- Joined
- Oct 2, 2019
- Messages
- 79 (0.05/day)
hello good day. i have a bunch of QTJ1 cpu's here to play with along with a maximus x hero z370 motherboard to overclock them with.
The QTJ1 and QTJ0 is a mix between a I9 9980hk And I9 10980hk. they are all produced right in the middle of the 2.
the I9 9900hk is pretty much a I9 9900k binned to be able to run at lower voltages as a whole. but if properly cooled (which it is) you should be fine running and exceeding desktop voltages. given it is actually properly cooled.
the versions i got here have a improved design and i am able to run 3400 mhz ram. i cannot run higher due to not owning 3600+ mhz stick currently.
they are direct die mounted to a artic freezer 2 420 and i never got them above 80c before. even at 300+ watts.
these chips tend to want more significantly more voltage than typical desktop chips when going above 4.6 ghz all core..
but in terms of performance they are equal to 9900k desktop chips.
the chip is 4.7ghz stable at small fft's at 1.45 core volts while keeping low 70's
i can run cinebench r23 fine at 5 ghz. (even at 5.1) pulling around 245/250 watts. (low 70's temps)
however i can't get that stable on Small fft's at 5ghz at any voltage the system would instantly BSOD/hang when you click start. not even giving a second to update HWINFO64 stats.
if i run core cycler/physics based stuff/any other cpu related benchmark. the system would pass with no issues.
i have been using LLC level 4 all the way up to max level 7 but no luck.
if i set a VRM current limit. the cpu will still do 5ghz during cinebench runs but itll run alot lower during FFT's due to the extreme power draw (i've got consumption up to 300-350 watts before)
i feel like that is not a proper way of trying to get a stable high OC though?
after some research i read that reducing LLC as much as possible can help with transient spike related crashes.
and i feel like those transient's from Smallfft's starting up is what i am suffering from. after all. the load is signficiantly more power hungry than anyhing else...
so after finding that out i have tried LLC1/2/3 and it seems to indeed give different behaviour..
i have to set the core voltage to max 1.54 volts so far and during load. this drops to 1.43/1.45 due to the low LLC.
i have noticed SmalFFT's actuall runs for a little longer before crashing in this situation. which means i might be a bit low on core voltage.
however during a single core benchmark the voltage does stay around 1.5v.
accoridng to intel documentation. a idle voltage of 1.52v is fine.
and even intel engineer's themselves seem to say that running chips at higher voltages is fine. given you can cool the chip
source : https://www.tomshardware.com/features/inside-intels-secret-overclocking-lab/7
they also state that mid 70's are fine. and anything above 80c at higher voltages is a dangerzone.
would it be fine to increase the core's static voltage to higher than 1.54v with LCC off?
and what about the single core workload's still being at such voltages?
any other options i could try to improve the possibly transient instability?
thank you in advance for your advice!
The QTJ1 and QTJ0 is a mix between a I9 9980hk And I9 10980hk. they are all produced right in the middle of the 2.
the I9 9900hk is pretty much a I9 9900k binned to be able to run at lower voltages as a whole. but if properly cooled (which it is) you should be fine running and exceeding desktop voltages. given it is actually properly cooled.
the versions i got here have a improved design and i am able to run 3400 mhz ram. i cannot run higher due to not owning 3600+ mhz stick currently.
they are direct die mounted to a artic freezer 2 420 and i never got them above 80c before. even at 300+ watts.
these chips tend to want more significantly more voltage than typical desktop chips when going above 4.6 ghz all core..
but in terms of performance they are equal to 9900k desktop chips.
the chip is 4.7ghz stable at small fft's at 1.45 core volts while keeping low 70's
i can run cinebench r23 fine at 5 ghz. (even at 5.1) pulling around 245/250 watts. (low 70's temps)
however i can't get that stable on Small fft's at 5ghz at any voltage the system would instantly BSOD/hang when you click start. not even giving a second to update HWINFO64 stats.
if i run core cycler/physics based stuff/any other cpu related benchmark. the system would pass with no issues.
i have been using LLC level 4 all the way up to max level 7 but no luck.
if i set a VRM current limit. the cpu will still do 5ghz during cinebench runs but itll run alot lower during FFT's due to the extreme power draw (i've got consumption up to 300-350 watts before)
i feel like that is not a proper way of trying to get a stable high OC though?
after some research i read that reducing LLC as much as possible can help with transient spike related crashes.
and i feel like those transient's from Smallfft's starting up is what i am suffering from. after all. the load is signficiantly more power hungry than anyhing else...
so after finding that out i have tried LLC1/2/3 and it seems to indeed give different behaviour..
i have to set the core voltage to max 1.54 volts so far and during load. this drops to 1.43/1.45 due to the low LLC.
i have noticed SmalFFT's actuall runs for a little longer before crashing in this situation. which means i might be a bit low on core voltage.
however during a single core benchmark the voltage does stay around 1.5v.
accoridng to intel documentation. a idle voltage of 1.52v is fine.
and even intel engineer's themselves seem to say that running chips at higher voltages is fine. given you can cool the chip
source : https://www.tomshardware.com/features/inside-intels-secret-overclocking-lab/7
they also state that mid 70's are fine. and anything above 80c at higher voltages is a dangerzone.
would it be fine to increase the core's static voltage to higher than 1.54v with LCC off?
and what about the single core workload's still being at such voltages?
any other options i could try to improve the possibly transient instability?
thank you in advance for your advice!