New build overclock ...

rpwarwood · 17 Jun 2011 at 19:09

Just got my new build up and running and it seems happy... currently installing drivers etc.

Idel temps, @ stock settings, with Silver Arrow and two TY-140s:
Temperature on each core is 28oC

Not bad starting opint, will let everyone know how it goes when I start the OC !

rpwarwood · 18 Jun 2011 at 14:57

Looks like temps will never be an issue, but doesn't seem that my Chip is a great one:

Can only get a stable Prime run at 4.5GHz with the following settings:

PLL overvoltage disabled
Vcore 1.31V
Everything else on Auto

Was hoping to get in under 1.3V, but core 3 constantly failed Prime within 5 seconds below 1.3. At 1.3 it failed after about a minute. 1.31 its been going half an hour and "looks" ok.

Temps max at 53/56 at load, and idle at 31/34 (all depending on which core)

Also, Core 0 is noticeable slower than Core 1 in Prime, idles hotter, but is cooler at load. Rather weird standard variation due to manufacturing process!

Anyone had similar resutls to this? What did you max your CPU at? I'm hoping to have mine running 24/7 @ 4.6GHz and no higher than 1.34v but will see how it goes.

Are there any other settings I should change to assist the overclock? My XMS RAM is rated at 1.65V, but I've read that this is not a recommended voltage setting. Can anyone confirm whether this RAM usually runs its rated timings at 1.55V ?

Thanks!

Phenomenologica · 18 Jun 2011 at 16:26

Core 0 normally runs slower in Prime because it's the one other processes are more likely to use, so some of its processing power is diverted elsewhere. It idles hotter for the same reason, gets more use at idle than the other cores do.

As to the overclock, you'll want to test for a full 8 hours with Prime to check stability. If it crashed that fast at 1.3 I'm not convinced it'll last more than an hour or so at 3.1, but time will tell. I haven't tried getting 4.6 stable on mine, just 4.5 and 4.7, but extrapolating from them I imagine 4.6 would require about 0.025 - 0.030 volts more than 4.5 Ghz on my chip. So it's possible you could (just) get 4.6 stable under 1.340 volts. However your mileage may vary, and you may have to go higher.

rpwarwood · 19 Jun 2011 at 10:30

Hmm, it made it an hour in then died, as you suspected. But ran it again and it died within a few minutes, so definitely not stable.

Prime now been running overnight at 1.32 and its fine. Doubt I'll bother getting it to run at 4.6. I'm sure the processor could handle the extra voltage, and the cooler is doing wonders, but I'm happy at 4.5. Max under load is 59 on my second core so thats more than fine.

May see how much it needs to get 5Ghz just out of curiousity, but think I've found the level. 1.2 for free is not to be sniffed at.

What temps are you getting on yours running at 4.7 and what cooler are you using?

Phenomenologica · 19 Jun 2011 at 10:50

Temps are in the high 60s/low 70s depending on ambient, hopefully I'll be able to drop that a bit when my new TIM arrives but it's taking its time. I'm using a Corsair A50 so hardly anything special, but my chip is reasonably good so I only need 1.325v to keep it stable at 4.7, which keeps the temperatures down a bit.

rpwarwood · 20 Jun 2011 at 12:36

Nice chip

...shame mine seems to run cool but doesn't look like it enjoys going above 4.5, let alone near 4.7.

I'm now happy with CPU @4.5, and I'm now at the stage where I'm looking to get the RAM running at its recorded timings (XMS3 C7 rated at 1.65V)... DRAM Voltage is currently 1.5, but I wil raise this up to 1.55 as needed to get the rated timings. If they don't work at 1.55 i'll aim lower. Has there been any conclusive proof that going over 1.58V causes longterm controller/cpu damage, or is it all ok provided the VTT is set high enough to maintain a 0.5V difference?

Should I set the VCCIO (VTT) to 1.1V to ensure the <0.5V difference is maintained?

rpwarwood · 22 Jun 2011 at 00:29

Ok, having real issues with the chip...

It is only stable @4.4GHz on 1792k in-place Prime95 FTT's, when I have VCore of 1.33V! Seems excessively high, although i have seen other (very few) users requiring higher. I get massive Vdroop under load, its dropping down to 1.304 under load according to CPUz!

This suggest I have a poor chip, which is just bad luck, but is there anything else I can change that will help?

I have every other voltage set to auto, and have set LLC to High (not willing to use any higher), and disabled PLL Overvoltage. Any suggestions to help stabilise my CPU?