I5-8600k Overclock Opinions

[grunto]

So I managed to run my 8600k at 5ghz on 1.33v using Prim95 26.6 for 10-15 minutes without crashing, and it floated around the mid to high 60's. Some people think it needs to run Prim95 for 24hrs before its considered stable, do I really need to push the CPU that hard for that long?

Also, I could probably get 5.1 at 1.36v, and 5.2 at 1.38/1.39v. Do you guys think I should push for those, or is 5ghz enough? I game mostly at 1440p on a 144hz monitor, mostly playing League, StarCraft 2, Gears 4, Civ 6 and SimCity.

Hardware:
I5-8600k
Asus Rog Strix Z370E
Corsair H100i V2
Sapphire R9 Fury
16gb Corsair Dominator Platinum 3000mhz
Corsair CX650m
Corsair Carbide 400c

 

[WildCard999]

Those temps/voltage are quite good and you could probably go further if you really wanted to. As for the 24 hour test it's not needed, what you did is sufficient. For a awhile I would just keep an eye on temps and log any BSOD/errors if they happen.

 

[Roland Of Gilead]

well, if you want to find out if the CPU is stable, yes, overnight is a good idea.

You wlll be very lucky to hit 5.2 at those voltages. FOr each increase in clockseed/multiplier, you will need an even bigger bump on voltage, so at 5.2 i suspect you will be pushing over 1.4v.

Stick with 5ghz. Your good. After that, it's a case of diminishing returns in heat generated v's marginal increases in mhz. Just not worth it for me.

 

[Roland Of Gilead]

Those temps/voltage are quite good and you could probably go further if you really wanted to. As for the 24 hour test it's not needed, what you did is sufficient. For a awhile I would just keep an eye on temps and log any BSOD/errors if they happen.

Why is a longer test not needed. What is the justification for that? So with 10-15 mins testing your happy that your CPU is stable?

 

[WildCard999]

Those temps/voltage are quite good and you could probably go further if you really wanted to. As for the 24 hour test it's not needed, what you did is sufficient. For a awhile I would just keep an eye on temps and log any BSOD/errors if they happen.

Why is a longer test not needed. What is the justification for that? So with 10-15 mins testing your happy that your CPU is stable?

Well you could go a bit longer for temps/stability but I would say an hour max as those stress test put a unrealistic load on the CPU that you'll rarely see in gaming. And from the testing I've done with Prime95 26.6 Small FFT's the temps I go at 10 min were practically identical as the temps I saw at 1 hour so it seems kind of pointless to keep pushing it further. Also Prime95 is more for temps and not stability, Asus Realbench would be better for that however I don't use so I'm not sure how long you'd test with it.
http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

 

[Roland Of Gilead]

i would agree that the load isn't everyday/gaming load, but the whole point of stressing specially with Prime/IBT or whatever app is to get max voltage/temps. So 10 mins or even an hour doesn't cut the mustard for me. That my personal preference. I like stability to be 99.9%

I would agree that in most cases the temps should even out after about 10 mins, but it will increase slowly over the first houe. At 10 mins it might settle at 75c (lets just say for arguments sake) but will still increase typically (and very slowly) for another hour and might get up to 76/77/78c. Thats my experience with Prime. So although I wouldn't run it for 24hrs. I run it overnight. Typically 11pm to 7am. If there are no issues i'm happy with that, and the implied stability. I don't use Realbench, never have. Prime is perfect for my needs.

Interesting to know your thoughts. Thanks for the explanation

 

[grunto]

Alright, I will run it at length then, just not 24 hours. Thanks you for your inputs.

 

[Roland Of Gilead]

Alright, I will run it at length then, just not 24 hours. Thanks you for your inputs.

no probs, you're welcome.

I think we are all on the same wave length, as are you.

You'e testing for stablility as you go, which is a good thing.

Keep in mind though, at 5ghz that chip is an awesome gaming chip. It will whack out the FPS like goodo

The difference with 5ghz, compared to 5.1/5.2 will probably not even garner just a FPS extra from GPU. At that speed, and with the R9 Fury, pushing the CPU that far is pointless. Be happy with the moderate OC, and keep temps and voltage low. Unless you are looking to get synthetic bench wins, or purely for the last 2-3 FPS you can eek out of the system, bumping clocks/voltage for an extra 200mhz is not worth it. You will see no benefit.

Whatever res you play at, and even if you upgrade monitor, and GPU, that CPU at 5ghz is plenty for driving any high end card out there, including a 1080ti or Titan XP, and/or any SLI set up too.

Enjoy your games, and report back with your insane FPS counts!

 

[grunto]

I ran Prime95 26.6 last night for about an hour, until it BSOD running 5ghz at 1.33v. I bumped the voltage up to 1.34v, which unfortunately pushed the temps from 67ish up to around 70° ± 2° at any given time, depending on the core. Only 1 core peaked at 74°. With that said, that core was usually around 72°.

With the second effort, the PC once again shut down at the 1 hour mark, but in this case the PC just went to sleep lol. Once I realized what happened, I ran it straight for another 3 hours, with zero issues, maintaining the previously stated temps.

Are these temps still viable? Do I still need to try for that 6hr mark, or is 3hrs at 1.34v enough based off how it handled 1.33v for an hour before BSOD? Either way, thanks to both of you for encouraging further stability testing regardless of duration.

 

[Roland Of Gilead]

absolutely, those temps are fine. I'm happy to push my CPU to 80c. But obviously that Prime Stressing temp. Normal everyday gaming use you shouldn't really push above 65c depending on the gaming load.

3hrs is solid enough. Its up to you really. Some people have a preference to up the speeds and not push the CPU to 80c+ because they don't want to leave performance on the table. My preference is max stability with min voltage and highest clockspeed. FOr others it's max speed, whatever voltage enables that speed, and then don't stress because they know they will BSOD or hit high temps and experience thermal throttling.

Different horses for different courses