Top 500 Supercomputers Ranked

Shadow Wrought writes "The Register is reporting on (alternate ZDNet article) the latest list of the top 500 supercomputers in the world. Top of the list is the Earth Simulator Center in Yokohama, Japan, with a benchmark performance of 35.86 Tflop/s. HP and IBM claim 159 and 158 of the systems respectively. I wonder how many teraflops Deep Thought could have done?"

Comments

Deepthought

[Duncan3]

Since it was a chess computer, I'd say it could probably do about 0.000 FLOPS.

If things were different in the 80's...

[lscotte]

This list would probably have been dominated by elxsi, Connection Machines, and Crays, if things had been different.

I never saw a live CM or Cray, but I did play on an elxsi, and it was a pretty hot system for it's time.

You have never been to japan

[gnovos]

For example, you won't see some Japanese Guy driving around in an Escalade with his girlfriend, "just cause".

Ironically, it's quite a common sight to see Japanese kids driving around in huge American monsters -- with the steering wheel on the wrong side for Japan even! -- "just cause" they think it looks cool.

Re:You have never been to japan

[pao93]

I agree. I hate it when people speculate about the Japanese young people. Yeah, maybe you see some dorky salaryman driving around in Hybrid. But if it's a young cool guy who likes cars, they'll be driving some monster classic from the US (slightly rare but not uncommon) or a heavily modded, suped up domestic car. Lots of Van type things with wings (?!). Lots of big station wagony cars. Lots of flashing lights. To insinuate that the dudes in these cars have to be Yakuza is fucking pointless. I have a good friend (Japanese) who loves classic american cars and has a huge old 69 Chevy.

Re:American Dominance in Supercomputers

[netsharc]

The 21st century is not PaxAsia. It is PaxAmericana. The hordes of immigrants flooding into this country to get the hell out of Asia should have been a big hint.

Gee whiz, what a grand conclusion from a simplistic argument. I wonder why people still find USA attractive, I come from the world's largest Muslim country (not Muslim myself), and I wouldn't want to go to a country where I would immediately be seen as a suspect terrorist and where your Agent Smiths can arrest me for just because I'm foreign. I don't even think they'll let me in if I wanted to visit the goddamn country. The same fate is suffered by a lot of students and scientists who study in the US and had helped the US to be the great technological nation that it currently is. Here's the article which mentions that fact, but I don't think you'd be able to read it - it's in German and you don't seem like someone who'd learn a foreign language.

That seems to be only a small, unimportant part of the US's big problem. Afghanistan is going to hell again, Iraq is still in chaos, and billions of your taxpayer money is being used in trying to stop that chaos, chaos which Bush created in hopes of rewards that he isn't going to share with you, because they're all going into his and his friend's and family's pockets.

Yea, vote Republican!

Hard to Tell Operating Systems...

[billstewart]

The Second Fastest Supercomputer, which isn't on the list, is SETI@HOME, with ~27TFLOPS, and most of the machines it's running on are Windows. Most of the machines don't say what OS they're running, but most of them are running some kind of hacked-up operating system to coordinate communications and tasks. The problem with Windows in this environment isn't whether it sucks, but that it's not open to hacking and customization to anywhere near the extent that most Unix OS relatives are (except apparently SCO, which has trouble running networks on /dev/lawsuit.)

Also, Windows wants to run on something that looks at least *vaguely* like a PC. Some of these supercomputers look like PCs with odd network peripherals underneath them, some look like clusters of multi-processor shared-memory PCs (sometimes with too many processors for Windows) with a communications layer between clusters. Some of these work ok for Windows (SETI, for instance), while others are too different.

Also, the communications patterns between nodes and between common applications programs are highly critical here. The tighter the coupling, and the finer-grained the parallelism, the harder it is to fit into whatever framework the operating system provides. Loosely-coupled systems can work just fine on Microsoft; very-tightly-coupled systems need more hacking. And a large part of the Windows plaform is really focused on desktop graphics applications, which simply aren't relevant for supercomputers. (There are people doing clusters with game consoles, such as the Sony Playstation, but that's because they want to use the fast parallel CPU in the graphics engine, not the boring CPU.)

Re:I nominate Pink

[DeathPenguin]

In many clusters today, each node relies on a hard disk to hold a distribution. This is a real big problem because hard disks generate a lot of heat. This is especially true with 10-15K RPM SCSI drives.

To add to that, most cluster node manufacturers that I know of design their 1U cases so that the hard disks are mounted on the front of the case for hot-swapping (Which in itself should suggest that people have problems with drives failing). The fans lie somewhere behind the hard disks. For example, this is a picture of an Appro 1122H dual Opteron server. Some cool air (But not a lot) is drawn from the vents on the front-right of the case, but a lot of hot air gets sucked away from the hard disks and blown over the CPUs. A better solution is just to take out the hot swap bays, make sure there's no SCSI backplane in the way, and let the cool air be drawn in from all across the front.

There are some manufacturers who will put a cardboard heat tunnel around the intake vent, lead it over some blowers, and eventually have to run over the CPUs. That's good design for keeping the CPUs cool, but it basically leaves the hard disks and other components such as north bridge chips and memory modules to fend for themselves when it comes to fighting thermal death. Often times they fail.

I have yet to see a hard disk that was smart enough to spin down to a lower speed when it reached a certain temperature. Even if they were that smart, who wants an HDD performing poorly when you've already spent ungodly amounts of money for bleeding edge 15K ultra-320 drives? As far as I'm concerned, they're practically built to destroy themselves. This is why diskless cluster nodes are so appealing to me.

Sorry for my little rant there. I hope it helped to clarify what I meant when I was emphasizing the advantages of having fewer moving parts.

Re:SETI@HOME is actually #2, with 27TFLOPS

[Noren]

Sorry to interrupt a rant with actual facts, but...

LANL and LLNL have actually done research on cancer, unlike SETI@HOME which has done no work at all on cancer.

The University of California is currently a 'Key Sponsor' of SETI@HOME and its Berkeley campus is home to the SETI researchers who set up and use SETI@HOME. The University of California also currently operates both LANL and LLNL.

I'm not familiar with Evil Linux, is it anything like Red Hat?