Anyone has overclocked an i5-12600k?

[Lunacy]

I have an i5-12600k with all P-cores and E-cores OC'd to 5Ghz and 4 ghz respectively, and CPU Vcore at 1.380V

My CPU cooler is an arctic liquid freezer II 280mm top mounted

And my case is a phanteks p500a w 3x 140mm intake front fans and 1x 140mm rear exhaust fan

 

My idle core temp is around 21C. CPU package temp is around 27C. Core max temp is 23C

When I run cinebench r23 CPU multi core test, the CPU package and Core max very quickly shot up to 100c and stayed there. whilst the Core temp touched 100c for a second and immediately hovers around 80c. I run this using win 11 as the OS.

 

What's the problem here? my component or is it a misread from hwinfo

[Shimejii]

I mean coolers are designed to react to heatloads, so when you suddenly spike it up the fans have to ramp up and the water pump if its not already at 100% has to spool up. That can take a few seconds to react. 

 

There really isnt a problem if it ends up at 80c. 1.38v is a bit high, what are your scores from leaving it on auto versus the manual OC?

[Lunacy]

24 minutes ago, Shimejii said:

I mean coolers are designed to react to heatloads, so when you suddenly spike it up the fans have to ramp up and the water pump if its not already at 100% has to spool up. That can take a few seconds to react. 

 

There really isnt a problem if it ends up at 80c. 1.38v is a bit high, what are your scores from leaving it on auto versus the manual OC?

I have only finished the single core test. 1959 points with avg CPU package and core max of around 49C, is that good?

 

[RONOTHAN##]

23 minutes ago, Lunacy said:

CPU Vcore at 1.380V

There's your problem.

 

Alder Lake is very thermally dense, and it doesn't really scale past ~1.3V at ambient since it just gets way too hot, so unless you've got a very droppy load line set, 1.38V is likely way too much. So yes, your CPU is probably thermal throttling. Lower your voltage or set a proper LLC (most Z690 boards actually do best when a very droopy LLC is selected, I know on MSI boards level 6-7 tend to be the best for voltage regulation) and odds are performance will increase since the chip will stop thermal throttling. 

 

Also, just saying, R23 isn't really a good stress test. It runs on relatively unstable hardware without too much of a fight (my delidded 7700K, for instance, can run that test at 5.6GHz on air cooling). If you actually wanna stress the CPU for daily operation and have it actually be stable, you need to use different tests like Linpack Xtreme, Prime95 SmallFFTs, or OCCT. 

[Shimejii]

22 minutes ago, Lunacy said:

I haven't finished it because in hwinfo it says yes on core thermal throttling, and 100c on CPU package and Core max. should i just let it be and finish the test?

 

Should at least let it run and finish. Then turn off all manual OC and turn it to the defaults, see if the result is about the same, and check thermals. 12th gen is a VERY hot set of chips, Intel  is pumping a lot more power through these chips to get the numbers up there so.

[Marko Del Vechio]

1 hour ago, Lunacy said:

I have an i5-12600k with all P-cores and E-cores OC'd to 5Ghz and 4 ghz respectively, and CPU Vcore at 1.380V

My CPU cooler is an arctic liquid freezer II 280mm top mounted

And my case is a phanteks p500a w 3x 140mm intake front fans and 1x 140mm rear exhaust fan

 

My idle core temp is around 21C. CPU package temp is around 27C. Core max temp is 23C

When I run cinebench r23 CPU multi core test, the CPU package and Core max very quickly shot up to 100c and stayed there. whilst the Core temp touched 100c for a second and immediately hovers around 80c. I run this using win 11 as the OS.

 

What's the problem here? my component or is it a misread from hwinfo

This is just a bad interface between your cooler and the cpu.. Temps spike fast as there is no heat transfer between the cpu die to the cpu metal cover to the aio coldplate..
Older gen intels had to be de-lided to fix the issue.

Try re-mounting the cpu and aio and check for paste spread when you remove it. Might be just a bad termal paste.

[Somerandomtechyboi]

3 hours ago, RONOTHAN## said:

R23 isn't really a good stress test. It runs on relatively unstable hardware without too much of a fight

^^^

As a beginner i thought r15 was enough to test stability but after a bunch of crashing and f@h just nuking itself randomly i decided to test p95 only to find out that my giga g31m s2c was unstable at 475 fsb (incapable of running rams high due to insufficient nb volt that needs vmod to be raised) so dialed it down to ~435 and it now works fine, that was using large ffts which is useful for when you are ram ocing (basically on old fsb/bclk platforms raising fsb = raise cpu freq but also ram freq), smallest ffts is great for burning the cpu and testing your cooling system but iirc there are better ways than p95 for cpu stability test due to p95 smallest ffts workload being unrealistic and is literally a burn test, i think ltt made a vid on stress testing software a few months ago

 

3 hours ago, RONOTHAN## said:

my delidded 7700K, for instance, can run that test at 5.6GHz on air cooling

Spoiler

*cries in 5.15g e8400 and x5660 bios boot at 1.7v

 

Golden sample? What kind of aircooler is even capable of that?

[Guest]

1 hour ago, Somerandomtechyboi said:

 

  Reveal hidden contents

 

 

Golden sample? What kind of aircooler is even capable of that?

Has to be a typo bro.

 

Now way 5.6ghz R23 through benchmate on air 7700k

I'd request an unedited video, but we know it won't be supplied.

[Somerandomtechyboi]

4 minutes ago, Guest 5150 said:

Now way 5.6ghz R23 through benchmate on air 7700k

Yea that makes sense, also thought it was kinda ridicolous since this is pretty much amd fx territory for overclocking

[Guest]

29 minutes ago, Somerandomtechyboi said:

Yea that makes sense, also thought it was kinda ridicolous since this is pretty much amd fx territory for overclocking

Not even FX chips without chilling. He must have meant 5.2ghz. That's about the top you're gonna do on air cooling with a 7700K if it even makes it that far. 

[RONOTHAN##]

5 hours ago, Guest 5150 said:

Has to be a typo bro.

 

7 hours ago, Somerandomtechyboi said:

Golden sample? What kind of aircooler is even capable of that?

It was a typo, it does do 5.3GHz in R23, and I have gotten 5.4GHz in the winter in my basement, but in tests like Linpack and Prime95 it can't do more than 5.0 without running into thermal issues. 

[Guest]

1 hour ago, RONOTHAN## said:

It was a typo, it does do 5.3GHz in R23, and I have gotten 5.4GHz in the winter in my basement, but in tests like Linpack and Prime95 it can't do more than 5.0 without running into thermal issues. 

Nah man I figured as much. Even 5.3ghz is pushing it.

 

My 8700K maybe 5.4ghz with HT off, bench able, but I wouldn't try prime95. That would overcome the loop quickly.

 

Edit, I mean chilled water btw. 

5ghz I can have 1.360v it's reasonable for prime.

[gamal]

i have 12600k and b660m mortar max dd4 oc by bclk to 111mhz all p core at 4.996ghz and e core to 3.995ghz with 4x8 crucial ballistix 3600 cl16 oc to 3885mhz with bclk as well and seems to be pretty stable with xmp settings

on cinebench r23 10 min stress test load vcore 1.204v peak power cons 164 was watts max recorded temp 83 dgrees

max recorded vcore when single threaded cinebinch r23 test running was 1.31 max recorded turbo boost was 5.439ghz

ran prime 95 small ffts peak power cons was 213 watts with vcore 1.176 after couple mins it thermal throttled and maintained 198 watts

keep in mind i have noctua redux cooler not that high end of a cooler and it held to itself due to whats considered low voltages 

i might have hit the lottery on that xd

if anyone wants the bios sittings reply to that