I'm not sure, but a better use of industrial gases might be this [dansdata.com] and probably would provide more perceived results.

(speaking as an ex LOX, LH2 and LN2 piping designer, of course, YMMV)

Dear Poor Reader,

The capacitors have liquid in them, he was mentioning the ice crystals to identify them. Try reading the parent again.

Sincerely,

Me.

I'm not surprised, the P4 is incredibility inefficient always has been. The thing that REALLy gets my goat are these POS Celerons Intel pushs in low end boxes. These cpu's are truely garbage. I'd say the Celeron is the biggest disservice Intel has foisted upon the public. Poor consumers are wasting millions because they are misled into thinking a 2.6GHz celeron is actually faster than a 1.6 Duron.

It's probably something like 3M Fluorinert [3m.com].

Some television engineers I know mentioned that they actually used this technique for mounting electronics in remote locations that needed to stay both temperature-controlled, and absolutely dust-free.

So they took refrigerators and removed all of the shelving from the interior, drilled holes through the side (around the coolant tubes) to bring in power cables, data cables and such (the holes were then filled with expanding foam to make them airtight), and plugged it in.

They said that every time they visited the site, everywhere else was dusty and dirty (and hot). Inside the fridge, it was cool (10c) and dust-free.

Cheap way of making sure that things in remote locations stay working :)

After watching those videos, I can't help but wonder why they were blocking out part of the screen on the CPU-ID program. What could've been so super-top-secret there?

N.

Liquid Nitrogen is cold when it's evaporating. You want it to be cold? Give it a flat surface to evaporate on, and keep pouring on the Nitrogen.

Basically, if you lay a piece of Saran Wrap on your motherboard, then let the LN2 drip on the CPU constantly, you can cool that bastard to -195.798C.

Making a big, tall tower just looks like a stupid Freudian mistake.

Sorry Germans. No wonder they've lost every war they ever started.

That's only because an engine is relatively cool. If you introduce forced induction the engine gets much hotter and a cooler intake will produce quite a few ponies.

He said: I should cool my VW with liquid nitrogen so that I can run the engine faster

Heh, that's easy - use an air/water Ic and stick ice in the reservoir.

The article did state that it was stable at 4700MHz.

Did an overclocker kill your parents or something, or are you just a pompous asshole?

A few years ago, I (and a lot of other people), bought a Celeron 366A for $70, and overclocked it by changing the frontside bus speed from 66MHz (the default for Celerons) to 100MHz (the default for Pentium 3s), making it run at 550MHz. The fastest available P3 at the time was 550MHz, and it cost something like $500.

This took about a minute to do, didn't require any extra cooling (except for $2 worth of thermal paste on the stock heatsink), and the chip has run flawlessly since then, giving me within 5% of the performance of a P3 550MHz in most applications.

To summarize, I bought $500 worth of performance for $70. Or, I saved myself $430 by overclocking.

Thousands of other people took advantage of this same underselling, it was a huge deal at the time. (Others might recount tales of overclocking the Celeron 300A to 450MHz in similar fashion -- it was a 'good year' for those chips.) Intel was selling these chips underclocked so as not to cut into the profits from the more expensive chips. (Today, in the case of new Celerons, they sell them with sky-high clock speeds to mask the fact that they've got horrificly poor performance.)

Part of the draw was that you didn't have to buy some fancy heatsink or run your motherboard at some strange frequency, or have to have any idea what you're doing beyond a few simple steps.

It's not usually quite that easy, and it's you're less likely to get the same 30%-50% clock speed gains with today's GHz+ chips, but there are still plenty of opportunities to get top-of-the-line performance from middle-tier chips without much cost or effort. Every time I build a new system, I look around to see whether any current chips in my price range are good overclockers. Sometimes there are, sometimes there aren't.

My point is that not all overclocking is the same. What Tom's Hardware, and a lot of other enthusiast sites do is just 'experimenting' to get the most performance out of what is already among the fastest and most expensive chips out there.

The article bills itself as a 'record attempt', not something practical or cost-effective. There is (as I've described) cost- and time-effective overclocking, but when someone breaks out the liquid nitrogen, it's pretty obvious that they're doing it just for fun.

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

That's among the stupidest things I have ever heard. That's exactly what Intel and AMD do! Intel especially is focused on ramping up clock speed to get more performance out of the same basic chip. The only difference is that they control over more variables in the process. Sometimes it's in the design phase, but a lot of the incremental speed ups (From say 2.5GHz to 2.6GHz) come from just cranking the clock speed up and seeing if it still works. That's overclocking if you ask me. They just happen to be the ones that decide (when they lock the multiplier and label the chip) what's "over" and what's not.

If that doesn't work, they rely on refinements in the manufacturing to give them more headroom. When they overshoot and make chips that can run much faster than they want to sell them -- well, that's where the overclockers come in.

Just an FYI, the boys in Japan have had a 5+ghz stable p4 since March.
http://son.t-next.com/
THG likes to say they do everything first, when in fact their p4 wasn't even stable at 5ghz. only 4.7ghz.
And yes. It is excessive.
-Zoson

From the article: > In plain English: 84 watts on a surface of 1.12 > square centimeters - the size of a fingertip! > Extrapolated to square meters that make 840,000 > watts or 840 kW. Not exactly true. The true number is 10000 / 1.12 * 84 = 750 KW