Exponential Voltage Increase

[skligmund]

Well, I have an AMD XP1800+ T-Bred B running on an A-Bit KT7e (KT133E). Right now, the core voltage is at 1.625 (1.600 is default), and the clock speed is 1696.8 MHz. If I bump it up to 1733, I need about 1.700 Volts. If I go to 1768, I need 1.825 Volts. I think it ran fine at a few MHz higher than 1768, but I felt the gains of speed vs. heat wasn't worth it. It takes 0.2 volts to overclock 100 MHz over my current speed. Of course this CPU originally ran at 1533 MHz, so the first 175 MHz was free.

Just as I write this, I wonder if my VIA Hardware Manager and BIOS are more accurate than CPUID, because CPUID says my core voltage is 1.575, a 0.050 volt difference. My BIOS says 1.625, and VIA HM says 1.640.

Wonder which one is closest to correct.

Water cooling is for the weak. Get liquid nitrogen.

 

[Teq]

Good question... software monitoring and the SMB's voltmeters are often not entirely accurate. The only real way to be sure is to put a voltmeter on the CPU pins.

I'm on an 1800+ like you and MBPROBE says 1.82 volts, AIDA says 1.66, my voltmeter fluctuates between 1.73 and 1.68 with load...

I tend to believe the voltmeter.


--->It ain't better if it don't work<---

 

[skligmund]

Never really thought about that, good idea.

What points do you use for testing your voltage?


Water cooling is for the weak. Get liquid nitrogen.

 

[Teq]

It depends on the chip and the board... I usually try to find a space where the probe tip won't cause a short. Usually it's right on the outer row of pins in the VCC group under the socket... sometimes from the voltage regulator...



--->It ain't better if it don't work<---