Adaptive Mode and Cascade Lake-X

[tistou77]

Hello

No one has managed to use Adaptive Mode with a Cascade Lake-X ?
It would seem that it was very complicated, because of the table of VID too high with this CL-X

Thanks

[Imperfections]

Hello

No one has managed to use Adaptive Mode with a Cascade Lake-X ?
It would seem that it was very complicated, because of the table of VID too high with this CL-X

Thanks

Well yeah. Of course you can use adaptive with it but it will be too high. Offset within stability margin also won’t help much.
Care to share what is your highest vid at 4.8? Mine went up as high as 1.52 v on worst core which is maximum I guess lol.

[tistou77]

Well yeah. Of course you can use adaptive with it but it will be too high. Offset within stability margin also won’t help much.
Care to share what is your highest vid at 4.8? Mine went up as high as 1.52 v on worst core which is maximum I guess lol.

Really a shame that it does not work, you also use the manual Vcore (fixed) ?

[H-de]

adaptive is the only way with manually vcore the vcore always set to maximum even in idle/low-load

hopeful ASUS will fix adaptive voltage and voltage table with cascade lake-x someday, currently i have -0.085 with 4.7GHz, i could even use -0.125 but the system will be unstable in idle/low load

The best way, ASUS let us decrease the turbo voltage, currently we can only increase Turbo Voltage

[tistou77]

adaptive is the only way with manually vcore the vcore always set to maximum even in idle/low-load

hopeful ASUS will fix adaptive voltage and voltage table with cascade lake-x someday, currently i have -0.085 with 4.7GHz, i could even use -0.125 but the system will be unstable in idle/low load

The best way, ASUS let us decrease the turbo voltage, currently we can only increase Turbo Voltage

With Cascade Lake-X, did you manage to use Adaptive Mode (but with a higher value) ?

[Imperfections]

Really a shame that it does not work, you also use the manual Vcore (fixed) ?

Yes, for now with this platform override is the way to go.

1. Adaptive - Only work if you want to set vcore higher than vid but who would do such thing when overclocking? Intel set one of my core to 1.516v at 4.8 while it is prime small fft stable at 1.3v... Another usage would be if you leave it at stock or super mild overclock.

2. Override - Fixed voltage even at idle. Well on the bright side it improve stability with supposedly slightly lower vcore as you don’t need lots of guardband if you set avx offset which can’t be avoided with this platform except you want to gimp the performance severely just for the sake of not using avx offset.

3. Offset - It may (but I’m sure it will) introduce instability at lower clock ratio. Not ideal to be used with this cpu especially because Intel set vid outrageously high than necessary. Then again if you can use the vid then better off with adaptive (it has offset too).
In my experiment at lowest ratio which is 12, I can have offset as much as -0.200 however it is not stable at 30 base clock which will allow only -0.75.

Asus AI overclock using mixed between override and adaptive. It set all 46 ratio cores using override and set the 4 best cores for 47 and 48 using adaptive. Which is clever but still in terms of lowest vcore possible and to reduce the heat override everything is better.

So among all of the method I decided the best way is to use override and per core overclock. I optimised the vcore per core and set 14 cores to 46, 2 cores to 47 and 2 cores to 48. Basically maxed out turbo boost all the time. I’ve tried 48 all cores however it warm up my working room significantly eventhough the temp of the core itself is still under 100 and it forced me to reduce my AVX overclock or else it gets to hot. So I settle with that turbo boost all the time oc and I get to keep my avx-512 at 38 and avx-2 at 42 which is slightly higher than single core stock. If I insist to go with 48, I can’t even run my avx-512 at all core 38. So I guess this is ideal.

Care to share snapshot of your vid and temp when running cinebench r20 at stock? And also if your bios has the silicon point score (extreme encore has it) please let me know, mine is only at 68.

[tistou77]

Thanks for your feedback

I hesitate to take a 10980XE, but this story of Additive Mode slows me down a bit

Are you at 68° on the hottest core with Cinebench (4800mhz) ?

[H-de]

With Cascade Lake-X, did you manage to use Adaptive Mode (but with a higher value) ?

nope only lower

[tistou77]

Ok, thanks

[Imperfections]

Thanks for your feedback

I hesitate to take a 10980XE, but this story of Additive Mode slows me down a bit

Are you at 68° on the hottest core with Cinebench (4800mhz) ?

Which cinebench? No way in hell with Cinebench r20. At 48 all cores it went to 90s.
I don't test cinebench r15 anymore as it is not so relevant for todays standard. But definitely lower temp.

Then again my chip is not the greatest it only scored 68 point on Asus prediction. With other chip went upwards of 90.
Also note that my vcore is for prime stable. If I were to lower my vcore to pass cinebench only then probably can lower the temp to low 80s. But 68? Nah. Or do you mean 86?