Opteorn 165, 1.6v with good temps = okay?

[Thasp]

This CPU is definately not going to see the light of day for more than another year, if that.

 

If I can dual prime at 1.6v with good temps, is this okay, or has anyone killed their opterons with that voltage?

[EllisD]

What temps are you hitting atm with 1.475V?

 

I know if i bump mine to 1.58V its in the Mid to high 50's and NOT prime stable (dies in about an hour) so i personally have no need to go that high (but i dont have a very good stepping)

 

1.6V is generally water territory from what i have seen.

[kitfit1]

For the longevity of a cpu, 1.6v on air may be pushing your luck a bit. Having said that, were all clockers around here so the cpu's long term life expectancy is'nt of any great worry. After all most of us arn't going to be running the same cpu in six months let alone a year. I'm firmly in the camp of "screw the nuts off it till it squeeks" then ask it for a bit more . on a serious note though, if you want it to see you through to retirement then back off.

[ButtChew]

Whats your idea of good temps at 1.6v? I cant imagine it being lower than high 40's and probably 50's ... especially under load.

[Thasp]

oh. I have to update my sig.

 

I don't have that RAM anymore, and 1.475v = unstable. Stable for dual prime for nine hours, not stable for me to plug a USB UMS device in. Go figure - prime can be such a waste of time, sometimes!

 

The heatsink is pretty huge here, so I don't think temps would be a big issue. I'm at 1.550v at 2.8 GHz now.

 

I'll try 1.6v later. I'm running memtest in a bit.. for some reason, I can encode for 11 hours straight, and not have an issue, but certain software freezes the computer up at 2.8 GHz. Only when the memory is at a higher speed, it's fine at lower speed. But the memory worked at that higher speed at 2.7 GHz!

 

I figured I could test memory than CPU stability, but it appears the speeds the memory worked at before goes down when the CPU goes up.

[dr_bowtie]

just because the ram passes memtest at a higher speed doesnt mean its totally stable... I learned that...it could be the ram controller on the chip..?

 

I could pass memtest with my XBLK at 2.5-3-3-7...2.9+.03 up to 294 with out error...I got the 1st error at 295 but could be tweaked from the secondary...But I could NOT pass prime at 280 or higher with those timings..?

 

I backed them off to 3-4-4-8 and all was fine...that 15mhz difference...but loosening them only cost me 200ms in super pi 1M.... and upping the fsb to 281 took care of that...

 

just another [email protected] to [email protected] is a pretty good jump...

 

you wont really notice much diiference in anything but encoding by staying with 2700....

 

I encoded and every 10fsb between 200 and 300 (10 multi) and I didnt see much difference in most of them till I got over 280...it jumped at 290 and a pretty good jump at 300....

 

I also tested the 11 multi at the same speed but lower fsb...it wasnt till i got over 2800mhz I saw the most gain...but not like I did with the 10 and higher fsb....make sence?

 

try loosening the ram a little and see what happens...

[gdogg]

for me on high end air, I get 41 Load at 1.6V on an opty 165

same with 1.55V 41 Load

 

this is @ 2772 Mhz 308X9.

 

oh I forgot room temperature 26C when computer is idle 29-31C when computer is load

 

cpu idle temps 29-31C

 

also I have high volts on the memory, which could be why i've been able to burn-in my cpu for many, many days at a time w/o killing it, on air.

[Thasp]

My Tuniq is basically the same thing as the thermalright, but my CPU gets much much hotter at 1.6v. However it also requires less voltage to get 2772 MHz, so I guess it's a fair tradeoff.