"If clustering doesn't work, and Supers are better / cheaper, explain why large companies (Pixar, NVidia,...) Government Labs (Los Alamos National Labs, Sandia National Labs,...) have invested, and are continuing to invest in and support their clusters."
Clusters work great for what a company like Pixar does. The DOE labs, on the other hand, have been using clusters for quite a bit of stuff, and people within DOE have actually been pretty unhappy about the performance they have been getting running some of their application codes on their cluster machines. The fastest cluster machine in the world, the ASCI-Q at Los Alamos, has actually been a bit of a disappointment in the minds of many people. Sure, its #2 on the top 500 list, but many folks were hoping to see a lot more TFLOPS out of it than they've been able to get.
There is a lot of interest in vector machines right now after seeing what Japan was able to do with the Earth Simulator.
While I certainly disagree that you can't build a very high performance computer out a cluster of computers (Linux or otherwise), there is a lot of merit to the fact that clusters just don't scale well for certain classes of applications. Hence the renaissance of the vector supercomputer (ala the Earth Simulator ).
Obviously, this guy is plugging the new Cray X1 architecture, which really is quite promising. For instance, check out this paper by some folks at Oak Ridge National Lab that appeared in Supercomputing 2003.
Of course, since this is Slashdot, I expect that there will be a deluge of posts decrying everything about the new Cray machine because it commits the cardinal sin of NOT USING LINUX. Oh, the horror!
This sounds like a fun event, but I'd be very surprised if they can make this work. Sure, if you get enough computers together, the "theoretical peak" (basically the aggregate FLOP rate of all the computers) can be really high, but for a problem with dependencies like the LINPACK benchmark, communication latency is what is going to limit the runtime when you try to hook a thousand or so laptop computers together using only fast ethernet. I think that even Gigabit ethernet would scale poorly in such a setting (decent bandwidth, but startup time for a communication is slow compared to something like Myrinet or Infiniband).
Maybe it's possible, but I think it would require a pretty creative network topology and some pretty clever re-structuring of the linpack benchmark (which is allowed by the Top 500 list rules, BTW).
I can't agree more with the parent post. Lycoris came bundled with my $200 Wal-Mart box (great buy, by the way!). At first I was impressed with the desktop setup, but then I found that I couldn't do anything useful with Lycoris without paying them extra money. No compiler? According to their bozos on the community website, normal users "don't need a compiler". And don't even think that you can get by without one. It is impossible to find binary packages for it, so you can't install anything. That is, unless you want to pay them money to send you a CD of a bunch of programs that are freely available, open source stuff. (But which you can't download and install yourself because, well, there's no compilers!)
Luckily, with a Knoppix CD my machine quickly because a very nice Debian install!
You're a bit quick to jump on this guy for stating that his wife was having a "boy" rather than just stating that she was having a "baby". Maybe the poster does have some gender bias, but maybe he was just being more specific by using "boy" instead of "baby".
My wife is a physicist and my sister is a mechanical engineer so I certainly don't hold with 1950's gender biases. But I didn't automatically get offended at the poster's possibly innocent statement.
I highly recommended the book mentioned above. It is written at a very intelligent and non-patronizing level, and is scientifically quite eclectic. It's thought-provoking and also a lot of fun. I really enjoyed the great diversity of possible "solutions" to the Fermi paradox that get discussed in the book. Lots of variety makes it hard to get bored, and some of the discussions are fantastic. Particularly good is the "percolation theory" explanation for why it may be impossible to hear from other intelligent life forms even though they may exist.
Give Joseph Black his due credit!
on
Happy Birthday, Atom
·
· Score: 3, Informative
"200 years ago today (Oct. 21) John Dalton revolutionized chemistry by starting the process of turning it into an exact science"
Can't argue with John Dalton having helped revolutionize chemistry, but he didn't start the process of turning it into an exact science. I think that the credit for that probably belongs to British chemist Joseph Black, who founded calorimetry and was one of the first scientists to emphasize quantitative experiments. (Interestingly, at Edinburgh his chemistry chair was unsalaried!)
Another journal such as this that doesn't just fill the coffers of Wiley or Elsevier is a good thing. As other posters have pointed out, though, there are similar free electronic journals out there. One that I haven't seen mentioned is the Electronic Transactions on Numerical Analysis (http://etna.mcs.kent.edu), which has actually been around since 1993!
A big problem with PLoS is that an author is charged $1500 (!!!) to publish in the journal. This is going to bar a lot of people who lack significant funding from publishing in the journal. I don't see how passing the ridiculous costs of journals from subscribers to authors is a very good fix! There are other free, electronic journals out there that don't cost anything to publish in (such as ETNA). I honestly don't know why PLoS charges so much money. The cost of running an e-journal can't be that high: authors don't get paid anything, reviewers don't get paid anything, and many editors don't get paid anything. Money for running a web/ftp server should just about cover it!
Oops. (Re:They forgot worst job number one....)
on
Worst Jobs In Science
·
· Score: 1
"(I mean, Jeez... they're possible treated worst than postdocs and get paid less.)"
OK, that looked illiterate.
possible --> possibly
worst --> worse
See the benefits that my 4+ years of Ph.D. study have gotten me?
"If it manages to complete the cluster before the cut-off date, it will score a Top 5 rank in the Linpack Top 500 Supercomputer List."
Err... I think somebody's getting a bit ahead of themselves here. =) Building parallel computing systems is complicated, and it may end up being quite a bit harder to realize the predicted performance than thought (not an uncommon occurrence). I'll believe it when they have the actual Linpack numbers.
>He had two choices: > >1) lose his life savings. >2) find a pro bono attorney and hope he doesn't lose his life savings and more.
The whole point of a settlement is to AVOID losing all of one's assets. How is agreeing to fork over everything a good settlement? I'm not just trying to argue... I'd like to know. I don't know much about lawsuits.
Also, I'd have to say that this is a high profile case and the student probably could have found a good pro bono lawyer: the case IS getting national coverage.
Could the ACLU have had any interest in this guy's case? It seems to be in line with their basic mission, but they usually seem to fight governments, not private corporations.
Yeah, encountering the "Flood" in Halo is one of the most deliciously terrifying video game experiences that I have encountered. I love how the up to that point you think that the Covenant are what's to be feared, but then you find that the shocking truth is that there's something much, much worse....
I think Strong Bad and company over at www.homestarrunner.com would disagree with you. Yeah, flash gets abused horribly but in the hands of creative people some great stuff can be done with it.
There are some fundamental problems with this proposed business model, but I won't get into those. My problem with this is that spare CPU cycles that they intend to use simply aren't worth very much because of the slow and unreliable nature of the network connectivity that most users will have. While SETI@home and distributed.net work on "embarassingly parallel" problems that require very little communication, many, many problems that people are interesting in paying money to solve require regular communication between nodes and thus some guarantee on the quality of network service. Some amount X of spare CPU cycles on machines using 56K modem connections (or even cable or DSL modems) just isn't worth nearly as much as an equivalent amount of spare CPU cyles on machines connected by something like gigabit ethernet... or even switched fast ethernet.
"I can imagine the use of laser turrets as protection against missiles, but I really can't see the use of a laser mounted in a 747. IMHO, it's way too slow compared to the missiles..."
But a stationary turret, now that's FAST compared to the missles, right?
Just so you know, the Pulitzer is awarded to an author for an individual work that author has done. The Nobel prize for literature is awarded to an author to recognize his achievements over a long career.
Wow, so software developers AREN'T like the ones portrayed in movies? Hard to believe, because all other occupations are portrayed spot-on in movies! That's why so many kids majoring in achaeology enter college with such strong bull-whip wielding skills. And aspiring paleontologists are ready to combat fearsom velociraptors. And so on... you get the idea.
Surely there must be some kind of laws that prevent inaccuracies in Hollywood movies. Right?
You're right, Ultima 9 was a completely sorry thing when it left the door. But years after its release, I'm playing through it now and am actually finding it enjoyable. The reason? The devoted fans who released patches for the game, the most important being the dialogue patch, which changes the dialogue, books, etc. (and to some extent, the plot) so that they reflect the rich history of the Ultima saga (and fix a lot of contradictions!). If you played Ultima 9 and found it as sorry as I did, give it another try with the fan patches. Sure, there's still plenty of problems, but the dialogue patch makes the game at least feel (mostly) like an Ultima game.
Check it out at
http://reconstruction.voyd.net/zips/u9fanpatch16 0. zip
Undoubtedly, Strong Bad's email is the best part of the site, but how can you not like "A Jumping Jack Contest" or "The Luau"? Some of the 'toons are pretty funny, and they add a lot to the appreciation of Strong Bad. Some of the menus are fantastic. Not to mention some of the incredibly stupid "music videos". And check out some of the old "Marzipan's answering machine" episodes under "games". And while you're there, try out the "Dancin' Bubs" game.
There's pleny of good stuff there. You just have to dig around for it, and there's a LOT of content to dig through.
Cool hack; very bad performance evaluation
on
Xbox Linux Cluster
·
· Score: 2, Insightful
It's cool that this fellow managed to hack this Xbox cluster together. But his methods for performance testing it are ridiculous. Beyond the lack of any meaningful benchmarking, hardware tuning, etc. there's a failure to consider why someone might want to stick a bunch of Xbox's together in the first place. Think about it...what is all of the hardware in the Xbox really suppose to support? GRAPHICS, duh. But this "performance evaluation" doesn't try to do anything with the graphics hardware. A more realistic evaluation of the benefits of clustering these would try to take advantage of the graphics processing power of the hardware. I'll guarantee that a $200 Wal-Mart PC would do much worse in parallel, realtime graphics rendering computations.
Very true (Re:Behind the times...)
on
Potato Bazookas
·
· Score: 1
Yes, when these become a popular thing that incautious people are playing with, it can be very dangerous. Some idiot frat boys in my home town were playing with one and seriously injured a passerby. The person they hit lost an eye, and suffered other injury as well.
I'm a nature enthusiast and a conservationist, and I agree wholeheartedly that areas devoid of life are worth preserving. Some of my favorite places on earth are alpine mountain summits, and there's certainly not much alive up there. But, honestly, the environmental impact of the proposed activities is just incredibly small. You must have some incredibly stringent standards for determining if something "violates" an area.
Slashdot Mirror
"If clustering doesn't work, and Supers are better / cheaper, explain why large companies (Pixar, NVidia, ...) Government Labs (Los Alamos National Labs, Sandia National Labs, ...) have invested, and are continuing to invest in and support their clusters."
Clusters work great for what a company like Pixar does. The DOE labs, on the other hand, have been using clusters for quite a bit of stuff, and people within DOE have actually been pretty unhappy about the performance they have been getting running some of their application codes on their cluster machines. The fastest cluster machine in the world, the ASCI-Q at Los Alamos, has actually been a bit of a disappointment in the minds of many people. Sure, its #2 on the top 500 list, but many folks were hoping to see a lot more TFLOPS out of it than they've been able to get.
There is a lot of interest in vector machines right now after seeing what Japan was able to do with the Earth Simulator.
While I certainly disagree that you can't build a very high performance computer out a cluster of computers (Linux or otherwise), there is a lot of merit to the fact that clusters just don't scale well for certain classes of applications. Hence the renaissance of the vector supercomputer (ala the Earth Simulator ).
Obviously, this guy is plugging the new Cray X1 architecture, which really is quite promising. For instance, check out this paper by some folks at Oak Ridge National Lab that appeared in Supercomputing 2003.
Of course, since this is Slashdot, I expect that there will be a deluge of posts decrying everything about the new Cray machine because it commits the cardinal sin of NOT USING LINUX. Oh, the horror!
This sounds like a fun event, but I'd be very surprised if they can make this work. Sure, if you get enough computers together, the "theoretical peak" (basically the aggregate FLOP rate of all the computers) can be really high, but for a problem with dependencies like the LINPACK benchmark, communication latency is what is going to limit the runtime when you try to hook a thousand or so laptop computers together using only fast ethernet. I think that even Gigabit ethernet would scale poorly in such a setting (decent bandwidth, but startup time for a communication is slow compared to something like Myrinet or Infiniband).
Maybe it's possible, but I think it would require a pretty creative network topology and some pretty clever re-structuring of the linpack benchmark (which is allowed by the Top 500 list rules, BTW).
I can't agree more with the parent post. Lycoris came bundled with my $200 Wal-Mart box (great buy, by the way!). At first I was impressed with the desktop setup, but then I found that I couldn't do anything useful with Lycoris without paying them extra money. No compiler? According to their bozos on the community website, normal users "don't need a compiler". And don't even think that you can get by without one. It is impossible to find binary packages for it, so you can't install anything. That is, unless you want to pay them money to send you a CD of a bunch of programs that are freely available, open source stuff. (But which you can't download and install yourself because, well, there's no compilers!)
Luckily, with a Knoppix CD my machine quickly because a very nice Debian install!
You're a bit quick to jump on this guy for stating that his wife was having a "boy" rather than just stating that she was having a "baby". Maybe the poster does have some gender bias, but maybe he was just being more specific by using "boy" instead of "baby".
My wife is a physicist and my sister is a mechanical engineer so I certainly don't hold with 1950's gender biases. But I didn't automatically get offended at the poster's possibly innocent statement.
I highly recommended the book mentioned above. It is written at a very intelligent and non-patronizing level, and is scientifically quite eclectic. It's thought-provoking and also a lot of fun. I really enjoyed the great diversity of possible "solutions" to the Fermi paradox that get discussed in the book. Lots of variety makes it hard to get bored, and some of the discussions are fantastic. Particularly good is the "percolation theory" explanation for why it may be impossible to hear from other intelligent life forms even though they may exist.
"200 years ago today (Oct. 21) John Dalton revolutionized chemistry by starting the process of turning it into an exact science"
Can't argue with John Dalton having helped revolutionize chemistry, but he didn't start the process of turning it into an exact science. I think that the credit for that probably belongs to British chemist Joseph Black, who founded calorimetry and was one of the first scientists to emphasize quantitative experiments. (Interestingly, at Edinburgh his chemistry chair was unsalaried!)
Another journal such as this that doesn't just fill the coffers of Wiley or Elsevier is a good thing. As other posters have pointed out, though, there are similar free electronic journals out there. One that I haven't seen mentioned is the Electronic Transactions on Numerical Analysis (http://etna.mcs.kent.edu), which has actually been around since 1993!
A big problem with PLoS is that an author is charged $1500 (!!!) to publish in the journal. This is going to bar a lot of people who lack significant funding from publishing in the journal. I don't see how passing the ridiculous costs of journals from subscribers to authors is a very good fix! There are other free, electronic journals out there that don't cost anything to publish in (such as ETNA). I honestly don't know why PLoS charges so much money. The cost of running an e-journal can't be that high: authors don't get paid anything, reviewers don't get paid anything, and many editors don't get paid anything. Money for running a web/ftp server should just about cover it!
"(I mean, Jeez... they're possible treated worst than postdocs and get paid less.)"
OK, that looked illiterate.
possible --> possibly
worst --> worse
See the benefits that my 4+ years of Ph.D. study have gotten me?
GRADUATE STUDENT.
(I mean, Jeez... they're possible treated worst than postdocs and get paid less.)
"Unless of course it's a $5000 ti or carbon bike then it's just better."
Sadly, its still cheaper when you factor in the cost of the six segway batteries.
BTW, if you can do the Mt. Washington climb in an hour, you ought to be a pro bicylist. =) Check out the times for the pros in the annual hill climb.
"If it manages to complete the cluster before the cut-off date, it will score a Top 5 rank in the Linpack Top 500 Supercomputer List."
Err... I think somebody's getting a bit ahead of themselves here. =) Building parallel computing systems is complicated, and it may end up being quite a bit harder to realize the predicted performance than thought (not an uncommon occurrence). I'll believe it when they have the actual Linpack numbers.
>He had two choices:
>
>1) lose his life savings.
>2) find a pro bono attorney and hope he doesn't lose his life savings and more.
The whole point of a settlement is to AVOID losing all of one's assets. How is agreeing to fork over everything a good settlement? I'm not just trying to argue... I'd like to know. I don't know much about lawsuits.
Also, I'd have to say that this is a high profile case and the student probably could have found a good pro bono lawyer: the case IS getting national coverage.
Could the ACLU have had any interest in this guy's case? It seems to be in line with their basic mission, but they usually seem to fight governments, not private corporations.
Yeah, encountering the "Flood" in Halo is one of the most deliciously terrifying video game experiences that I have encountered. I love how the up to that point you think that the Covenant are what's to be feared, but then you find that the shocking truth is that there's something much, much worse....
...look so much like "Shrek II"? Maybe the movie will be good, but, come on, the Hulk looks like a big cartoon.
I think Strong Bad and company over at www.homestarrunner.com would disagree with you. Yeah, flash gets abused horribly but in the hands of creative people some great stuff can be done with it.
There are some fundamental problems with this proposed business model, but I won't get into those. My problem with this is that spare CPU cycles that they intend to use simply aren't worth very much because of the slow and unreliable nature of the network connectivity that most users will have. While SETI@home and distributed.net work on "embarassingly parallel" problems that require very little communication, many, many problems that people are interesting in paying money to solve require regular communication between nodes and thus some guarantee on the quality of network service. Some amount X of spare CPU cycles on machines using 56K modem connections (or even cable or DSL modems) just isn't worth nearly as much as an equivalent amount of spare CPU cyles on machines connected by something like gigabit ethernet... or even switched fast ethernet.
"I can imagine the use of laser turrets as protection against missiles, but I really can't see the use of a laser mounted in a 747. IMHO, it's way too slow compared to the missiles..."
But a stationary turret, now that's FAST compared to the missles, right?
Just so you know, the Pulitzer is awarded to an author for an individual work that author has done. The Nobel prize for literature is awarded to an author to recognize his achievements over a long career.
Wow, so software developers AREN'T like the ones portrayed in movies? Hard to believe, because all other occupations are portrayed spot-on in movies! That's why so many kids majoring in achaeology enter college with such strong bull-whip wielding skills. And aspiring paleontologists are ready to combat fearsom velociraptors. And so on... you get the idea.
Surely there must be some kind of laws that prevent inaccuracies in Hollywood movies. Right?
You're right, Ultima 9 was a completely sorry thing when it left the door. But years after its release, I'm playing through it now and am actually finding it enjoyable. The reason? The devoted fans who released patches for the game, the most important being the dialogue patch, which changes the dialogue, books, etc. (and to some extent, the plot) so that they reflect the rich history of the Ultima saga (and fix a lot of contradictions!). If you played Ultima 9 and found it as sorry as I did, give it another try with the fan patches. Sure, there's still plenty of problems, but the dialogue patch makes the game at least feel (mostly) like an Ultima game.
6 0. zip
Check it out at
http://reconstruction.voyd.net/zips/u9fanpatch1
Undoubtedly, Strong Bad's email is the best part of the site, but how can you not like "A Jumping Jack Contest" or "The Luau"? Some of the 'toons are pretty funny, and they add a lot to the appreciation of Strong Bad. Some of the menus are fantastic. Not to mention some of the incredibly stupid "music videos". And check out some of the old "Marzipan's answering machine" episodes under "games". And while you're there, try out the "Dancin' Bubs" game.
There's pleny of good stuff there. You just have to dig around for it, and there's a LOT of content to dig through.
It's cool that this fellow managed to hack this Xbox cluster together. But his methods for performance testing it are ridiculous. Beyond the lack of any meaningful benchmarking, hardware tuning, etc. there's a failure to consider why someone might want to stick a bunch of Xbox's together in the first place. Think about it...what is all of the hardware in the Xbox really suppose to support? GRAPHICS, duh. But this "performance evaluation" doesn't try to do anything with the graphics hardware. A more realistic evaluation of the benefits of clustering these would try to take advantage of the graphics processing power of the hardware. I'll guarantee that a $200 Wal-Mart PC would do much worse in parallel, realtime graphics rendering computations.
Yes, when these become a popular thing that incautious people are playing with, it can be very dangerous. Some idiot frat boys in my home town were playing with one and seriously injured a passerby. The person they hit lost an eye, and suffered other injury as well.
I'm a nature enthusiast and a conservationist, and I agree wholeheartedly that areas devoid of life are worth preserving. Some of my favorite places on earth are alpine mountain summits, and there's certainly not much alive up there. But, honestly, the environmental impact of the proposed activities is just incredibly small. You must have some incredibly stringent standards for determining if something "violates" an area.