Oc'd 8700k can handle cinebench and Intel extreme utility stress test, but blue screens on heaven benchmark

[LaboonTheWhale]

So I delidded and Liquid metal my 8700k and Temps are great but when I benchmark it seems to blue screen on heaven benchmark. This obviously doesn't put as much of a load as cinebench or Intel stress (100% load} but for some reason blue screens on heaven.

 

The error is clock_watchdog_error so looks like I need to turn up the voltage for that.

 

But why does it handle 100% but not gaming load

 

[hconverse02]

Well what are your temps, clocks and voltages?

[Jurrunio]

handles non-gaming load probably because of AVX offset. That lowers your multiplier when AVX is used, which reduces your clocks.

[LaboonTheWhale]

9 hours ago, hconverse02 said:

Well what are your temps, clocks and voltages?

Oh yeah that's important. 5ghz 1.31v. Z370 strix itx with a noctua u14s push pull. Temps idle at 34. 68C during cinebench. 58C during Intel extreme. Around 50 C during heaven.

 

9 hours ago, Jurrunio said:

handles non-gaming load probably because of AVX offset. That lowers your multiplier when AVX is used, which reduces your clocks.

When watching hardware info 64 app it shows my clocks being synced across the board at 4.9xx or 5ghz.

[hconverse02]

3 minutes ago, LaboonTheWhale said:

Oh yeah that's important. 5ghz 1.31v. Z370 strix itx with a noctua u14s push pull. Temps idle at 34. 68C during cinebench. 58C during Intel extreme. Around 50 C during heaven.

That's really good.  Try 1.32v or 1.325v and see what happens EDIT: It could just be an issue with heaven though.  How is it in game testing? Any bsods? 

[LaboonTheWhale]

Just now, hconverse02 said:

That's really good.  Try 1.32v or 1.325v and see what happens

I believe when I tried 1.32v it blue screen. I didn't try the latter yet

[TheGlenlivet]

Could also be something with the PCI lanes.  The other tests aren't using the GPU but Heaven does.  MORE VOLTAGE!!!!

[.Apex.]

1 minute ago, LaboonTheWhale said:

I believe when I tried 1.32v it blue screen. I didn't try the latter yet

probably because with Heaven you're stressing your GPU too and its pulling power from the PSU, so there's tiny voltage fluctuations or something, bumping up the voltage a bit more will give you more headroom for those fluctuations so you wont blue screen.

also there's many other things in BIOS that will get you stable at that voltage, but seriously the easiest thing to do is bumping up a bit more voltage, 1.325 or 1.33.

[LaboonTheWhale]

Just now, TheGlenlivet said:

Could also be something with the PCI lanes.  The other tests aren't using the GPU but Heaven does.  MORE VOLTAGE!!!!

Higher Temps lol.i have an 8600k that can handle it at that voltage so maybe I'll just use that instead. Delid and all.

[porina]

Different applications use a different mix of instructions, so you do need to try a blend before declaring any OC as potentially stable. If you get repeatable instability, you have two choices: increase voltage or lower clock, assuming heat is controlled adequately.

 

For comparison, I have an 8350k which needs over 1.4V to be bench stable, so I think you have headroom to try higher voltages.

[TheGlenlivet]

1 minute ago, LaboonTheWhale said:

Higher Temps lol.i have an 8600k that can handle it at that voltage so maybe I'll just use that instead. Delid and all.

I'd be comfortable with another 10 degrees (78) before I was concerned and honestly even that is probably conservative.  I bet you got another .25v in there...

[LaboonTheWhale]

13 minutes ago, TheGlenlivet said:

I'd be comfortable with another 10 degrees (78) before I was concerned and honestly even that is probably conservative.  I bet you got another .25v in there...

I'm downsizing to sffpc using a thermalright axp-100RH so every degree counts lol.