Strange limit on Epox 9nda3+... on two processors!

japesgalore

Distinguished
Sep 21, 2008
1
0
18,510
I know I have a fairly old motherboard here but recently ended up buying a HD3850 AGP ('cos my X850XT PE couldn't handle Bionic Commando - no SM3) and realised it was being seriously limited by my Athlon XP3200+ (overclocked to 2500mhz). So, hunted down an Opteron 165 that was listed as being capable of 2880mhz. Well, in my computer, for all my efforts, I could not go beyond 2700mhz. It would either not post right, BSOD just after the windows logo or give me a message saying that the Windows system file was either missing or corrupted. Any options on dividers, multipliers and voltages made no difference - I had hit a wall. Thinking I was lied to I went and hunted down another Opteron 165 with supposedly a good stepping. Initially was was very pleased, and relieved because it reached 2700mhz without an increase in VCore. But then... EXACTLY the same problem!

At least I know this isn't an issue of the CPU limiting the overclock. I have 2x1gb GSkill PC3200 sticks that operate up to ddr500 with loose timings. I can boot with FSB well above 300 with reduced multiplier and memory divider. Can there be anything else I've missed here? Does anyone have this board and experience with overclocks this high? Could it be an issue with the location of my SB Audigy (it's in the PCI slot second from the end) or the RAM (they're in slots 3 and 4, the ones further away from the CPU). My PSU is a Tagan 480Watt. Somehow I doubt it's the cause since the wall is so damn stubborn, whatever the VCore. Please, any suggestions? I gotta crack this... :fou:
 
The Opteron 165 is a 1.8GHz piece and you are complaining about not getting to 2.88GHz, merely 2.7GHz...on an Nforce3 AGP board with air cooling. Let's make it clear you are asking how to push past the envelope for your years old gear.

Not to mention you do not increase the VCore on Opterons generally, certainly not under the conditions you are describing, it doesn't help.

In your case I would run the FSB as high as possible, up to 250MHz and then increase the CPU multiplier until the CPU conks out keeping that final frequency as your ceiling. All the while keeping the RAM at CAS 2.5 and whatever other timings the ram can muster.

I would not expect 2.88GHz, nor would I expect you to be able to run the 165 at the highest multiplier, nor would I suspect you to be able to run hypertransport at 5x or 4x....3x is plenty sufficient and stable in that scenario.

If you are still of the mind that you need to increase voltage...the nf3 chipset itself does need more current at those speeds, here's a link to some results, from the wayback!

http://www.pcstats.com/articleview.cfm?articleid=1682&page=2