Skylake Voltage

AiiR · 21 Apr 2017 at 18:00

I know a few people will cringe at this but I'm fearless so here we go.

I've been testing my 6700k to it's limits. The best speed I've gotten out of it is 4.8ghz which is fully stable, but only at 1.5v. I've done some reading and Intel's max recommended voltage is 1.52v, so technically its within spec. My question is to people who have run CPUs at such high voltages for an extended period of time, has there been any noticeable degradation to the chip? How much am I shortening the lifespan by running at this high voltage.

I'm personally not worried about a shortened lifespan, the chances of me having this chipset in 1 - 2 years time are fairly slim. Temperatures are okay, low mid 60's in gaming which is all I do. Did a quick IBT run and it got to the high 80s but I'm never doing anything more intense than a 64 player Battlefield game so I'm not concerned by that.

DarkBahamut · 23 Apr 2017 at 00:29

AiiR said:
I know a few people will cringe at this but I'm fearless so here we go.

I've been testing my 6700k to it's limits. The best speed I've gotten out of it is 4.8ghz which is fully stable, but only at 1.5v. I've done some reading and Intel's max recommended voltage is 1.52v, so technically its within spec.

Put simply, no it isn't.

The maximum VID Intel will program a CPU is 1.52v, that's got nothing to do with what Intel 'recommend' or what the specs allow when overclocking. That comes from a lack of understanding about what the VID is and people reading a PDF to make them feel good about using crazy volts. The documentation is clear. If you are outside the stock clockspeed (Overclocking) and not using the stock VID then you are outside the recommend voltage from Intel (and thus out of spec). This is what the datasheet actually says, people gloss over that part though

Either way, that's some serious voltage and well into the degrading CPU range. That said I would think it would last the year or two you want though, but it may be showing signs of wear by then (instability - increasingly needing more voltage). People running older chips at lower voltages than you have even noticed signs of the chips degrading, so it will certainly happen. Is that extra 100 or 200MHz really worth pushing the voltage so high?

mattyfez · 23 Apr 2017 at 10:31

Have you tried updating motherboard bios? I couldn't get past 4.7 without silly voltage, as in 1.4.

Updated 2 days ago and I'm on 4.8ghz at 1.375v

VKleita · 23 Apr 2017 at 10:48

Just please tell me when you "Don't have this chip in two years" it's not because you sold it to some poor sap who doesn't know better....

mattyfez · 23 Apr 2017 at 13:10

Out of interest (I know it's probably impossible) it would be interesting to see how voltage affects long term stability on mainstream chips if the temperature is kept safe.

AiiR · 23 Apr 2017 at 16:28

VKleita said:
Just please tell me when you "Don't have this chip in two years" it's not because you sold it to some poor sap who doesn't know better....

If I did, I would disclose the history, price it accordingly and let the buyer decide.

I've since dialled back the speed and voltage to 4.6 at 1.35v, but am still curious on how long it would cope with 1.5v. I know I've seen people pumping 1.5v or more through Sandy CPUs since day 1 and they're still going strong.

mattyfez · 23 Apr 2017 at 16:42

There's not any data really, that's the problem.
It's all here say and let's face it, no one is going to to a proper test, you'd have to have a few identical rigs running high volts with reasonable temperature for a couple of years, by which point the platform will be obsolete and no one will care anymore because it's not relevant to the new chips.

Doug2507 · 24 Apr 2017 at 13:45

mattyfez said:
Out of interest (I know it's probably impossible) it would be interesting to see how voltage affects long term stability on mainstream chips if the temperature is kept safe.

You can still easily kill a chip with high voltage even if temp is deemed as good. Iirc there's a vid on YouTube of a well know bencher killing an older gen chip in 10 minutes with too much volts whilst temps remain good.

fastboy · 28 Apr 2017 at 16:26

so whats a safe oc voltage for 24 hr useage?
I'm on 1.35 at 4.3ghz but I would drop the oc if the voltage was hurting the chip

pastymuncher · 28 Apr 2017 at 19:03

mattyfez said:
Out of interest (I know it's probably impossible) it would be interesting to see how voltage affects long term stability on mainstream chips if the temperature is kept safe.

Temps don't matter, you can be sub zero but the voltage will still kill the chip. I know because I did it with the original launch of the Core2Duo E8500 after being used to being able to put 1.5v through the old 65nm cpu's. I was using 1.45-1.5v trying to hit 4.5Ghz and it degraded within 2 weeks. I had to either drop the overclock or increase the voltage to get it to run stable. In the end it would only run at stock with increased voltage.

Yutang13 · 29 Apr 2017 at 12:23

fastboy said:
so whats a safe oc voltage for 24 hr useage?
I'm on 1.35 at 4.3ghz but I would drop the oc if the voltage was hurting the chip

1.4V is the generally agreed safe 24/7 voltage for skylake so 1.35 is just fine.

mattyfez · 29 Apr 2017 at 16:04

It's a matter of opinion, I'll use 1.4 volts when testing, but for some and myself, it's a little high, 3.5 to 3.75 is pretty safe for general use.

You don't really get much more gaming performance for an extra 100mhz of cpu speed or so. So if that's the difference, it's better to sacrifice a little bit of clock speed for better temperature and lower voltage.

fastboy · 29 Apr 2017 at 16:08

well I can oc my 6400 to 4.5ghz at 1.4v but dropped it back to 4.3 and 1.35 for a bit of safe headroom,as you say the last bit of oc isn't going to be much difference as the overall percentage of oc goes down as you go up.

benjii · 5 May 2017 at 11:07

In articles written by ASUS and MSI on overclocking 6700k's, they both recommend 1.4v as a good limit for 24/7 use.

mattyfez · 5 May 2017 at 22:19

AiiR said:
I know I've seen people pumping 1.5v or more through Sandy CPUs since day 1 and they're still going strong.

24/7/365, or one off 'suicide runs' for benchmarking?

Fishpan · 6 May 2017 at 17:22

I use adaptive voltage in the ASUS in combination with EIST, speedstep etc. Obviously you will have to do a lot of stress testing and determine you final voltage vs clock speed first. Then introduce the adaptive voltage which is the max voltage you want when CPU is running OC'd with the turbo multiplier. I have an innate fear of temperatures wrecking electronics ever since I lost my R9 290X to excess heat and will do anything I can to bring temps down including this method.