How come you can't constantly overclock

Tab

Lifer
Sep 15, 2002
12,145
0
71
I know for a fact you can't overclock a 166mhz Pentium to 2Ghz. I don't know why, I just know you cant. My friend doesn't think so. He says with proper cooling you can get it to any performence. I said the FSB Speed wont be enough, he said he could overclock that too. How can I prove him wrong?
 

Pink0

Senior member
Oct 10, 2002
449
0
0
You can prove him wrong by giving him an IQ test.
There are several reasons you can't overlclock a 166 to 2000mhz. First and foremost, the multiplier is locked. In order to get the chip to those speeds you would have to run the FSB at 800mhz. Obviously this is an impossibility in itself since no p1 board supports this and no chipset for the p1 would be able to handle even 1/10th that. secondly, this would force the PC66 ram in the system to also run at 800Mhz. Completely impossible.
The voltage that would be required for such a ridiculous overclock, even if the multiplyer weren't locked, would be so tremendous that current would leap from one gate to the next and the processor would be dead within a microsecond.
No matter how much cooling you give it the .25u pentium 1 could not reach even 400mhz let alone 2000.
 

Tab

Lifer
Sep 15, 2002
12,145
0
71
How could I explain this to someone who 'thinks' they know alot about computers?
 

cheapgoose

Diamond Member
May 13, 2002
3,877
0
0
lol..don't we all have friends like that?

tell him to post his theory, or what he claims he can do on here.
and we'll just let the forum take care of him.
 

mechBgon

Super Moderator<br>Elite Member
Oct 31, 1999
30,699
1
0
You should make sure you don't inhale any of whatever your friend is smoking... it's probably NOT good for you! :Q
 

zsouthboy

Platinum Member
Aug 14, 2001
2,264
0
0
Isn't there a quote about this?

"There's a right way and a wrong way, but you can't overclock a 386 to 1 ghz ANY way." :D
 

CTho9305

Elite Member
Jul 26, 2000
9,214
1
81
There are a few limiting factors. At the extreme limit is the speed of light. 186000 miles per second, so at 1ghz you can go .000186 miles, or 1 foot. Electrical signals are slower, so at some point, you just can't get a wire to carry a signal across the whole chip in one clock cycle.

Before that comes the transistors. A "big" transistor takes time and current to switch. As I understand it, you can increase the voltage and that decreases the time required. However, it raises the current (and thus heat), and at some point, the electricity will just spark across the transistor and destroy it. Another thing to consider: transistors havesome cutoff voltage for on/off. in a 5V 386, this might be 2.5V (I have no idea). normally, signals would be .1v for a 0, and 4.9v for a 1 (also a wild guess). if you run it at 50V, leakage currents might mean that a 0 is over 2.5V, but the transistor would see that as a 1. Think of a light switch - you can only turn it on/off so fast. If you go too far, you'll just break the switch.

Also, the clock does not tick at the same time at all parts of the chip. Lets assume that in a 386, the "clock skew" as it is called is 1ns. If you try to run the chip at 1ghz, some parts of the chip are a whole clock cycle ahead/behind others. That is a bad thing ;)

There are a lot more things that come into play, but I think that is enough info for now ;)
 

DAPUNISHER

Super Moderator CPU Forum Mod and Elite Member
Super Moderator
Aug 22, 2001
28,275
19,922
146
Originally posted by: Tabb
How could I explain this to someone who 'thinks' they know alot about computers?
You don't! Ignorance is it's own punishment ;)
 

Tab

Lifer
Sep 15, 2002
12,145
0
71
My friend's theory on how he is going to overclock?

1)He will up the voltage

2)Overclock the BUS

3)Get a big heatsinks etc...

He is trying to go from 1.1 Ghz to anyspeed

He has a k7s5a
 

masterxfob

Diamond Member
May 20, 2001
7,366
3
81
My friend's theory on how he is going to overclock?

1)He will up the voltage

2)Overclock the BUS

3)Get a big heatsinks etc...

He is trying to go from 1.1 Ghz to anyspeed

He has a k7s5a
i guess he'll be making a keychain out of the cpu after he's done testing his theory?
 

SupermanCK

Platinum Member
Mar 31, 2000
2,264
0
0
ROFLMAO!!!
p1 166 to 2GHz...if that could happen...intel would either be stupid or really smart
stupid part being they can't even do it without new tech advances
smart is that they are rebadging all p1 to p4s...and selling it at $600+
 

Insidious

Diamond Member
Oct 25, 2001
7,649
0
0
so at 1ghz you can go .000186 miles, or 1 foot. Electrical signals are slower, so at some point, you just can't get a wire to carry a signal across the whole chip in one clock cycle

Call the CPU 1/2"............. sounds like................. 48GHz *I could live with that* :D



Actually, stray capacitance will limit your speed. (as technology has advanced, reduction of this capacitance by better device fabrication methods has been a significant contrubutor to the gains in speed. )

capacitive reactance is a positive function of frequency. As reactance increases, so must voltage to maintain the same current. You reach a point where the voltage required to provice sufficient signal strength is beyond that which the materials of the semiconductor are insulative. From this it is clear that manufacturing process control being imperfect yields some chips which overclock better than others because of differences at the molecular level which affect capacitance.
 

kursplat

Golden Member
May 2, 2000
1,547
0
0
hey whats with you guys . this isn't the anandtech way. we got to help this guy .
now then, i'm willing to bring the liquid nitrogen to help with the cooling , if somone else can get a flux capacitor (second hand is OK).
also i might be able to get us access to a 5kv line for a bit to run the coolers .
sombody want to pick-up some snacks ? going to be a long night....
 

teqwiz

Senior member
Sep 8, 2002
603
0
0
Buy him the 166 on EBAY. Let him prove his theory. That is how you prove him wrong.

BTW The cost of a 166 is plenty worth the amusement factor you'll get watching him struggle with this one!:)

Ask youself why you would even need to cosider proving this theory wrong? Hm.....

lol;)
 

FluffyBunny

Member
Apr 21, 2002
29
0
0
WOW! This thread really speaks me! I'm goin' to blow the dust off my 450 and super tune that bad boy. If a 166 can do 2.0Gig I should easily get 3.0Gig with it! And imagine I wuz going to spend a couple bills on a 2.53 P4 for nothing. While I'm @ it I might as well turbo charge the 8MB on board video chip to at least match the ATI 9700 and save myself another couple of hundred. And if i put my IDE cables upside down on my 13 Gig Hard Drive it should yield another 20-30 Gigs right? Man! I'm saving a fortune!!! Lets see now, take the cover off the DVD ROM; put some jelly on the laser lens and BAM-O, instand DVD BURNER!! Hmmm...lets see...maybe I can convert my 3.5 drive into a zip drive...that should'nt be too hard...

ADVICE: FIND SMARTER FRIENDS
 

kursplat

Golden Member
May 2, 2000
1,547
0
0
While I'm @ it I might as well turbo charge the 8MB on board video chip to at least match the ATI 9700 and save myself another couple of hundred.
oh ya thanks for reminding me . i have 2 voodoo 2's SLI goodness. should be like 2 9700's together.....wow
 

CrazySaint

Platinum Member
May 3, 2002
2,441
0
0
Originally posted by: chuck340
hey whats with you guys . this isn't the anandtech way. we got to help this guy .
now then, i'm willing to bring the liquid nitrogen to help with the cooling , if somone else can get a flux capacitor (second hand is OK).

I'll call up Doc Brown and see if I can borrow the one off his Delorian.
 

rbV5

Lifer
Dec 10, 2000
12,632
0
0
I know for a fact you can't overclock a 166mhz Pentium to 2Ghz
Sure you can, I overclocked my cordless phone to 2.4GHz, you just need a bigger HS/F and bump the FSB way up.