Here's the text of Linus' message (slighly edited to get around the "Lameness filter", which itself is pretty lame):
Ok. I didn't make 2.4.0 in 2000. Tough. I tried, but we had some last-minute stuff that needed fixing (ie the dirty page lists etc), and the best I can do is make a prerelease.
There's a 2.4.0-prerelease out there, and this is basically it. I want people to test it for a while, and I want to give other architectures the chance to catch up with some of the changes, but read my lips: no more recounts. There is no "prerelease1", to become "prerelease2" and so on.
One thing other architectures will want to catch up with is the changes to handle 2GHz+ machines, which due to overflow issues caused "loops_per_sec" to become "loops_per_jiffy". And some architectures have not had much chance to synchronize with me due to other fires to put out.
Give it your worst. After you recover from being hung-over, of course.
You're missing my point. I'm saying that for many many applications, a hot-swappable RAID array build out of FibreChannel or what have you gets you nothing.
If all you need is a place to temporarily stash some DNA input files (which are certainly duplicated somewhere else) than cheap-as-shit IDE
drives are just fine for what you're.
And I haven't forgotten about controllers, cabinets, powersupplies, etc. Assuming 1,200 machines, you can put 40 gig IDE drives in each machine and have 48 terabytes of disk. Use a 2U or even a 1U case and you don't need anything else, other than the few buck IDE cable.
1. It won't be the second fastest supercomputer ever built until it performs a benchmark that proves it. Both ASCI Blue machines are theoretically *faster* than ASCI Red, but they still haven't beaten Red in Linpack.
2. 50 Terabytes of disk isn't really that big of a deal. 80 Gig Drives are $250 off pricewatch, so we're talking roughly 3k per terabyte, or only $150,000 for the disk, which in the high-performance community really isn't all that much money. (And don't whine to me about "Well, super-duper-ultra-scsi7 isn't that cheap" because many many apps simply don't need fast disk - they may very well be CPU bound or maybe memory bandwidth bound.)
It's all a matter of how secure your sandbox is. In fact, it doesn't matter whether your code is Java or in native code - the only reason all these
distributed pay-you-for-cycles companies are choosing Java is not because it's more secure, but because they've got someone else to blame when a security hole is discovered. It doesn't really matter if it's the JVM or something else that's overwriting my files.
VMware, plex86, User-Mode-Linux - they're all on the right track: trap on the privilegded instructions and safely emulate them. Every bit as
safe as Java, but without the speed penalty. (This assumes of course that VMware, plex86, and UML have all the instructions handled safely, which can be hard to do on the X86 instruction set. But I'd guess that Sun's JVM has at least as many bugs and security problems as VMware does.)
What you couldn't pay me enough to do is to let anonymous third parties run my code. These distributed-processing people need to take a lesson from the online gaming community: Your clients will cheat, and they are smarter than you. The only way to catch them cheating is to re-run the computation, and don't bother with encryption or obfuscation, because remember, your clients are smarter than you. Also, don't even bother to try and keep "sensitive" data hidden from the people running your computation: you can't. Only send things to the remote clients that you don't care if they see, because they're going to.
The problem with re-running your computation is, of course, that it takes more Compute Time. Sure, you can just check up a percentage to take a statistical guess, but how much is your computation worth to you? If you're willing to pay for it you're better off to check everything again, because processing power is "cheap." Even still, for my serious calculations I'll do them in-house, thank you.
Umm, if you're getting Cease-and-desist letters, you really probably shouldn't just ignore them. Even if they are totally full of shit, the US legal system can be nasty.
IANAL, and Ask Slashdot most certainly is not either.
You just run a renderer per machine. (Of course, you've got to be able to afford a license per machine, but that's got nothing to do with it's abaility to scale)
People are working hard, and spending plenty of money solving these problems - check out the Alliance - particularly Globus and Condor. We're doing real-world science now. The other day we solved QAP30, which is was a big problem in the optimization field. We've got people doing particle physics simulations, protein conformation, computer architecture simulation - the list goes on and on.
People need to stop looking at the d.net/Seti@home problems as the only model for Internet computing. They're not that hard of problems. What makes them neat is that they've got lots of CPU's. (SETI is cool because it's space and aliens and everything, but RC5-64 is just plain stupid - they're proving that 64 bit RC5 is 256 times harder to crack than 56bit RC5. Yawn.)
Numerical accuracy is a concern. Latency is a concern - but not for a a huge set of problems. You don't need a T3E for Monte Carlo simulations, and you shouldn't try and put your finite-element simulations all around the world. Networks are getting faster and faster, so code size is really not an issue today for anyone on a real network (ie vBNS.) Data size can be a problem, but again, networks are getting faster, and you can prestage a lot of the data. If your code is too sensitive to risk distributing, then no amount of technological progress is going to change it. User security is not that difficult of a problem - it's not too hard to sandbox an application on a decent OS. And as for FORTRAN, I don't see what the problem is. Processors don't run C or FORTRAN or Pascal, and the FORTRAN compilers still produce some pretty tight code.
The Internet makes great sense for high-performance computing, for the right problems.
Record companies are outdated. Why keep them around? A CD costs $15 - and plenty of people are willing to pay that. How much of that money do you see? A buck or two, if that?
In the music business, the only people that really matter are the band and the recording engineer. The technology exists now that you can tell the record company to fuck off, and you can go straight to us. We're still willing to pay $15 bucks for your album, and you'd get to keep all of it.
The RIAA is fighting technology tooth and nail because they know that they're irrelavant now, and the billions of dollars the music industy produces is going to stop going to the stockholders and start going to the artists. They don't care about you. They don't care about us.
In the end, they're going to lose - this is an arms race, and if the lawyers destroy Napster, another will take over, immune from whatever destroys Napster. If the lawyers destroy that one, it will be replaced again.
Why not be the band that goes down as the band that changed the way the industry workes, and gets rid of the cancer of record companies?
That's true in closed-source software as well. The solution is to never trust the client in a multiplayer game, and do all the game simulation server-side. If you don't want a client to potentially know where something is, don't tell them.
Security through obscurity is never the right answer.
A lot of people make claims like "games can't be free software" for various reasons, but I've always thought it's because we've never had a hugely successful Free-speech game -- when I say successful I mean on the scale of Quake or Warcraft or NHL '99. It seems that most of the "free software community" is perfectly willing to give games an exception to the "we won't use it unless it's free" philosophy.
Is there anything different about the Gaming marketplace that prevents it from being free? Should the community refuse to play non-free games (ie no Quake III - only Quake I, etc.) Will you go to Texas and try and convince Carmack to open the source to Q3? (The industry is pretty good about following in whatever he does)
Condor, from the University of Wisconsin, should have been listed on page two. Condor is a high-throughput computing system, that runs on UNIX (virtually all flavors) and NT. We support MPI and PVM. We can run regular jobs, or you can relink with our libraries and get transparent checkpointing and remote I/O. You can use sockets in your job.
You don't need to have a dedicated cluster - Condor started life as a scavenger of idle workstations. We run Condor on every workstation here at CS, and routinely recover several thousand CPU-hours a day that otherwise would have been wasted. You can configure Condor to run with any policy you want on a per-workstation level - only run jobs at night, only run jobs from this group, only run jobs if the wind is blowing from the west - whatever makes sense to the workstation's owner.
I've been on-again, off-again working on clone/successor for Tradewars for sometime now. I've been planning on calling it Tradewars 2112, and it's under the LGPL. It's a client/server game, currently with a multi-threaded UNIX server, and a crappy Java client (though there will be no reason why you can't write a Win32 or GTK or Cocoa client)
I work for a research project in distributed systems, and some of our biggest users are 3D shops, (We're working on turning the entire UW Comp Sci and College of Engineering labs into one massive 1000-node render farm) so I know all about render farms.
This does not enable a cheap render farm - you've been able to do this with NT for a long time. Yes, it's nice because it's on Linux, so the cost per node is an NT license cheaper, the nodes are more stable, yada yada yada - #include . But it really doesn't suddenly enable real cheap render farms. (Check out Lemon)
People are singing and dancing in the streets because they think this signifies some sort of major effort and commitment to Linux by Alias/Wavefront. It doesn't - if this took more than a few days of an engineer's time then they have a really fucked up renderer. The real show of a commitment to Linux is the creation tools - even just a target date would be something. Until then it's really just hype. (Though granted, nice to hear for render farm admins)
This isn't really that big of a deal, because it's just the renderer - you still need an NT or IRIX box to actually create any content. When they port all of their tools over, then it will be a big deal. (The maya renderer is a command-line tool, and if it was any big deal to port then I'd be really suprised.)
My hometown is Chippewa Falls, WI, where Cray has all of their manufacturing facilities, so I've seen quite a few of the machines. I didn't think the X-MP was that comfortable, actually.
Far cooler is the Cray 2 - the one that had the clear case and the waterfall heat exchanger for the florinert.
As far as finding a cabinet, you might try comp.sys.super...
No one's made any noise about filtering in Wisconsin - yet Sometimes I think I'd like to head them off at the pass. My state legislature meets just up the block, and they're in session now.
My proposal:
All libraries must provide filtering software.
That software must be off by default.
Anyone may request the software be turned on for their session
The software must be turned on for anyone under 18, unless they have permission from their parents to have it turned off. (The parents can file a blanket permission form for every session, or they can have it for individual sessions)
The list of all sites and keywords blocked must be made availble upon demand by any library patron
The library must have the ability to add or remove sites easily
(This one isn't terribly important, but would be nice) The source code must be availble.
If no such software exists, then the state shall provide funds for it.
This should satisify everyone who has legal rights - No one is forced to use it, but it's availble for those people are offended easily. It also "protects" children, but provides a way for more enlightened parents to allow their children to see everything, and make their own choices as to what's appropriate. It doesn't give children a real say in the matter, but since they have a different legal standard, that's OK.
I really thought the best superbowl commerical I ever saw was the one for Compuserve a couple years back - there was nothing but a black screen, and the busy signal sound - this was back when AOL had just switched to flat rate, and no one could get though.
At least with Wisconsin's laws. Wisconsin allows you to register at the polls on Election day, establishing residency can be done by just showing a lease or even a piece of mail. In Madison, most everyone lives within a mile or less of their polling place (I live about 300 feet from mine) Voter turnout in my district is so heavy that they ran out of ballots in the last Congressional elections!
America will fight online voting for the same reason it will profess support for the family farm - going to the polls is part of our national identity. Sure, the future may be megafarms and evoting, but you can be damned sure a lot of us are going to fight it tooth and nail.
Fluent, the CFD code that was used to do the simulation, has a version that runs on clusters. (We had combustion people doing it all the time on our cluster)
It wouldn't really matter for this country. Realistically, the world is one big warez site - if you don't want a legal copy of NT or 3DSMAX, you can get a pirate copy. Fortunately for software companies, most companies try to make sure that they're not violating any copyright laws, and have licenses for everything that they use. As long was we have Lawyers and the SPA, there'll always be a US market.
Slashdot Mirror
Ok. I didn't make 2.4.0 in 2000. Tough. I tried, but we had some last-minute stuff that needed fixing (ie the dirty page lists etc), and the best I can do is make a prerelease.
There's a 2.4.0-prerelease out there, and this is basically it. I want people to test it for a while, and I want to give other architectures the chance to catch up with some of the changes, but read my lips: no more recounts. There is no "prerelease1", to become "prerelease2" and so on.
One thing other architectures will want to catch up with is the changes to handle 2GHz+ machines, which due to overflow issues caused "loops_per_sec" to become "loops_per_jiffy". And some architectures have not had much chance to synchronize with me due to other fires to put out.
Give it your worst. After you recover from being hung-over, of course.
Linus
You're missing my point. I'm saying that for many many applications, a hot-swappable RAID array build out of FibreChannel or what have you gets you nothing.
If all you need is a place to temporarily stash some DNA input files (which are certainly duplicated somewhere else) than cheap-as-shit IDE
drives are just fine for what you're.
And I haven't forgotten about controllers, cabinets, powersupplies, etc. Assuming 1,200 machines, you can put 40 gig IDE drives in each machine and have 48 terabytes of disk. Use a 2U or even a 1U case and you don't need anything else, other than the few buck IDE cable.
1. It won't be the second fastest supercomputer ever built until it performs a benchmark that proves it. Both ASCI Blue machines are theoretically *faster* than ASCI Red, but they still haven't beaten Red in Linpack.
2. 50 Terabytes of disk isn't really that big of a deal. 80 Gig Drives are $250 off pricewatch, so we're talking roughly 3k per terabyte, or only $150,000 for the disk, which in the high-performance community really isn't all that much money. (And don't whine to me about "Well, super-duper-ultra-scsi7 isn't that cheap" because many many apps simply don't need fast disk - they may very well be CPU bound or maybe memory bandwidth bound.)
It's not difficult to safely run untrusted code.
It's all a matter of how secure your sandbox is. In fact, it doesn't matter whether your code is Java or in native code - the only reason all these
distributed pay-you-for-cycles companies are choosing Java is not because it's more secure, but because they've got someone else to blame when a security hole is discovered. It doesn't really matter if it's the JVM or something else that's overwriting my files.
VMware, plex86, User-Mode-Linux - they're all on the right track: trap on the privilegded instructions and safely emulate them. Every bit as
safe as Java, but without the speed penalty. (This assumes of course that VMware, plex86, and UML have all the instructions handled safely, which can be hard to do on the X86 instruction set. But I'd guess that Sun's JVM has at least as many bugs and security problems as VMware does.)
What you couldn't pay me enough to do is to let anonymous third parties run my code. These distributed-processing people need to take a lesson from the online gaming community: Your clients will cheat, and they are smarter than you. The only way to catch them cheating is to re-run the computation, and don't bother with encryption or obfuscation, because remember, your clients are smarter than you. Also, don't even bother to try and keep "sensitive" data hidden from the people running your computation: you can't. Only send things to the remote clients that you don't care if they see, because they're going to.
The problem with re-running your computation is, of course, that it takes more Compute Time. Sure, you can just check up a percentage to take a statistical guess, but how much is your computation worth to you? If you're willing to pay for it you're better off to check everything again, because processing power is "cheap." Even still, for my serious calculations I'll do them in-house, thank you.
-Erik
NASDAQ
The name changed to be the "Annual Linux Showcase", not the Atlanta linux showcase. My guess is because it's not going to be in Atlanta
next year.
How long did that take? The slashcode did not take a very "development out in the open" sort of approach.
IANAL, and Ask Slashdot most certainly is not either.
You just run a renderer per machine. (Of course, you've got to be able to afford a license per machine, but that's got nothing to do with it's abaility to scale)
People need to stop looking at the d.net/Seti@home problems as the only model for Internet computing. They're not that hard of problems. What makes them neat is that they've got lots of CPU's. (SETI is cool because it's space and aliens and everything, but RC5-64 is just plain stupid - they're proving that 64 bit RC5 is 256 times harder to crack than 56bit RC5. Yawn.)
Numerical accuracy is a concern. Latency is a concern - but not for a a huge set of problems. You don't need a T3E for Monte Carlo simulations, and you shouldn't try and put your finite-element simulations all around the world. Networks are getting faster and faster, so code size is really not an issue today for anyone on a real network (ie vBNS.) Data size can be a problem, but again, networks are getting faster, and you can prestage a lot of the data. If your code is too sensitive to risk distributing, then no amount of technological progress is going to change it. User security is not that difficult of a problem - it's not too hard to sandbox an application on a decent OS. And as for FORTRAN, I don't see what the problem is. Processors don't run C or FORTRAN or Pascal, and the FORTRAN compilers still produce some pretty tight code.
The Internet makes great sense for high-performance computing, for the right problems.
It seems to me that Dijkstra's algorithm should be on there. It's certainly as important as quick-sort...
In the music business, the only people that really matter are the band and the recording engineer. The technology exists now that you can tell the record company to fuck off, and you can go straight to us. We're still willing to pay $15 bucks for your album, and you'd get to keep all of it.
The RIAA is fighting technology tooth and nail because they know that they're irrelavant now, and the billions of dollars the music industy produces is going to stop going to the stockholders and start going to the artists. They don't care about you. They don't care about us.
In the end, they're going to lose - this is an arms race, and if the lawyers destroy Napster, another will take over, immune from whatever destroys Napster. If the lawyers destroy that one, it will be replaced again.
Why not be the band that goes down as the band that changed the way the industry workes, and gets rid of the cancer of record companies?
That's true in closed-source software as well. The solution is to never trust the client in a multiplayer game, and do all the game simulation server-side. If you don't want a client to potentially know where something is, don't tell them.
Security through obscurity is never the right answer.
A lot of people make claims like "games can't be
free software" for various reasons, but I've always thought it's because we've never had a hugely successful Free-speech game -- when I say successful I mean on the scale of Quake or Warcraft or NHL '99. It seems that most of the "free software community" is perfectly willing to give games an exception to the "we won't use it unless it's free" philosophy.
Is there anything different about the Gaming marketplace that prevents it from being free? Should the community refuse to play non-free games (ie no Quake III - only Quake I, etc.) Will you go to Texas and try and convince Carmack to open the source to Q3? (The industry is pretty good about following in whatever he does)
You don't need to have a dedicated cluster - Condor started life as a scavenger of idle workstations. We run Condor on every workstation here at CS, and routinely recover several thousand CPU-hours a day that otherwise would have been wasted. You can configure Condor to run with any policy you want on a per-workstation level - only run jobs at night, only run jobs from this group, only run jobs if the wind is blowing from the west - whatever makes sense to the workstation's owner.
Best of all, we're free-as-in-beer.
If you have any questions, send us mail at condor-admin@cs.wisc.edu
The page is here
No, it wouldn't, unless you put all of your program into the kernel. You don't want to pay the price of going into and out of the kernel.
Ideally you'll implement everything userlevel, including your networking, so you never go into the kernel.
This does not enable a cheap render farm - you've been able to do this with NT for a long time. Yes, it's nice because it's on Linux, so the cost per node is an NT license cheaper, the nodes are more stable, yada yada yada - #include . But it really doesn't suddenly enable real cheap render farms. (Check out Lemon)
People are singing and dancing in the streets because they think this signifies some sort of major effort and commitment to Linux by Alias/Wavefront. It doesn't - if this took more than a few days of an engineer's time then they have a really fucked up renderer. The real show of a commitment to Linux is the creation tools - even just a target date would be something. Until then it's really just hype. (Though granted, nice to hear for render farm admins)
This isn't really that big of a deal, because it's just the renderer - you still need an NT or IRIX box to actually create any content. When they port all of their tools over, then it will be a big deal. (The maya renderer is a command-line tool, and if it was any big deal to port then I'd be really suprised.)
My hometown is Chippewa Falls, WI, where Cray has
all of their manufacturing facilities, so I've seen quite a few of the machines. I didn't think the X-MP was that comfortable, actually.
Far cooler is the Cray 2 - the one that had the clear case and the waterfall heat exchanger for the florinert.
As far as finding a cabinet, you might try comp.sys.super...
My proposal:
All libraries must provide filtering software.
That software must be off by default.
Anyone may request the software be turned on for their session
The software must be turned on for anyone under 18, unless they have permission from their parents to have it turned off. (The parents can file a blanket permission form for every session, or they can have it for individual sessions)
The list of all sites and keywords blocked must be made availble upon demand by any library patron
The library must have the ability to add or remove sites easily
(This one isn't terribly important, but would be nice) The source code must be availble.
If no such software exists, then the state shall provide funds for it.
This should satisify everyone who has legal rights - No one is forced to use it, but it's availble for those people are offended easily. It also "protects" children, but provides a way for more enlightened parents to allow their children to see everything, and make their own choices as to what's appropriate. It doesn't give children a real say in the matter, but since they have a different legal standard, that's OK.
Comments?
I really thought the best superbowl commerical I ever saw was the one for Compuserve a couple years back - there was nothing but a black screen, and the busy signal sound - this was back when AOL had just switched to flat rate, and no one could get though.
At least with Wisconsin's laws. Wisconsin allows you to register at the polls on Election day, establishing residency can be done by just showing a lease or even a piece of mail. In Madison, most everyone lives within a mile or less of their polling place (I live about 300 feet from mine)
Voter turnout in my district is so heavy that they ran out of ballots in the last Congressional elections!
America will fight online voting for the same reason it will profess support for the family farm - going to the polls is part of our national identity. Sure, the future may be megafarms and evoting, but you can be damned sure a lot of us are going to fight it tooth and nail.
Fluent, the CFD code that was used to do the simulation, has a version that runs on clusters. (We had combustion people doing it all the time on our cluster)
So yes, this can be tied into a Beowulf.
It wouldn't really matter for this country. Realistically, the world is one big warez site - if you don't want a legal copy of NT or 3DSMAX, you can get a pirate copy. Fortunately for software companies, most companies try to make sure that they're not violating any copyright laws, and have licenses for everything that they use. As long was we have Lawyers and the SPA, there'll always be a US market.