Team Blue's 10900K Overclocks

[Grim5]

Paul has an interesting way to overclock the 10900k he just removed the pl1 and 2 and tau limiters and current, gave vcore an 0.6 offset bump and got 5.1 all core and 5.4 single (or is it two?) core:

https://youtu.be/NPk2lnCFoUA 22:00

it seems he used AIDA64 stresd test

Oh hold on he used an MSI motherboard...

When it's "single core" workload it's actually 2 cores that alternate between the load - I don't know if this is a window scheduler thing or if it's hard coded into the CPU, but Ryzen CPUs do the exact same thing when you run Cinebench single thread bench, it puts alternating load on 2 cores. With the 10900k I've read that it's always core 7 and 8 that gets used for single thread loads - I assume it's cause Intel bins their cores and so core 7 and 8 is the highest quality of the lot.

As for Paul, his review has some weird data like getting 20fps more than a 9900k in tomb raider and not a single other reviewer achieved the same results even with a all core 5.3ghz overclock

 

[FireTwo]

interesting way to oc I'll have to check this out!

 

[klumpo]

buildzoid overclocks the 10900K and ram 5,1GHz 4266mhz ram ring ratio 4500mhz linpack stable

Turn up thermal throttling a bit or else it can get unstable when the CPU clocks down (transient response) 2:00
linpack hits cpu, cache, memory controller and ram, prime95 not as much 3:00
trcd and trp on intel is linked will be the same 9:00
Don't stress test with Cinebench or RealBench 13:00
high memory oc sets very high auto system agent and vccio volt 16:00
vcore 1,405v (fine with much vdroop), system agent 1,380v, IO 1,280v for 5,1GHz 4266mhz ram linpack stable 33:00

https://youtu.be/WtucSUA5TOI

 

[Destination]

he achievs what with this out of interest?
is the don't test with cinebench due to instability?

 

[klumpo]

he achievs what with this out of interest?
is the don't test with cinebench due to instability?

I think he says that cinebench and realbench doesn't push the cpu enough.

I passed1 hour stress in RealBench but got BSODs in Fallout 4 and sure enough, Prime95 blend with avx2 instant error. So therefore I am interested in linpack and prime95 results

 

[Destination]

fallout 4 is a good test of anything, it could blue screen a red wall….

 

[deuse]

Hey guys.

Had my 10900KF for a few weeks on a Asrock Z590.
I have never seen it go past 4889 ghz running intel burn test at it's highest.

I thought it was auto 5.1-5.3
What am I doing wrong?
Thanks

edit. everything is on auto in the bios.

 

[kitfit1]

Hey guys.

Had my 10900KF for a few weeks on a Asrock Z590.
I have never seen it go past 4889 ghz running intel burn test at it's highest.

I thought it was auto 5.1-5.3
What am I doing wrong?
Thanks

edit. everything is on auto in the bios.

As long as you have very good cooling, enable MCE in your bios.

 

[deuse]

As long as you have very good cooling, enable MCE in your bios.

Thanks.

I wonder what asrock call it.

 

[darket]

Hey guys.

Had my 10900KF for a few weeks on a Asrock Z590.
I have never seen it go past 4889 ghz running intel burn test at it's highest.

I thought it was auto 5.1-5.3
What am I doing wrong?
Thanks

edit. everything is on auto in the bios.

Boosting to 5.1-5.3 is only in light loads and only with TVB enabled . And that UP TO 5.3 is not guaranteed, it depends on the SP rating of the cpu, the temps and voltage tables.

And may I ask as to why you are running IBT ? Testing for stability with a crazy power virus on a chip that will gladly draw up to 300w won't gain you anything other than a lot of heat and a degraded chip. If you want to test stability use Prime non avx, Occt Large date set , realbench or cinebench etc etc. Testing a modern Intel cpu with insane AVX loads is like holding your car at the red line for hours on end. Its unrealistic and doesn't do it any good.

 

[darket]

As long as you have very good cooling, enable MCE in your bios.

That won't have the desired effect.

 

[deuse]

Boosting to 5.1-5.3 is only in light loads and only with TVB enabled . And that UP TO 5.3 is not guaranteed, it depends on the SP rating of the cpu, the temps and voltage tables.

And may I ask as to why you are running IBT ? Testing for stability with a crazy power virus on a chip that will gladly draw up to 300w won't gain you anything other than a lot of heat and a degraded chip. If you want to test stability use Prime non avx, Occt Large date set , realbench or cinebench etc etc. Testing a modern Intel cpu with insane AVX loads is like holding your car at the red line for hours on end. Its unrealistic and doesn't do it any good.

It never went above 70c

I went with 5.2ghz on all cores @1.35. But in 4K gaming it never made a difference.
So it's back to stock

 

[humbug]

superPI anyone? 32M
https://ibb.co/FwwhfHq

I'll take your 6m 30.6s and give you 6m 28.4s

5800X locked in low power mode at 4.7Ghz, i'm experimenting with it, 94 Watts total package power, 1.28v at 4.7Ghz all core locked... HA! this thing is ####### nuts