lowering the voltage for intel's also?

[nicco eneidi]

Hey, On the front page there is a little blurb saying that some people seem to overclock better with lower that normal voltage for the amd cpu's.

Now i am wondering if anyone has done this with Intel's chips. :)
I mean has anyone had better overclockinbg results with lower the default voltage?

[Spode]

I know Daniel had a 500E that would only do 733 when using less voltage. I remember when he told me this I was very surprised - this concept was new to me. I wonder what the physics behind it is?

[vipez888]

Well, I tried lowering voltage on my p3-700@994mhz and Windows gave me the BSOD, it obviously didn't work at 1008mhz, looks like my FCPGA Cb0 oem likes higher voltages, at least up to 1.9V, anything higher will get halts or freezes before Windows finishes loading up eveything.


Maybe AMD Thunderbirds work better with lower voltages when OC'd, but for my intel, lower voltages won't cut it(even though the temps lowered).

[zeisel]

if you check a few posts down you will see that is what is happening with my pentium 2. i kept getting registry and explorer errors the farther i went (fsb). the lower voltage booted fine where the higher wouldnt. right now i have my pII 350@452 with 1.85v (2.00 default) and my I/O voltage at 3.3 (3.5 default). it wont boot at this speed with default voltages or higher, weird huh

[Mr. Potatoehead]

zeisel,
great pic!

[wild_andy_c]

Spode - you wanna know the Physics.

Gate bouncing - in just 2 words. The change in polarity of the ionic electron shells in the silicon gates can be overcharged by over-voltage - causing the gates to vibrate out of control rather than simply opening and closing approximatley 1 million times a second. This causes a different effect to a lockup whereby heat is produced due to over movement but a protection error will occur due to the nonsensical computation of the CPU.

When a chip will move easily, the lower voltage will definatley not over move gates - thus making them stable at their job.

This is all besides the heat dissipated in the simple equation of power related from input efficiency and work done.

This is something I took up with my lecturer before Christmas. We talked for hours and drank lots! My mentor - my Obi Wan

He has theories like no other man. He suggested atomic mass 29 or 28 silicon to me about 12 months ago - that should give you an idea of his calibre

[Spode]

I see. That makes sence. But why are some susceptable to it and some not?>

[DaveB]

Nice theory, but in practical terms you're talking about over-volting which is more likely to cause permanent damage than simple instability. The effect being explored in this thread is more about running the system for a while at a high voltage and then setting it back. This type of "seasoning" can often allow you to run at the same rate, but at a lower voltage. And probably more due to a decrease in heat generated at the lower voltage, you might even geta little further. Especially with those little space-heating T-Birds. As we all know, keep it cooler and it goes further. That's why all the extreme guys go for Peltiers and Water Cooling rigs. Heat is the enemy, not voltage. But, lower the voltage and you'll lower the heat.

My PIII 700, with it's nominal 1.65V, needs 1.95V to get stable at 994 MHz where it's been running for months. Any attempt to lower the voltage simply results in crashes. It never crashes at 1.95V. The Alpha keeps the temp at 34C max, but the CPU is more sensitive to temperature when it's overclocked more. That's why I can't run stable any higher. It will run at 1008 MHz for Windows apps, and even 1050 MHz for a little while, but will eventually crash. I'm sure if I went for a Peltier, I could get to 1050 MHz but it isn't worth it for a lousy 50 MHz.

[bas]

wild_andy_c (Feb 12, 2001 02:54 p.m.):
Spode - you wanna know the Physics.

Gate bouncing - in just 2 words. The change in polarity of the ionic electron shells in the silicon gates can be overcharged by over-voltage - causing the gates to vibrate out of control rather than simply opening and closing approximatley 1 million times a second. This causes a different effect to a lockup whereby heat is produced due to over movement but a protection error will occur due to the nonsensical computation of the CPU.

When a chip will move easily, the lower voltage will definatley not over move gates - thus making them stable at their job.

This is all besides the heat dissipated in the simple equation of power related from input efficiency and work done.

This is something I took up with my lecturer before Christmas. We talked for hours and drank lots! My mentor - my Obi Wan

He has theories like no other man. He suggested atomic mass 29 or 28 silicon to me about 12 months ago - that should give you an idea of his calibre


wild Andy talking like that scares me and confuses me:). Man thats some deep ****. alot of smart guys on the boards. IM the kind of guy that cant count past 10 unless I take off my shoes or pull down my pants;). hehe Later

[Spode]

DaveB (Feb 12, 2001 05:28 p.m.):
Nice theory, but in practical terms you're talking about over-volting which is more likely to cause permanent damage than simple instability. The effect being explored in this thread is more about running the system for a while at a high voltage and then setting it back. This type of "seasoning" can often allow you to run at the same rate, but at a lower voltage. And probably more due to a decrease in heat generated at the lower voltage, you might even geta little further. Especially with those little space-heating T-Birds. As we all know, keep it cooler and it goes further. That's why all the extreme guys go for Peltiers and Water Cooling rigs. Heat is the enemy, not voltage. But, lower the voltage and you'll lower the heat.

My PIII 700, with it's nominal 1.65V, needs 1.95V to get stable at 994 MHz where it's been running for months. Any attempt to lower the voltage simply results in crashes. It never crashes at 1.95V. The Alpha keeps the temp at 34C max, but the CPU is more sensitive to temperature when it's overclocked more. That's why I can't run stable any higher. It will run at 1008 MHz for Windows apps, and even 1050 MHz for a little while, but will eventually crash. I'm sure if I went for a Peltier, I could get to 1050 MHz but it isn't worth it for a lousy 50 MHz.

Thats all well and true that in most cases you need MORE voltage to overclock, and I can see what you are saying that we can only overclock more becuase less voltage=less heat, but I think this is a another issue all together. something different is going on to make you need less volts to overclock more.

[Newbie_Doo]

Could also be that as the manufacturing process matures and yields improve, what once was a spec voltage for a processor at a given speed now creates crosstalk. in the paths. Increases in the voltage increase the crosstalk until a chip becomes unstable. When the voltage is lowered, as long as the current is sufficient, reduces the crosstalk and improves chip performance. Think of signal-to-noise ratio in audio. Above a certain point in an amplifier's output, you may get more "sound" but at an unacceptable level of distortion.