Inquiry on OC - 9700K north of 5.1Ghz

[Alcatraz1015]

I have my new 9700K successfully OC’d to 5.1Ghz @ 1.31v (it’s mated with an MSI Z390 Edge mobo and a Noctua NH-U12A tower cooler). During Cinebench R15 highest temp was a quick spike to 79c on one core (scored 1666). During PassMark CPU tests hit 80c, again a quick spike on one core (scored around 20,300). I have a Corsair H115i Pro 280mm AIO on the way. I know it’ll shave some Celsius on the top end.

 

This was my first OC. What I’m curious about is 5.2Ghz or higher. Hitting 5.1 @ 1.31 seemed like a pretty solid feat compared to other 5.1 results I’ve seen. What’s the worst that could happen if I tried for say 5.2Ghz @ ~1.36v?

 

Any OC knowledge will be welcome.

[Levent]

Nothing? worst case scenario is setting up your bios settings all over again (only if you need to clear cmos). Go as high as you want as long as you can keep it stable (Meaning WHEL errors what so ever) and cool.

[Alcatraz1015]

3 minutes ago, yaboistar said:

just the basics here but how long are you testing for? CBR15 is a pretty short test these days and isn't really enough to confirm stability for anything more than a minute or so at full load. I'd suggest using Aida64's stress test for at least an hour to put it under load for longer to see if that cpu is truly stable where it is now before looking to push it onwards and upwards

 

 

worst case when pushing up clocks and small voltage bumps is you need to clear cmos. 1.36 isn't going to nuke the die, 3.36 might lol

lol @ 3.36v

 

Yeah so my first test was R15 which like you said lasts for just a minute maybe. I did run it three times... but then I peg it with PassMark’s CPU bench which lasts about 8-10 minutes maybe? But then I was following that up with a few game benches (which are obviously more GPU oriented...) So yeah didn’t peg it for an hour for sure. 

 

I was planning on purchasing AIDA 64 just for that purpose. So now I’ll for sure do that. Educate me on CMOS please? Is resetting CMOS only necessary if I experience instability in an OC attempt or must that be done each new attempt regardless? Because I definitely just plugged in 5.1Ghz @ 1.36v and lowered voltage from there in between each test without touch CMOS.

[Alcatraz1015]

12 minutes ago, yaboistar said:

cmos reset is for when your OC settings aren't even bootable due to instability and you can't get into the bios to change them

I’ve researched it a little, but if worse comes to worst and I need to reset CMOS, hopefully my RAID array (which I have Windows installed on) will be fine. Seems like I just need to re-enable the raid controller after a CMOS reset and it should detect my array. 

[jones177]

I run one of my i7 8086ks at 5.1ghz and there are 2 resounds I don't run it at 5.2.

The first is that there is no gain in the overclock on my CPU except a better Cinebench R15 single core score. 

The second is that it puts the idle temps into the high 30s low 40s with my cooling solution so it is unacceptable to me.

 

Fortunately Gamers Nexus fully tested my CPU and they found that the only real performance gain after 5ghz was at 5.35ghz. Since that is way past my cooling solution there in no pint in me going further.

 

I suggest you do some research to see if it is worth it when it comes to performance. It is not on my CPUs, but it may be on yours.

 

 

 

 

[Alcatraz1015]

7 minutes ago, yaboistar said:

most modern uefi bios' have a feature to save the current bios config to a profile. i suggest making a profile with the cpu at stock speeds and voltages and your raid array etc set up so that you have a baseline to revert back to if you need to clear cmos to get the machine to boot again. also helps with not having to set fan speeds etc up all over again

Nice. Yeah I know I saw options to set up several profiles. I’ll take advantage of that. 
 

Do you have any experience with overclocking RAM? I have a nice kit. I use XMP. I don’t really want to mess with it because I only have passive cooling for it but in the off chance it’s fine to try for an extra 100Mhz or something without worry... then I’d like to take advantage of it. 

[Alcatraz1015]

11 minutes ago, yaboistar said:

you can usually tighten down 1 step in timings without needing to add much, if any, voltage. windows has an onboard memory checking tool (type windows memory diagnostic into the start bar) but you can also use a bootable USB of memtest86

 

ram overclocking doesn't give massive performance bumps on intel (unlike ryzen where ram speed defines certain speeds within the cpu package) so i'd personally leave it on XMP unless you're looking to extract every ounce of performance possible

 

i've found that going up 1 step in ram speed is usually pretty easy (3000 to 3200, 3200 to 3666 etc) and you'll only have to add a little voltage. your latency WILL increase as speed goes up so don't try and fight that. if you're stepping speed up a little then step the latency up by 1 and work back from there, if you're going, say, from 3000 to 3666, step it up by 2, maybe 3 and work back from there. but - memory overclocking does carry the risk of "silent" errors that even memory testing suites won't always detect but can cause kooky behaviour when you're actually using the machine, so personally I suggest that you just go up one step in memory speed and play with the latency to get it to work as going too far in speed, too tight in timings or too much/little voltage can all have adverse affects - seeing as intel doesn't much mind ram speed anyway there's no need to shoot for the moon with it

Cool deal. I think I’ll just let XMP do it’s thing and call it good. 
 

I appreciate all of your help. It was all very valuable info and I’ll use it going forward.