Overclockers Australia!
Make us your homepage. Add us to your bookmarks  
Major Sponsors:
News
Current
News Archive

Site
Articles & Reviews
Forums
Wiki
Image Hosting
Search
Contact

Misc
OCAU Sponsors
OCAU IRC
Online Vendors
Motorcycle Club

Hosted by Micron21!
Advertisement:

OCAU News
GeforceMX Core Voltage (0 Comments) (link)
 Monday, 26-March-2001  12:27:12 (GMT +10) - by Agg

This info from Les Tutt - he of the holden radiator watercooler from nearly 18 months ago:

The last email (about a year ago) was about raising my vcore on my banshee cpu. This time I've got some info on my SUMA geforce MX. I'm still running one of the K7M's that came in an unbranded box, so have some probs with my geforce.

It seems to be an interface voltage problem, as the onboard 3.3 volt rail ( for AGP , PCI and ram) is default set at 3.45v. This is great for overclocking the ram, but no so good with a geforce MX (ie video ripping and jags in 3d). The general fix on the net is to wind the voltage to stock 3.3. No good for me (do you think I's wind my old 550 athlon back from 920, no way, and won't even post at 3.3v), so I started looking for alternatives. I confirmed the prob by taking the mb 3.45 to 3.6 and made the prob worse. Obviously the problem is the slightly changed voltage switching characteristics at 3.45v (ie the GPU core runs at 1.9v standard, and the switching time in relation to 3.3v is slightly different than in comparison to 3.45v).

I made the assumption it was the core rather than the ram, as the ram is running at 3.45 from the mobo-generated rail, not a regulator on the vid card. I tracked down the info on my cards regulator, a UniSem US1150 (I've seen this reg on many mx's). I paralleled 1 resistor to lift the GPU core .15 volts to 2.05 (hoping this would retime the bus to the GPU) and my problems dissappeared. I then also noticed I could for the first time ever turn the AGP 2x on and not crash as soon as initialising the vid card into 3D mode. Also has had the really 'bad' :-) side effect of allowing my core to o'c another 10 mhz (probably just due to the raised vcore rather than the timing).


Nice one. Good to see you're still out there tweaking, Les. :)



Return to OCAU's News Page

Advertisement:

All original content copyright James Rolfe. All rights reserved. No reproduction allowed without written permission.