Hey, I've got a question for you clustering experts. How hard would it be for Virginia tech to upgrade Big Mac in the future? Is it possible to just add another row of G5's to their network ifrastructure and/or swap out the dual 2 giggers when something signifcantly better comes along?
There are many wonderful pictures of the 'Terascale Project'
I quite like the first "G5 Ordering" pic. I guess after they used the university credit card to order 1100 G5s they figured a couple songs at 99¢ wouldn't hurt
There is likely to be some final tweaking of the installation, but it will surely rank at least fourth in the supercomputing results. The second place machine is the ASCI Q at the Los Alamos National Laboratory, a weapons lab operated by the U. S. Department of Energy. Clustering -- the linking together of multiple processors -- has come to the fore in recent years as a supercomputing technology. Other processors successfully used for supercomputing clustering include the Alpha processor, recently discontinued by the Hewlett-Packard Co., Intel's Xeon, and AMD's Opteron. HP's Itanium processors are likely to take a place high up on the supercomputer speed list when an Itanium-based installation enters the supercomputer sweepstakes.
That article was a copy & paste job from about 5 other articles. Sound familiar to anyone?
I know it's tough to be worried about accuracy when you're pasting a dozen articles and press releases a day into the forums without adding a word of your own, but you could at least try dude.
The university's computer science department linked 1100 Apple Computer G5s together several weeks ago to create the installation at a price of less than one-twentieth the cost of comparable supercomputers.
wrong it was more like half as much.
Quote:
The installation has been ranked at least fourth in the constellation of supercomputer compiled by Jack Dongarra of the University of Tennessee.
wrong it's at least third, and although it won't pass #2, #4 is no where near it.
Quote:
The machine can handle 17 trillion operations a second, still short of the Earth Simulator in Japan, which can handle 35 trillion operations a second.
wrong it's right around 10 Tflops, not 17. if it were 17 it would be in 2nd place. these people are idiots.
Quote:
The big difference is price -- $5.2 million for the Virginia Tech installation and more than $250 million for the Japanese installation.
misleading the 5.2 million was for machines and wires, but not the physical housing. do the earth simulator's figures just account for the cost of machines, or the facilities as well?
80% of the first two paragraphs were wrong. i'd do the rest but this post would be two pages long.
if you can't bother yourself to actually post content with these simple article links, at least do everyone the favor of not posting one line links to articles that suck.
Just heard a story on NPR about our favorite supercomputer. Nothing new was said that hasn't been stated before, but i just thought i bring it up because of the attention this story is getting. They made apoint about the cost and rapid assembly. they mentioned that this may spur further installations of such clusters by companies etc that previously felt they couldn't afford such computng power in house. I reaaly consider this absolutely great advertising for Apple. Even if experts want to debate the claim to "the world's fastest personal computer" the average joe who hears or reads about this supercomputer that is so fast and cheap will obviously conclude that Apple must be making one great machine.
Since some of you think the article from Tech Web sucks I guess I must have been wrong to post a link to it. On the other hand, perhaps those of you with enormous know-it-all egos should chill out and try posting articles yourself instead of pontificating on things as if you were always experts on the subject at hand. Also, where does it say I or any other poster has to comment on an article when he or she posts it?
Since some of you think the article from Tech Web sucks I guess I must have been wrong to post a link to it.
the annoying thing about this article is two-fold. first, it's horribly inaccurate. the crap in that article was so far off base as to laughable. if this were one of the few articles on the Big Mac cluster that might be OK, but this is the 20th. by this point in time, if you're just going to regurgitate facts, they should at least be accurate.
then to top it all off, you of all people posted it. this is a thread you started! how in the world can you think this is a good article if you've been following the VTech cluster at all!
i'm sorry but i can't believe you actually read through that article before posting a link to it. that doesn't make any sense. to anyone following Big Mac that article was just a bunch of gibberish.
i probably wouldn't have cared if it was some random person who just popped into the thread w/o having read any of this stuff, but for the original poster to throw this crap in here is insulting.
Quote:
Originally posted by MacsRGood4U:
Also, where does it say I or any other poster has to comment on an article when he or she posts it?
from the PG (albeit in the AO specific area. probably something that should be included across the board.
Quote:
- Threads with no original content will not be allowed. Example: posting a link or quoting and article while contributing little to nothing of your own.
note: going through this thread, up until that last article post, 6 of your 9 posts in this thread were just links to articles or excerpts of articles with no original content.
Kickaha and Amorph couldn't moderate themselves out of a paper bag. Abdicate responsibility and succumb to idiocy. Two years of letting a member make personal attacks against others, then stepping aside when someone won't put up with it. Not only that but go ahead and shut down my posting priviledges but not the one making the attacks. Not even the common decency to abide by their warning (afer three days of absorbing personal attacks with no mods in sight), just shut my posting down and then say it might happen later if a certian line is crossed. Bullshit flag is flying, I won't abide by lying and coddling of liars who go off-site, create accounts differing in a single letter from my handle with the express purpose to decieve and then claim here that I did it. Everyone be warned, kim kap sol is a lying, deceitful poster.
Now I guess they should have banned me rather than just shut off posting priviledges, because kickaha and Amorph definitely aren't going to like being called to task when they thought they had it all ignored *cough* *cough* I mean under control. Just a couple o' tools.
Don't worry, as soon as my work resetting my posts is done I'll disappear forever.
What did ASCII White cost?? Or Ascii Q which is more expensive still. Oh, I guess 1/20th for slightlylower/comparable performance isn't so far off.
The 17 Tflops are the theoretical max of the G5 cluster in a perfect world, stated so in many of those articles you have already read. Gee, let the PR departments have their day in the sun! We know they will eventually (probably/hopefully) top out around 12-13 Tflops, but maybe not in the ranking tests or in the near future.
the cost comparison comes from Varadarajan himself. he just priced them out a few months ago. it was 2x, not 20x for a comparable product that wasn't a G5.
they're comparing a theoretical 17Tflops max for the Big Mac against the real world performance of the Earth Simulator. also stupid.
Comments
There are many wonderful pictures of the 'Terascale Project'
Originally posted by bauman
One of my friends just sent me this: http://don.cc.vt.edu/
There are many wonderful pictures of the 'Terascale Project'
Wow, did he send those to you through a rift in space-time?
Originally posted by bauman
One of my friends just sent me this: http://don.cc.vt.edu/
There are many wonderful pictures of the 'Terascale Project'
I quite like the first "G5 Ordering" pic. I guess after they used the university credit card to order 1100 G5s they figured a couple songs at 99¢ wouldn't hurt
Originally posted by Eugene
Wow, did he send those to you through a rift in space-time?
Is it that old? I've never seen them linked to from here... I might have missed it though.
Originally posted by bauman
Is it that old? I've never seen them linked to from here... I might have missed it though.
I've never seen them either, thanks for the link
There is likely to be some final tweaking of the installation, but it will surely rank at least fourth in the supercomputing results. The second place machine is the ASCI Q at the Los Alamos National Laboratory, a weapons lab operated by the U. S. Department of Energy. Clustering -- the linking together of multiple processors -- has come to the fore in recent years as a supercomputing technology. Other processors successfully used for supercomputing clustering include the Alpha processor, recently discontinued by the Hewlett-Packard Co., Intel's Xeon, and AMD's Opteron. HP's Itanium processors are likely to take a place high up on the supercomputer speed list when an Itanium-based installation enters the supercomputer sweepstakes.
Full articlehere.
that article sucked
BIG Time....
I know it's tough to be worried about accuracy when you're pasting a dozen articles and press releases a day into the forums without adding a word of your own, but you could at least try dude.
The university's computer science department linked 1100 Apple Computer G5s together several weeks ago to create the installation at a price of less than one-twentieth the cost of comparable supercomputers.
wrong it was more like half as much.
The installation has been ranked at least fourth in the constellation of supercomputer compiled by Jack Dongarra of the University of Tennessee.
wrong it's at least third, and although it won't pass #2, #4 is no where near it.
The machine can handle 17 trillion operations a second, still short of the Earth Simulator in Japan, which can handle 35 trillion operations a second.
wrong it's right around 10 Tflops, not 17. if it were 17 it would be in 2nd place. these people are idiots.
The big difference is price -- $5.2 million for the Virginia Tech installation and more than $250 million for the Japanese installation.
misleading the 5.2 million was for machines and wires, but not the physical housing. do the earth simulator's figures just account for the cost of machines, or the facilities as well?
80% of the first two paragraphs were wrong. i'd do the rest but this post would be two pages long.
if you can't bother yourself to actually post content with these simple article links, at least do everyone the favor of not posting one line links to articles that suck.
Originally posted by MacsRGood4U:
Since some of you think the article from Tech Web sucks I guess I must have been wrong to post a link to it.
the annoying thing about this article is two-fold. first, it's horribly inaccurate. the crap in that article was so far off base as to laughable. if this were one of the few articles on the Big Mac cluster that might be OK, but this is the 20th. by this point in time, if you're just going to regurgitate facts, they should at least be accurate.
then to top it all off, you of all people posted it. this is a thread you started! how in the world can you think this is a good article if you've been following the VTech cluster at all!
i'm sorry but i can't believe you actually read through that article before posting a link to it. that doesn't make any sense. to anyone following Big Mac that article was just a bunch of gibberish.
i probably wouldn't have cared if it was some random person who just popped into the thread w/o having read any of this stuff, but for the original poster to throw this crap in here is insulting.
Originally posted by MacsRGood4U:
Also, where does it say I or any other poster has to comment on an article when he or she posts it?
from the PG (albeit in the AO specific area. probably something that should be included across the board.
- Threads with no original content will not be allowed. Example: posting a link or quoting and article while contributing little to nothing of your own.
note: going through this thread, up until that last article post, 6 of your 9 posts in this thread were just links to articles or excerpts of articles with no original content.
Now I guess they should have banned me rather than just shut off posting priviledges, because kickaha and Amorph definitely aren't going to like being called to task when they thought they had it all ignored *cough* *cough* I mean under control. Just a couple o' tools.
Don't worry, as soon as my work resetting my posts is done I'll disappear forever.
What did ASCII White cost?? Or Ascii Q which is more expensive still. Oh, I guess 1/20th for slightlylower/comparable performance isn't so far off.
The 17 Tflops are the theoretical max of the G5 cluster in a perfect world, stated so in many of those articles you have already read. Gee, let the PR departments have their day in the sun! We know they will eventually (probably/hopefully) top out around 12-13 Tflops, but maybe not in the ranking tests or in the near future.
the cost comparison comes from Varadarajan himself. he just priced them out a few months ago. it was 2x, not 20x for a comparable product that wasn't a G5.
they're comparing a theoretical 17Tflops max for the Big Mac against the real world performance of the Earth Simulator. also stupid.
Originally posted by alcimedes
they're comparing a theoretical 17Tflops max for the Big Mac against the real world performance of the Earth Simulator. also stupid.
That's all they can do for the time being so why is it stupid?
they know what the Big Mac is pushing. it's right around 10Tflops.
Varadarajan said at 9.4Tflops they might get another 10% out of it. pushing it up to about 10.3 - 10.5Tflops.
they're not going to hit 17Tflops.
Originally posted by alcimedes
they know what the Big Mac is pushing. it's right around 10Tflops.
Varadarajan said at 9.4Tflops they might get another 10% out of it. pushing it up to about 10.3 - 10.5Tflops.
they're not going to hit 17Tflops.
I know the 17 is wrong, but I was just thinking that the 9.something is wrong too. Err on the side of happiness I guess.
Originally posted by bunge
I know the 17 is wrong, but I was just thinking that the 9.something is wrong too. Err on the side of happiness I guess.
Well they could have taken the most up-to-date info from here.