5ghz extreme overclocking

L5

These crazy guys at toms harware guide did this as a test, using liquid nitrogen. Insane! Check out the videos that go along with it.

http://www.tomshardware.com/cpu/20031230/index.html

bizmark

crazy stuff

i dont see the point anyway in 1 year 4 to 5 ghz pcs will be out anyways me things and you wont need super cool liquid

po0k

I feel multi-core cpus and SMP will have more of an impact.
The ceiling for curretnt tech is being reached, as the processes get smaller, the leakage between the silicon layers increases and leads to increased power consumption and less returns per watt as the frequency increases.
Course, there area few nice techs coming into the mainstream now aswell, a few based around low-k dielectrics

Matt Simis

Originally posted by L5
These crazy guys at toms harware guide did this as a test, using liquid nitrogen. Insane! Check out the videos that go along with it.

http://www.tomshardware.com/cpu/20031230/index.html

When I checked that article yesterday they had:

No Benchmarks at all
All the Asus motherboard mod pics removed
The Intel processors ID blured out (most likely isnt a normal P4)

Didnt check again (too many bloody pics), but it wouldnt be THG if their wasnt that air of suspiciousness around would it?

Also, why bother with the P4? A64s at much lower clocks are actually faster. Check out: http://www.xtremesystems.org/forums/showthread.php?s=&threadid=25313&highlight=K8NNXP

People need to put this mhz thing to rest.


Matt

joePC

I think there missing the point, This type of test is excellent, it shows how a 3.2Ghz P4 can do 5.25Ghz with some modding & extreme cooling,

I think SyxPak has some good points the P4's are made using the 0.13 Micron process & the P5's will be 0.09 there getting to the stage where there writing on atoms -_-

Optic processors are the future there in use by the millitary & are dam fast but cost a **** load of €€€€€€€€€.

Multicore sound like the next step.

Thanks joePC

astrofool

there is 0 chance the military would be using any type of cutting edge tech

most military type hardware will be running on old 8086's and at newest pentium 1's. Chances of failure are too high with new processors esp. when it comes to being in control of weapons (mass destruction or otherwise), added to the fact that most military type os's will be written at assembly level so they can see everything that is going on.

Tom's hardware also didn't explain how the ice build up all over the motherboard (visible in the pictures) didn't cause the motherboard to fail, despite saying earlier that they prevented any condensation on the motherboard.

Matt Simis

Water isnt *that* conductive (it gets more conductive thanks to the salt\other particles left by handling components), and ice is far less conductive. On my numerous overclocking experiments Ive had ice build up on components too, its fine. Its only a problem when it thaws and turns to water, but even then Ive ran pcs with (filterered, bottled) water dripping on them. The water was actaully quite warm when I noticed, but the equipment was still going (sounds card, mobo and gfx).


Matt

joePC

>>astrofool

Optical Processor

http://www.computerweekly.com/Article126338.htm

joc_06

Originally posted by astrofool
there is 0 chance the military would be using any type of cutting edge tech

most military type hardware will be running on old 8086's and at newest pentium 1's. Chances of failure are too high with new processors esp. when it comes to being in control of weapons (mass destruction or otherwise), added to the fact that most military type os's will be written at assembly level so they can see everything that is going on.

Ah how do you know this and can you explain please??

8086's ran at 50 mhz. i doubt any company in the world use these for any thing important anymore. I dubt a military would be using them.

Where did you get your info? maybe you can prove me wrong...

Capt'n Midnight

Originally posted by joePC
>>astrofool

Optical Processor

http://www.computerweekly.com/Article126338.htm

Actually it's an analog computer so is really only use for doing certain type of mathematical calculations within certain tolerances (at a guess 0.1% accuracy)

astrofool

do a bit of research into it joc, i read an article on it a while ago. I remember a couple of years ago they "upgraded" the hubble space telescope from a 386 to a 486, and they said at the time they'd never use pentiums for it.

Capt'n Midnight

For JOC - http://www.boards.ie/vbulletin/showthread.php?s=&threadid=126187

elexes

Originally posted by astrofool
do a bit of research into it joc, i read an article on it a while ago. I remember a couple of years ago they "upgraded" the hubble space telescope from a 386 to a 486, and they said at the time they'd never use pentiums for it.

ye same with the shuttles . but im sure they will be forced into making a move to a better processor that can take a lot of punisment and that would be able to last 14 years without probloms . maby make there own ?

cerebus

Originally posted by SyxPak
The ceiling for curretnt tech is being reached, as the processes get smaller, the leakage between the silicon layers increases and leads to increased power consumption and less returns per watt as the frequency increases.
Course, there area few nice techs coming into the mainstream now aswell, a few based around low-k dielectrics

Low-K has been a bit of a disaster for the merchant fabs. TSMC had huge problems qualifying their .13u Low-K process, for example - lots of dfficulty getting parts to yield. I think Intel and IBM did a bit of a better job...

Even with these problems, 90nm is going to be Low-K all the way for TSMC - they aren't offering FSG at all, as far as I know. Here's hoping it goes a bit better for them this time around.

On the leakage power front, things are getting quite scary. Low-K won't help very much here, but people (e.g. Intel) have done some cool things like changing bulk/body bias voltages in power-down to reduce leakage. As Syxpak mentioned, effects like gate tunneling (electrons tunnel through the gate oxide) are starting to dominate over the well-understood source-drain leakage.

One of the more frightening leakage scenarios is thermal runaway - leakage currents consume power, this power is dissipated as heat which raises temperature, leakage currents have an exponential dependency on temperature so these increase, this causes more power => more heat => higher temp => higher leakage => ad infinitum

Eventually the oxides will break down and things don't work any more. Nasty.

zAbbo

Originally posted by joc_06
Ah how do you know this and can you explain please??

8086's ran at 50 mhz. i doubt any company in the world use these for any thing important anymore. I dubt a military would be using them.

Where did you get your info? maybe you can prove me wrong...

8086`s ran at 5,8 and 10mhz, had 29,000 transistors(1979 release i think).
Not sure how many of them are around( a fair few id say). From experience id say there is a good amount of pre Pentium cpu`s running in Irish Companys. We have 2 Intel i486DX`s(50mhz ones, originals were 33mhz) at work, not used now but were up and runnng until 6 months ago. 80586`s were the first Penitum(pentium being 5) ran at 60-200 MHz, also have one of them knocking around as well.

I`d be guessing Military are running anything but Pentiums