Pushing P4 to 5.25GHz with Liquid Nitrogen

SkywalkerOS8 writes "The folks at Tom's Hardware have an article up about their attempt to overclock a Pentium 4 over 5 GHz using liquid nitrogen as cooling. A DivX video is available along with pictures of the custom copper cooling head they made."

Eschew Obfuscation

[anaphora]

Liquid Nitrogen? Compressors? Huge heatsinks? Wouldn't it have been cheaper just to beowulf cluster a few systems together?

Then again, I guess that wouldn't be as 1337, and we wouldn't have this slashdot story over it.

Re:cost?

[Liquidrage]

The point isn't to have *this* much processing power in a home computer.
It's more like climbing a mountain. You do it because you can and you enjoy doing it.

Re:5+ GHz

[niko9]

The question is, how fast did it play solitaire once Windoze was booted?

The real question is; how fast did Windows crash before you even loaded solitaire?

Hardware damage!

[starsong]

Heh, this looks like a lot of fun, but that board's not going to last long. Look at the picture on the first page. See the capacitors next to the socket with little ice crystals growing on them? Those are electrolytic caps; they use a liquid electrolyte which doesn't take kindly to being frozen solid. I'm amazed they didn't split open. Colder isn't always better; some components will simply fail at liquid-N2 temperatures. At least they took steps to deal with condensation.

Looks like something any ordinary plumber could do

[glenebob]

With all that fancy talk about tolerances and only one company in the world that could make the aparatus, you'd think it would be bit fancier... Nope, just a coper plate with a copper tube sticking up off of it that you fill with nitrogen, and it cools via evaperation. I could build it with some 2-inch copper pipe, a torch, and some soldier... 5 GHz is cool and all, but come on, is there really the need to make it sound so difficult?

Shorts

[king-manic]

with that amount of ice crystals, I'm surprised it didn't short? I know it's distilled water but you figure minerals from the metallic elements on the silicon would contaiminate it and cause shorts?

Overclocking is stupid--No, make that "insane"

[shanen]

This overclocking stuff is REALLY stupid to the point of insanity. My conclusion is that it's a weird fantasy about the lone DIY (do-it-yourself) tinkerer.

First, consider the economic side. For all of the special efforts and costs needed to cool down, test, and monitor an overclocked CPU, you could just buy a couple more for the same speedup effect. No special anything required. At the same time, there is no real need for all those cycles--we have a glut of cycles now. If it were really cost-effective to overclock and use special cooling systems, then the very few people who actually do need lots and lots of cycles would be using overclocking for their supercomputers--and they don't. They just buy more CPUs and run them the way they were designed.

The design question leads to the second point. Building a modern CPU is not a hobby for amateurs. It is an incredibly complicated device involving the efforts of large teams of very clever people using very fancy design tools. No one person could even know all the details of a modern CPU. Far too many details. They may know some of the higher level features, or know a lot of detail about a tiny section, but no one really understands all of it. However, they are doing the best they can to insure that it will work reliably, and that includes MANY design considerations that are related to the clock speed.

So back to my main conclusion: Overclocking is a fantasy of the DIY tinkerer "beating" the experts. Actually, it's nice when it happens, but overclocking is NOT one of those cases. The overclockers fantacize about some form of "delivering more bang for the buck", but they are competing directly against professionals with the same goal. The pros win, especially in Intel's case where their development costs per CPU are almost negligible. As the joke goes, "The race is not always to the swift, nor the battle to the strong, but that's the way to bet." The overclockers already lost. (By the way, I think this is also an expecially American fantasy, a kind of "independence" thing, and that there are very few non-American overclockers.)

One more technical aspect as a fairly concrete example. Overclocked computers can become unreliable. Many overclockers limit their testing to "Does it boot and seem to run the OS properly?" However, the OS is not using the floating point resources the same way that true numeric applications do. The machine may seem okay as far as the OS is concerned, but actually be producing gibberish results. (There was actually a probable example of this published by seti@home. I'm tempted to diverge into the psychological relationships there...)

Ergo, I've never heard of Intel hiring someone for their expertise in overclocking, and I don't expect to.

Talk about journalistic integrity!

[Hobophile]

I saw this article over lunch today, and when I checked back just now I noticed they'd removed the page of benchmarks. One of the interesting results shown was that the Athlon 64 FX-51 managed to beat this overclocked behemoth in a couple tests.

Only one or two, mind you, but it still boggles the mind that this Pentium running 2.5x faster than the Athlon chip didn't utterly dominate all comers.

Given the history of THG and their decidedly negative (some might say Intel-funded) view of the Athlon 64 chips, it's not particularly surprising they'd choose to pull that page, but it does cast further doubt on the continued relevance of what was once a high-quality tech reporting site.

The few posts questioning this on the THG forums seem to have disappeared in the time it took me to write this. Strange...

Benchmarks?

[Aaron England]

Where are the usual pretty Tom's Hardware graphs? What the hell is a 5.25 GHz processor good for, if we can't awe over benchmarks like "time it takes to process a SETI unit" or its score in Sandra 2004?

Color me unimpressed

[xtal]

This is basically dumping liquid nitrogen onto processors outside and clocking them up. There's not much of an achivement there. You can soak LEDs in liquid nitrogen and make them do all sorts of interesting tricks too. Whoop.

Why not wait until someone comes up with a indoor version, properly vented and pumped, with a compressor cycle that you can actually use on a long-term basis? That would be an achivement I'd like to see. Of course, it's orders of magnitude more difficult and dangerous, too.

Relevance & Powermac

[reignbow]

A lot of people have asked about the relevance of this: Basically, there is none. But that's all right. It's a nice story to entertain their readers, and I'm willing to bet it was a lot of fun for them, too. Not everything needs to have a point, you know.

That said, there's one thing that would still interest me: Now that we've seen them overclock that wimpy Pentium 4 (I hate that architecture! How can anyone build a 20-step pipeline?), let's have some real techno-porn: Liquid Nitrogen-cooled 2x2.0GHz G5 Powermac! That would be quite a sight to behold. Especially with that nifty 1Ghz FSB.

Re:Looks like something any ordinary plumber could

[Slack3r78]

You must never have read Tom's Hardware before. *EVERYTHING* they do is played up to man-on-the-moon levels, regardless of how trivial. You can either get used to it, or do like me and simply avoid Tom's as much as possible. :)

Re:But is it really a good idea?

[Hoser McMoose]

Of course it's not a good idea! Geez, did you look at the pictures? The voltage regulators were covered in ice after a half hour or so. PCs are NOT designed for these sorts of systems, it's unlikely that it would run for a day, let alone any significant amount of time.

Besides which the cost to buy all that equipment, get a customized motherboard, have someone mill the heatsink and attachments, etc. etc. would surely make this this a ridiculously expensive system.

However, when it comes right down to it, it sure does get the website a lot of hits, and that was the goal all along.