Slashdot Mirror

Believe it or not, the US government has had this concern for a while now and is trying to do something about it:
https://cmi.ameslab.gov/
https://cmi.ameslab.gov/materi...

Believe it or not, the US government has had this concern for a while now and is trying to do something about it:
https://cmi.ameslab.gov/
https://cmi.ameslab.gov/materi...

Our work to date shows that it is not an insurmountable task to extract very high efficiency from a massively-parallel ensemble, for the reasons presented here. We feel that it is important for the computing research community to overcome the "mental block" against massive parallelism imposed by a misuse of Amdahl's speedup formula; speedup should be measured by scaling the problem to the number of processors, not fixing problem size. We expect to extend our success to a broader range of applications and even larger values for N.

...it seemed to me that Amdahl's law was still alive and kicking.

CUDA is already doing great things in molecular dynamics, which bears some similarity to FEA:

HOOMD Benchmarks

A single 8800 GTX reaching 75% of the performance of a 32 node cluster isn't bad. I imagine the GTX 280 would easily beat the cluster.

Not to mention that that extra dollar cost may push a Mac over the line into capital equipment where overhead isn't charged. My first laptop for work ran into that issue: I picked a nice model for ~$1800, and was told I wasn't spending enough. As it turned out, the $2600 ibm was cheaper, because the lower cost one came with 50% overhead attached.

You could have also ssh'd into a real cluster, or built a Mac cluster for a price similar to an Opteron system, and just quietly integrated it with your desktops (how my lab runs now). They just work, and they just work smoothly. It's also nice that tools like VMD come in native form, and run very smoothly. It's nice taking VMD, GAMESS, and Amber on the road with you, running in native mode the same way they run on the big clusters back home, just in case. (yes, I know about the windows ports or cygwin, but they always feel somewhat clunky)

Finally, sometimes the commercial software really does just work better, and fighting a journal over file formats is an exercise in futility.

John Gustafson has a different take. In a nutshell, he says two things. First, when many processors are available, what you really want to do is solve a bigger problem in the available wall clock time. Second, when you grow the kinds of problems we are talking about, the parallel part grows faster than the serial part. For the right kinds of problems, these two things make Amdahl's law not applicable.

At one point, the Democrats were the party that was solidly behind states' rights and the Republicans were all about centralized power.

Actually at first there was the Democratic-Republican party which Thomas Jefferson was a member of. Back then the other major party was the Federalist Party, then some of it's members became members of theWhig Party and others joined the Democratic Party.

I tend to think the authors of the article are refering to the problems of a "useable form" for the structures and easy access of many of these databases. The first problem is mearly a problem of converting between the various structural file formats out there, something a good programmer (or grad student) can solve is a few weeks or less. The second is a bureaucrat issue and not a scientific one.

The ENIAC was the first real computer, you know.

Of course, I'm kidding. Everyone should know this. :-)

Honestly, I don't understand how a conservative government can increase the size of government this much, and ask for internet regulations, I mean it does not follow the philosophy at all. Am I the only libertarian here?

Actually you like many others have switched what liberal means. Going back to the revolutions in the USA and France to be liberal meant to be for small government and Liberty, ie "liberal". A prime example of this in the USA was Thomas Jefferson. As a Liberal TJ wanted small government. Of course today most people don't use the word to mean the same thing, instead Libertarians come the closest to what it means to be a Classical Liberal.

It's just a darn shame they stole all that technology, eh? Although Eckert disputes it at the end of the interview, the court found that: "...John Vincent Atanasoff and Clifford Berry had constructed the first electronic digital computer at Iowa State College in the 1939 - 1942 period. He had also ruled that John Mauchly and J. Presper Eckert, who had for more than twenty-five years been feted, trumpeted, and honored as the co-inventors of the first electronic digital computer, were not entitled to the patent upon which that honor was based. Furthermore, Judge Larson had ruled that Mauchly had pirated Atanasoff's ideas, and for more than thirty years had palmed those ideas off on the world as the product of his own genius." Full Q&A can be found here: http://www.scl.ameslab.gov/ABC/Trial.html Court documents can be found here: http://www.cs.iastate.edu/jva/court-papers/index.s html

Thats not a fact by a far cry. Never heard of superlinear speedup?
http://www.scl.ameslab.gov/Publications/Gus/Superl inear/Superlinear.html

If you split your problem up in 7 subproblems, and they now all of a sudden fit in cache - you are gonna see a tremendous speedup versus a 7 times faster processor where the problem does not fit in cache.

Close, but not quite.

Electronic digital computers were first invented to calculate artillery tables for naval guns, which told the gunner what angle to fire the gun so the shell would reach a certain distance. The tables were originally calculated with logarithm tables and slide rules, by hordes of "calculators" (human calculators, that is). Then John Atanasoff invented a method of calculating them electronically with the Atanasoff-Berry Computer. Unfortunately, Atanasoff abandoned his project when WWII started and went to work for the Naval Ordinance Laboratory.

Sorry if this sounds insulting, but your attitude seems really narrow-minded and short sighted. The whole reason the computer is such an incredibly useful tool is that it is so flexible and extendable. YOU might manage to get everything you need out of the software included in your distro, but do you really expect the big distros to anticipate every single need of every single user? A lot of people who are not computer experts have specific application needs that the vast majority of users don't share. Should a good distro include a version of GAMESS just because I want to do a theoretical chemistry calculation? Or maybe the people who make distros should assume (correctly) that if I am one of the .0001% of computer users who would want to use that program, I should just go download it myself?

"This may sound elitist of me, but if you can't figure out how to do it now, you probably aren't capable of making that sort of decision."

Yes, you sound incredibly elitist, as if it is impossible to be smart and NOT a computer expert. There is a big difference between knowing enough about one's Linux distro to install a program and having enough common sense to find programs on the internet with minimal risk of installing malware. If I google search for software that simulates microwave spectra of asymmetric top molecules (and by the way there are quite a few) what are the odds I'm going to find spyware masking itself as what I'm looking for?

I have several harddrives that top out around 14-20Megabytes per second, which turns into roughly the speed rating you are talking about.

I doubt your running out of PCI bandwidth.

It could be the latency, or that you have a poorly tuned network stack. I know that using NFS, getting 12-15Mbit/sec was considered pretty good given the inherient latency of the protocol.

I had similar problems no matter what protocol I was using FTP, HTTP, or scp. What I found was that, I needed to use a network speed tool: NPtcp, which is part of the netpipe tool set.

http://www.scl.ameslab.gov/Projects/old/ClusterCoo kbook/nprun.html

The other thing is to figure out if your cards support Jumbo frames. If they do, it can be a boon to go change your MTU, and modify specific parameters in your TCP stack, and in your application to change the socket options used (specifically, to use a packet size larger then 8k). I'm not sure how to do this under windows, but I've found it readily available under Linux on google searches.

More information is more useful. Knowing what chipset it is based off of, which drivers you are using, what OS, would be mighty helpful to helping you solve your problem.

Kirby

The adulteration of orange juice: In 1990 an employee of Sun Up Foods, an orange juice processor, told the Food and Drug Administration (FDA) that the company had set up secret rooms in its facilities to hold tanks of liquid beet sugar, a common adulterant of orange juice.

Why would anyone use rabbit shit in coffee? Chicory is the standard item used to dilute or adulterate coffee.

We can contemplate the Ford Pinto without much further comment.

How about a little Alfatoxin in your peanut butter!

Need a chest X-Ray or a really fast sun tan?

Or maybe you want to take a trip in Sir Geoffrey de Havilland's Comet?

And if you still think that you can consume or use products or services without paying close attention, I have a bridge to sell you.

According to the Smithsonian, U.S. Patent Office, U.S. Courts and many others, the honor goes to John Victor Atanasoff

According to the Smithsonian, U.S. Patent Office, U.S. Courts and many others, the honor goes to John Victor Atanasoff

It appears this "Computer History" attempt overlooks John Vincent Atanasoff, credited by most reliable sources (Smithsonian, etc.) as developer of the first electronic digital computer" years before the ENIAC. In fact, the ENIAC was derived from Atanosoffs's ABC Computer at Iowa State after an ENIAC developer visited Atanasoff (stayed several days in Atanasoff's home), and "stole" his ideas and proposed a larger verssion as the ENIAC to the army. Atanosoff's ABC computer was the first to solve Schroedinger's equation represented by the solution of a 39x39 system of matrix equations. However, time caught up with the ENIAC visitor, and the notebook he kept when he visited Atanasoff was his undoing when the U.S. Court in Minneapolis overturned previous patent rulings for computer developments and ruled they were all derived from Atanasoff's ABC computer. Hopefully, this attempt at a computer museum will soon be updated to accurately reflect the original development of the electronic computer by Atanasoff at Iowa State in 1942.

If it was "built at the same time during the second world war," then they weren't built any sooner than 1939. If they weren't built any sooner than 1939, then one John Vincent Atanasoff had been working on a computer for two years. According to this page, the Colussus was built in 1943-1945, which means it was started a year after the Atanasoff-Berry Computer was finished.

And before anyone brings up the ENIAC, Honeywell Inc. vs. Sperry Rand Corp. overturned the patent due to prior art from the ABC.

Tragically, both the Colussus and Atanasoff-Berry computer have been destroyed. The Colussus because it was top secret, and the Atanasoff-Berry because Atanasoff was convinced the standard door size was 36", when they're actually 33". It was dismantled to free up office space.

*DING DING DING DING*

EXACTLY.

We buy about 2 new macs each time apple releases a new machine, and put it in our 32 machine Mac cluster http://www.scl.ameslab.gov/Projects/PPC_cluster/, and someone gets an old cluster machine. In virginia Tech's case, they can buy 30 or 40 ;)

FYI, our cluster runs linux, and yes, I just got infiniband working on linux on a G5.

Benchmarks to follow ;)

Troy Benjegerdes (hozer-nospam@hozed.org)

The first electronic digital computer. Well, the rebuilt replica anyway if you happen to come through my home state. =)

No one had contested that Eckert and Mauchly had designed the first electronic computer, but instead had hooked onto details in the patent file.

Except for Honeywell who filed the lawsuit against Sperry on the basis that Eckert and Mauchly had derived their work from that done by Professor Atanasoff and his student Clifford Berry at Iowa State University.

ENIAC was not the first electronic computer, it was rather the ABC. That's what the lawsuit established.

More information here

The unfortunate thing is that the news broke the same day as the "Saturday Night Massacre", where Nixon told Robert Bork(yes the same Bork who Reagan later tried to appoint to the SCOTUS) his new Attorney General(after the other two guys had quit) and then had him fire the independent counsel investigating Watergate. So it didn't receive a whole lot of coverage.

I'm unclear on exactly what you think the similarities are here. I'm familiar with the Atanasoff story as I'm an alum of Iowa State.

Being from the east coast, I'm sick of people with the attitude that the only thing that comes out of iowa is corn. Iowa has one of the highest SAT averages in the nation, and some of the best universities, such as Iowa State and University of Iowa. Beleive it or not, there is a world outside of California, New York, and Philadelphia.

Global Cooling on the otherhand produce rival products to Medis El based on the Free piston Stirling Engine.

Despite being some impressive technology, Free Piston Stirling Engines haven't really been taken up to well. Its a shame because they do seem to be much more efficient.

If you are really interested then you might want to check this out At Ames Lab. Gschneidner's work on the giant magnetocaloric effect is REALLY impressive. Its all about the exchange of entropy between magnetic and kinetic forms. Damn cool.

In practice, gig ether (even over fiber) achieves only 400 to 600 megabits/sec. SCL tested some of the first gigE cards and got terrible performance. Their cluster cookbook is rather dated, but compares some of the popular interconnects (Myrinet, Fast Ethernet, and FDDI).

Few cards that I have seen support jumbo frames; even then, I've only seen cards that go up to 9000-octet frames. Face it: Ethernet doesn't scale well past 100mbits/sec.

In practice, gig ether (even over fiber) achieves only 400 to 600 megabits/sec. SCL tested some of the first gigE cards and got terrible performance. Their cluster cookbook is rather dated, but compares some of the popular interconnects (Myrinet, Fast Ethernet, and FDDI).

Few cards that I have seen support jumbo frames; even then, I've only seen cards that go up to 9000-octet frames. Face it: Ethernet doesn't scale well past 100mbits/sec.

...there are many of these clusters around the U.S. This one was built in 1999 at Iowa State University/DOE. It consists of 48 G4 processors, runs Black Lab Linux, and was never given any attention by the press. This new cluster is just a more commercialized version of the same thing.

1) yes, the digital computer.

2) Iowa produces Corn. Corn = Ethanol. Ethanol = lessened dependance on oil. This makes oil prices drop, and it lessens our dependance on foreign oil. Not to mention that it burns cleaner and more efficent than regular oil.

3) You want something that we've done? How about the first six-sided 'cave' virtual reality system in the united states?

I resemble that remark not to mention some other guys I know (search for "javasoft" for some humorous anecdotes).

Our heroic New Yorker author, with a single leap, bounded right over duos like the Wright Brothers and Atanasoff and Berry as well as small hunting packs like Id Software and small tribal clans like the Seymour Cray 34.

Yet again, someone [this column] lists the "first modern computer" as ENIAC--yet England and Iowa State University both built predecessors:

Colossus
Atanasoff-Berry Computer

As established in court the computer was invented before WW II at Iowa State College of Agriculture and Mechanical Arts in Ames, Iowa

I believe they have finally gotten the credit they deserve for their invention even though most people still think it was Eckert and Mauchly.

This is correct--the ABC was acknowledged as the first digital computer over the ENIAC. Significantly, this acnoledgement came from a federal court in 1972 in a case about... wait for it... patents on certain aspects of electronic computing. The so-called "ENIAC patents" were bought from Eckert and Mauchly by Sperry Rand. Other computer makers, specifically Honeywell in this case, were willing to go to any expense to overturn the validity of the patents. As it happened, the ruling was favorable, with the judge concluding that the ENIAC patents were based directly on prior art by Atanasoff et al.

Sperry Rand's computer division has grown over the years to form... Unisys, the GIF patent people.

More info about the case here

--Tom

There is some information on what channel bonding is here. I have not yet set up my system to handle this, but I intend to.

Everyone knows about caching on a web server, so I won't bore you with that. Since I am using a java application server (Weblogic) to serve my site, I can configure the container to load and serve a connection pool to my applications. This means that the connections are never closed, and so I only have to spend the overhead of opening the connection one time. I am serving from the same machine(s) so my reuse of connections is very high.

It's really quite fast. I'll probably write a howto on this for others to follow as well.

"The fact that a manual is shorter doesn't mean that it is a better or easier to install program."

While this is true, it's not even to the point. They didn't compare manuals. They took a book written on building a Linux cluster, and compared it to what is basically a step by step outline for for plugging together a G4 cluster. There are similar outlines out there for Linux clusters, too:

But the point is, this article was written by pro-Mac people, so obviously they're going to take a pro-Mac stance. I mean, if these G4 clusters get to be useful, someone is going to write a 230 page book on how to build one of them. Right now, all the documentation that may be out there could be contained in this one page outline. The books come later, if the technology becomes accepted.

• Here is a not-too-technical report on buckyballs, their properties, etc.

• Pretty pictures!!!

• According to
• this article, buckyballs hold the record for highest-temperature superconductor.

• A report (fairly technical) on research into building buckballs...

• And
• here's a report on single buckyball transistors.

Based on the description of the article, I looked up some things. What can I say? Somebody modded me down, so I'm at 49, and I'm incomplete without that karma point.

Amdahl's law

Amdahl's law

On chip multiprocessing

Simultaneous multithreading

There are some nice G3 and G4 clusters out there, they are just not very cost-effective. Here's a howto on building a G4 cluster from a national lab, and there are some prebuilt systems like those running Black Lab Linux they were showing at MacWorld New York last summer.

Also in contention for the distinction of being the first digital computer is the Atanasoff-Berry Computer built at Iowa State University between 1939 and 1942 by John Vincent Atanasoff and Clifford Berry. Supposedly, Atanasoff conceived the plan of the machine drinking bourbon in a roadhouse bar somewhere in Illinois in 1937.

In 1973, after a lengthy court trial, a federal judge declared the Eckert-Mauchly (Eniac) patent invalid and named Atanasoff the inventor of the first digital computer. It should also be noted that it was the first digital computer to use dynamic RAM. Lots of good information on the ABC and many more links can be found here and also here. Photos and diagrams can be found here.

If one closely examines this period of history, they find that it is a time that is just chock full with all kinds of convergences between mathematics, physical science, engineering and materials technologies that make the digital computer almost inevitable. After all, this is a device that had been conceived of, at least in part, as early as the Victorian age and the birth of the industrial revolution. Really, it was just a matter of time before somebody produced a working model, and as so often happens many people took different paths to the same end.

Also in contention for the distinction of being the first digital computer is the Atanasoff-Berry Computer built at Iowa State University between 1939 and 1942 by John Vincent Atanasoff and Clifford Berry. Supposedly, Atanasoff conceived the plan of the machine drinking bourbon in a roadhouse bar somewhere in Illinois in 1937.

In 1973, after a lengthy court trial, a federal judge declared the Eckert-Mauchly (Eniac) patent invalid and named Atanasoff the inventor of the first digital computer. It should also be noted that it was the first digital computer to use dynamic RAM. Lots of good information on the ABC and many more links can be found here and also here. Photos and diagrams can be found here.

If one closely examines this period of history, they find that it is a time that is just chock full with all kinds of convergences between mathematics, physical science, engineering and materials technologies that make the digital computer almost inevitable. After all, this is a device that had been conceived of, at least in part, as early as the Victorian age and the birth of the industrial revolution. Really, it was just a matter of time before somebody produced a working model, and as so often happens many people took different paths to the same end.

..also known as ABC, constructed in 1942 at Iowa State University. It's such a shame the ENIAC overshadowed this wonderful project. The machine had a precursor of todays Dynamic RAM (used charged capacitances to store digital information).

Actually, I've seen the ABC pictures and replica, and it DOES use tubes. See this site for some cool pictures of the replica they built, including actual vacuum tubes manufactured in the 40s!

http://www.scl.ameslab.gov/HINT/

This benchmark measures performance with multiple dimensions (ie. performance with respect to the size of the problem, the number of processors available, and the size of the cache) It's numerical intensive so it gives a good idea of what scientific programming performance would be like, not say graphics performance...

There is also no mention of the first electronic digital computer, the Atanasoff-Berry Computer (ABC). After all, the Courts have decided that the ENIAC was a derivative of the ABC.

There is also no mention of the first electronic digital computer, the Atanasoff-Berry Computer (ABC). After all, the Courts have decided that the ENIAC was a derivative of the ABC.

Lovely, I missed a key in the href

www.scl.ameslab.gov/ABC

If you are interested in the history of the first electronic digital computer you really should check out the site

Being an alumnus of Iowa State University, it always irks me when the ABC goes completely unmentioned in the history of computing.

Acutally, I think the first computer was the Atanasoff Berry Computer (Ames Lab). A court trial voided the patent awarded to ENIAC. There's a lot at the site, but it's a must-read if you've ever belived that ENIAC was in any way the first computer. I've seen the ABC replica (at the Iowa State Fair, laugh, laugh) and it's not a very big machine, and it can do not a whole lot, but it's an electronic digital computer.

I've always thought that the true "First Digital Computer" should go to the ABC. It is true that because it was built a physics prof and a grad student for research, stability wasn't the best but it was more than enough to prove digital computing theories they put forth.

Here is the Ames Lab's web page on their project to construct a replica the ABC machine. Each drum could hold up to 30, 50 bit numbers...that is only 3,000 bits(.4K). That is computing power! :-)