My initial Opteron 170 overclocking results

[Phew]

I got an 0546 Opteron 170 (2*2.0 GHz, 2*1MB) from Monarch in the mail yesterday, and popped it in my DFI NF4-Ultra D (after a BIOS flash). It is cooled by an XP-90 with a (slow/quiet) Panaflo 3000 RPM fan.

I got up to 2.5 GHz on stock voltage (1.33 V), dual Prime small FFT stable all the way up. I was using a 166 memory divider, so when I bumped it up to 2.6 GHz, my Patriot 2-3-2 RAM was then running at 216 MHz, which was probably too fast for CAS2 on stock memory voltage because the computer hung up overnight while I was priming. So I dropped the divider down to 150, and upped the CPU voltage to 1.4V (just in case that was the problem), and I am currently 3 hours dual Prime stable at 2.6 GHz.

I don't remember the idle temperature, but I think it was mid-30s C. The load temp @ 2.6 GHz is 49 C, which is a little high for my taste. I might find out the limit for this processor and then either bump down to a cooler overclock or bite the bullet and buy a louder/faster fan.

I'll keep you updated. Overall, I'd say these dual-core Opterons are real gems.

 

[Puffnstuff]

I've ran mine up to 2.5 with the memory still at 1:1 however stock is really fast enough for me so that is where I leave it most of the time.

 

[Leper Messiah]

is that stock volts? Well, anyways, congrats on your FX-60 equivalent.

 

[Phew]

No, it's a 5% voltage increase (from 1.33V to 1.4V) to get to 2.6 GHz. I might be able to get by with less, but I'm not in fine-tuning mode yet (just seeing what this chip can do). I'll post more updates tonight after I push it further and/or tweak it some more at 2.6 GHz.

 

[Leper Messiah]

Thats pretty good, my x2 3800 does 2.6 at 1.4 vcore, haven't taken it any higher from there.

 

[nealh]

from what I have seen and experienced with my opteron 165...it appears they oc well to a point at low voltage then you start to need more vcore toget much more out of the chip

I hit 2.5ghz @low vcore...I cant recall now whether it was 1.35v or 1.4v...(my opteron has default vcore of 1.35) so small bump

but for me to get to 2.6, I need 1.46 vcore in my bios/1.44v in winxp..and temps starting getting hotter as I move closer to 1.5-1.53v...not worth 100mhz IMHO

So I bet most will find they will overclock to certain point on low vcore then any extra mhz will take a much larger proportional jump in vcore

and of course OCing is always YMMV

looks like these 170 should hit 2700 easily

 

[Phew]

UPDATE: I have experienced the exact behavior nealh is referring to.

I finished 12 hours of dual Prime stable @2.6 GHz, which required only 1.38V. No errors, max temp 49 C.

Any higher requires big voltage per MHz increases. 2.7 GHz @1.43 will boot into Windows, but locks up after a few minutes of priming with the Temp approaching 52C. I am sure I could get it stable with 1.48V or so, but 100 MHz is not worth 7% more voltage and the associated heat.

I am very happy with this stable, relatively cool, low voltage 30% overclock.