Maximus XI Hero & i9 9900k oc

[djmarks]

Hi all. I've been using some info from here to oc my pc.
I've used x-rated's recommended settings from a previous post ....

(EXTREME TWEAKER:
AI Overclock Tuner.................................XMP I
SVID Behavior......................................Best-Case Scenario
1-Core Ratio Limit.................................50
Max CPU Cache Ratio................................50
CPU Core/Cache Voltage.............................Adaptive Mode
Additional Turbo Mode CPU Core Voltage.............1.350


EXTREME TWEAKER -> EXTERNAL DIGI+ POWER CONTROL:
CPU Load-line Calibration..........................Level 6
CPU Current Capability.............................170% (or as much as possible)

EXTREME TWEAKER -> INTERNAL CPU POWER MANAGEMENT:
Long Duration Package Power Limit..................4095
Package Power Time Windows.........................127
Short Duration Package Power Limit.................4095
IA AC Load Line....................................0.01
IA DC Load Line....................................0.01)


and got a nice stable 5.1 oc. I've been dropping the Additional Turbo Mode CPU Core Voltage down by 0.1 then stress testing with XTU but I'm not sure if I'm reading the readings right.

When stress testing with XTU, I'm reading from cpuid hwmonitor the vid voltages as 1.329v with max value of 1.340v
VCCIO = 1.328v
Sys Agent = 1.216v
VCORE = 1.270v
PCH = 1.0565v
VCCPLL_OC = 1.216v

CPU-Z shows core voltage as 1.144v

Although with my limited knowledge, and with the powers of deduction, I'm assuming the VID voltages in cpuid hwmonitor is related to adjusting the Additional Turbo Mode CPU Core Voltage as my initial value was 1.350 and I witnessed that value in cpuid hwmonitor VID values.

Question is though, why is it refusing to go lower than 1.329 when I've it set to 1.310 in bios?

Edit - just noticed that VCORE in cpuid hwmonitor the max value is 1.314v. Is this due to my adjustment in bios?

Extra edit - I'm getting a nice max cpu temp of 65c - would the thought of pushing to 5.2 be too much?

Edit again - Scrath the above. The stress tests I ran last times were for 10mins and they all went ok. I then tried to run then for 30mins and they failed unmtil i returned voltage to 1.34. 10mins into a 30min test now and so far so godd!

[x-rated]

your vcore should fluctuate from ca. 0,7V to your desired value in bios
dont mix vid and vcore values - vid is what cpu would like to use, vcore is what it gets
from what i see try to adjust vccio and system agent voltage manually - you can try default values for 3,2 GHz (0,95V IO, 1,05V SA), then you can test stability in y-cruncher benchmark quite fast settings 0, 1, max. available memory

[djmarks]

your vcore should fluctuate from ca. 0,7V to your desired value in bios
dont mix vid and vcore values - vid is what cpu would like to use, vcore is what it gets
from what i see try to adjust vccio and system agent voltage manually - you can try default values for 3,2 GHz (0,95V IO, 1,05V SA), then you can test stability in y-cruncher benchmark quite fast settings 0, 1, max. available memory

Thanks x-rated!!

I've just had to up the voltage again as when I do an extended stress test (30mins) it has failed on 1.32, 1.33 and now 1.34. So I'm back to 1.35 at 5.1 and will run an extended stress test on this for a while in XTU.

Hoping this will be stable as I'd liek 5.1

[djmarks]

So ran many tests and my cpu doesn't like 5.1ghz much so I've dropped down to 5ghz and ran a test at 1.34v on prime for 40mins and everything went ok.

I'll run it for longer when I get a chance but for now, but for now my stress test will be (trying) to play a game of fo76. That games' also a stress test for me - will the server dc, will my mods mysteriously disppear from stash box etc? lol

[rjbarker]

Dont be deterred....according to numbers by Silicon Lottery only about 50% of CPU's are stable at 5 Ghz...drops to something like 17% for 5.1Ghz and 3% for 5.2 Ghz......take what you see on many of the OC'ing Forums with a big grain of salt....there appears to be a LOT of folks claiming stable @ 5.1 - 5.2 with there 9900K's ......as you saw a 5 min stress test isnt a 45 min stress test ....

[Silent Scone@ROG]

As above, 5.1. and 5.2 is still considered a lottery win. My personal sample which is a tray CPU needs 1.34v for 5GHz (no AVX offset) and fails to do 5.1GHz below 1.4v.

Rule out memory instability by reverting to default, too. It can also help to reduce the uncore ratio.

[djmarks]

Thanks silent scone and rjbarker.
I've stayed with 5ghz at 1.34v as everything seems to be stable with this. If I drop the voltage any lower, it doesn't seem to make any difference as the vid voltage for each core still shows 1.349v under stress.

I also tried using the rog realbench stress test but that looked like it was cooking my cup with temps hitting 100c so I stopped that right quick lol.

My cpu is a tray jobbie too.

I've kept the memory at it's stock - 3200mhz and I'm happy enough with that for the time being.

Just would like to maybe get the voltage lower.

[rjbarker]

Thanks silent scone and rjbarker.
I've stayed with 5ghz at 1.34v as everything seems to be stable with this. If I drop the voltage any lower, it doesn't seem to make any difference as the vid voltage for each core still shows 1.349v under stress.

I also tried using the rog realbench stress test but that looked like it was cooking my cup with temps hitting 100c so I stopped that right quick lol.

My cpu is a tray jobbie too.

I've kept the memory at it's stock - 3200mhz and I'm happy enough with that for the time being.

Just would like to maybe get the voltage lower.

That all sounds great...if you can reach 5 Ghz with reasonable temps consider it "all good"....my chip reaches 5 Ghz "relatively stable"...not P95 Stable though...but gets thru the standard Realbench....Intel Extreme....Cinebench....etc etc.....

[Falkentyne]

Thanks silent scone and rjbarker.
I've stayed with 5ghz at 1.34v as everything seems to be stable with this. If I drop the voltage any lower, it doesn't seem to make any difference as the vid voltage for each core still shows 1.349v under stress.

I also tried using the rog realbench stress test but that looked like it was cooking my cup with temps hitting 100c so I stopped that right quick lol.

My cpu is a tray jobbie too.

I've kept the memory at it's stock - 3200mhz and I'm happy enough with that for the time being.

Just would like to maybe get the voltage lower.

Please IGNORE the VID.
VID is not vcore.
VID is based on the CPU cache ratio.
Don't believe me? Set the cache (RING) ratio to x36 and watch what happens.

[rjbarker]

Ive been overclocking for many years....and part of the problem is folks getting a chip...setting a fairly substantial OC....getting thru one or 2 benchmarks and suddenly they announce a stable OC......stick it in there Signature.....start telling others there settings (and go by my settings) when in fact they're not "really stable" at all....this misinformation runs rampant on these OC Forums and folks starting thinking "everyone" should be reaching X.XX Ghz .....

I remember getting my 980X Chip many years ago.....couldnt get it past 4.2 Ghz ( I think....) ...meanwhile according to all the OC Forums everyone was 5 Ghz OC Club.....take it all with a grain of salt....theres a reason there are no 5.2 Ghz 9900K Chips available at Silicone Lottery.......

If you can reach 5 Ghz all cores with x45 - x47 Uncore.....around 1.3v....thats awesome!!!!

Ive had my 9900K for maybe 7 weeks....but only have been home from work enough to play with it a few days....and can already see its going to be a 5 Ghz Chip......was hoping for 5.1 Ghz......but the added voltage isnt worth the 100 Mhz