undervolting novice

The forum for non-component-related silent pc discussions.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
ajagodnik
Posts: 19
Joined: Mon Mar 15, 2004 12:37 pm

undervolting novice

Post by ajagodnik » Tue May 11, 2004 9:39 am

Hi,

I have an epox 8rda+, athlon xp 2500+.
I would like to optimaly/safely undervolt CPU a bit to reduce heat that is produced in order to run TT silentboost's fan a really low RPM (800-1200) using Zalman fanmate.

By optimaly/safely I mean the system has to be 100% stable and I want to get just enough low temperatures needed for low RPMs, without losing taht much CPU power.

Thanks.

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Tue May 11, 2004 10:05 am

Here's the method I use to undervolt:

Start dropping the vcore one notch at a time, then run Prime95 for an hour to check for stability problems. (a search here will net you lots of info on Prime95, it's an excellent way of testing for stability)

I repeat the above step, dropping a notch, running an hour, dropping a notch, running an hour, until I hit a vcore where Prime95 fails during the hour. Then I raise the vcore a notch and run Prime again. If Prime 95 will run >12 hours without and error, I consider it stable.

Your 2500 will probably be able to drop .1-.25 volts. That'll reduce your heat ouput by a very noticable amount, something in the neighborhood of 25%, all without compromising performance or stability.

geordie
Friend of SPCR
Posts: 175
Joined: Mon Jun 09, 2003 4:00 am
Location: Reading, UK

Post by geordie » Tue May 11, 2004 11:06 am

Yep. That's what I'd do too.
All 3 of my 2500's are stable with 1.525v, so if you want to save a bit of time you could use something round there as a starting point.

ajagodnik
Posts: 19
Joined: Mon Mar 15, 2004 12:37 pm

Post by ajagodnik » Wed May 12, 2004 3:33 am

I have undervolted the cpu to 1.55 V, that's approx. 0.1V lower than default (1.64).

I will do some testing and measurements to see if it is enough cool...if I get lm_sensors to work :(.

Thanks.

burcakb
Posts: 1443
Joined: Tue Mar 09, 2004 9:05 am
Location: Turkey

Post by burcakb » Wed May 12, 2004 5:44 am

8rda+, especially rev 1.1 is known as a cheapo overclockers board and does not like undervolting. So be very sure about your stability. I have the same board too (in wotan) and 0.1V is the max I got. Also run memtest. epox doesn't need to use the CPU a lot to fail, run anything and it'll probably crash/lockup within 15 min.

ajagodnik
Posts: 19
Joined: Mon Mar 15, 2004 12:37 pm

Post by ajagodnik » Thu May 13, 2004 1:05 am

I have 8rda+ rev 2.2,
recently someone advised me too, that 8rda+ is not that keen on changing processor voltage. It can "burn" one particular chip one can change, but still...
???

silvervarg
Posts: 1283
Joined: Wed Sep 03, 2003 1:35 am
Location: Sweden, Linkoping

Post by silvervarg » Fri May 14, 2004 7:16 am

It can "burn" one particular chip one can change, but still...
I fail to see you dropping the vcore could burn anything. The only component that might possibly work harder would be the voltage regulators/FET, and personally I don't refere them as chips.
I guess that the advice comes from someone who as raised the vcore, so it would not apply to you.

Undervolting typically causes instability that can make the computer crash, but as soon as you restore the settings you are good to go again.

ajagodnik
Posts: 19
Joined: Mon Mar 15, 2004 12:37 pm

Post by ajagodnik » Sun May 23, 2004 8:15 am

well, I didn't meant burn in literal sense, but as you said...he told me this "thingy" could work more and thus fail and then needed to replace, a bit annoing after experiecing my vendors's policy on warranty/service.

at 1.64 stock voltage I get. around 47C idle and some 53-55C under load (1h of mp3,mpeg,3d).
at 1.55V it was around 43C and less idle.

Post Reply