happy birthday Opera and thanks, but no thanks.
A college roommate was impressed by how easily I built my own computers and got too anxious to install his DDR memory in his new PC when he got home that he didn't wait for an expert to arrive. He managed to jam the memory in backwards against the key as hard as he could until he heard a snap. Both the motherboard and memory started on fire.
This is a great site to get other people's stories from: http://rinkworks.com/stupid/
Prior art is not affected by first-to-file. You can still invalidate any patent with prior art under a first-to-file system. The bill (at least in the version I read) also allows unrelated third party prior art challenges, which weren't allowed before.
http://democracyrising.us/content/view/196/164/ This is an excellent article and it's the best I've read on the draft rumors. Summary: The rumors of secret draft plans are only rumors, but the military can't keep using mercenaries to account for their poor recruiting numbers. Sooner or later, the merc recruiters are hiring the same people at higher cost who could've served in the military and bodies are needed on the ground in the field. Like with police, you need support people for every person on duty. You can retrain some troops for infantry combat and ship them, but you can't cannibalize all corps and lose the support people to win the war. Nevermind that we went to war without any clear victory conditions set.
I suggest to Lucas that he target the Slashdot audience directly by making Natalie Portman naked on screen. That is if they can still stomach her after she turns evil.
You're implying that if it doesn't take place on top of or underneath a typical hardcore Slashdot user's desktop, it matters. If a tree falls in the forest and it doesn't run Debian...
Opera 8 is a free upgrade to Opera 7 so the point is moot.
Firefox costs money, too. A 1GB dual channel RAM kit costs about $200.
Why does Open Source have to beat competitors who fill a niche and really try to do their jobs well? Do we have to release a faster, lighter, Minimo (Moperilla?) browser simply because commercial competitors are evil or is it because of the spirit of advancing the art? It's fine to admit that the competition is better at something and in fact, it's required in order to improve. Why not admit that the Firefox interface is still pretty ugly?
Who cares if they aren't open source? I just need something fast with real sessions and MDI that doesn't hog my memory or invite spyware in. Web standards only matter when things actually don't work. Is anyone reading this tripe an engineer who actually cares about best tool for the job anymore? Some people use browsers and operating systems to do paying work. So what's wrong with using Linux and Opera?
That's funny. I paid that for Opera a few years ago rather than use Firefox for free and came out ahead.
If you look at the Nullstone website, they have an excellent list of optimizations to expect from a compiler and a few to hope for. It must be remembered, however, that some of these can produce wrong answers such as illegal hoisting if the compiler isn't tested thoroughly enough. You must remember that you can make code very, very fast if you don't care about correct answers and useful results. Remember your priorities!
Whether you support Badnarak (as so many meta-moderators), Nader, Cobb, or even Kerry, I think we can all agree that the wrong issues are being discussed in the media and that this election is a sham. As Ralph Nader said in Harlem a couple weeks ago, there are thirty times as many serious birdwatchers in America as Congress-watchers. Don't be a fool and let yourself be sucked into minor differences between the two candidates when both support unjust war and the continuing destruction of our democracy. Counterterrorism starts at home.
Their bottom line to the research: "A well-balanced architecture is nearly insensitive to communications overhead. By contrast a system with weak communications can lose half its power for applications where communications are important."
Conclusion: "For most large scientific and engineering applications the performance is more determined by parallel scalability and less by the speed of individual CPUs. There must be balance between processor, interconnect, and I/ O performance to achieve overall performance. To date, only a few tightly-coupled, parallel computer systems have been able to demonstrate a high level of scalability on a broad set of scientific and engineering applications."
I'll leave your knee-jerk reaction of free trade libertarianism alone since it's off-topic, but your criticism of the effort is open season. I wouldn't say this bill represents "the Democrats." The bills are sponsored by the Congressional Black Caucus, the most oft-ignored group in the House. These are genuine attempts at reform by folks who are very cynical of the system and are not merely griping that Al Gore didn't get in. Why don't you RTFA before taking such a jaded view?
Similarly, IRV isn't perfect and known academic breaks exist, but for all intents and purposes, people know that voting your conscience is still the safest way to make your candidate win in any large IRV election. After all, Australians have experience running IRV since the 20's and they know there's no perfect cheat vote.
Also, there would be several major benefits:
- The duopoly would become optional for once.
- Americans would become open to further voting reforms.
- Voting equipment would support optical scanners and exclude vendors like Diebold who can hardly certify their equipment for duopoly voting, let alone IRV.
- People would feel like their vote counted.
These rabid Condorcet advocates would rather see no progress at all in spite of all the hard work and direct lobbying it took for people to get this far in spite of real world opposition from the two parties. If you actually gave a damn, you'd push for an amendment to the bill, not spam slashdot to write armchair letters opposing our hard work.No, the Founding Fathers weren't stupid, but your response is a common myth. I recommend that article highly for its insight into why the arguments for the electoral college fail. They were making a concession so the southern slaveholding states would receive more votes and support the constitution. I assume you don't intend that we revive slavery, so that there's still a purpose for the electoral college? Since when do State's rights trump fair federal elections anyway?