Baffled with my Core i9-9900K overclock (Asus Auto OC > Manual VCore)
EDIT: Here's where I got to, tl'dr I have a bad silicone lottery: https://www.overclock.net/forum/28277336-post29.html
I've recently upgraded from my long-standing Asus Sabertooth Z77 with a de-lidded Core i7-3770K (running at 4.5GHz) to an Asus Maximus Hero XI Wi-Fi with a Core i9-9900K.
As I learned 6+ years ago, it's always good to test the system on stock first; and I did that with the Core i9 system - Upon installing Windows I hit up the BIOS, tinkered with a few settings and put it on an old non-AVX built P95. Then, I ran the test for around 2hrs without any throttling or problems; temps were in the high 90s and I presumed the system was overshooting my voltage in order to guarantee a 5GHz OC (Asus MultiCore Enhancement set all 8 cores to 5GHz) - CPU-Z reported a VCore of 1.252V; HW reported a max of 1.332V-1.368V.
Below is a pic of my BIOS; essentially most things were on stock/auto; XMP1 + the Asus OC.
After testing my 'stock' system, I decided to start my OC'ing journey. I watched a multitude of guides, read a bunch of them - most had contradicting information, so I stuck with what I knew best - the same sort of settings that applied to my core i7-3770K (almost identical to a few guides I read of the i9-9900K).
Long story short, 5GHz ran into plenty of BSOD/auto-restart/P95 crashing but the PC staying in desktop; and right now I'm a bit stumped.
To put it into perspective, I ran manual VCore to 1.4V and it was still crashing (let alone the sky-high temps, which put me off 5GHz all together).
Even more of a mystery to me is my VCore readings in CPUZ and HW: If in the BIOS I set 1.32V, in the software programs without ANY load, I get around 1.28V; add 100% P95 load on it and with LLC Level 6-7 it drops to around 1.25V.
Okay, so what are the questions?
I want to know the following:
1. Why is my BIOS Vcore drastically different from my CPUZ/HW readings?
2. Why is my manual OC struggling, while the Asus auto OC seems to run fine in P95?
3. What can I realistically achieve with my Noctua D15 air cooler with this chip? I don't need 5GHz, but it would be a nice sweet spot :)
4. Why do I need 1.31V to run 4.8GHz?
5. Am I doing anything wrong with my settings/BIOS etc?
Just before answering, I've attached all my BIOS screenshots, below for my 4.8GHz 'OC'; it's currently running P95 as I type this - it failed at 1.3V after 2.5hrs.
FYI - I'm now running the latest P95 version and have disabled AVX2 and AVX1 in the custom field.
Any help would be greatly apprecaited :)!
look in HWInfo for a "VR VOUT" sensor. Some boards don't have it but if you do then that's what you should be looking at.
Software readings always have some level of inaccuracy but the VR VOUT sensor is a direct readout from the VRM and should be accurate enough.
vDroop is normal and it's a good thing (albeit annoying), use LLC to limit it if it's excessive but you shouldn't try to completely remove it. Overshoot is more of an issue than needing slightly higher voltage to offset the vDroop.
As for the readings - I can only go based on CPU-Z and HW.
VDroop is normal, I know, and I do know about LLC; but running Lvl 8/8 LLC isn't advisable - due to the extremely hot temperatures; LLC also accounts for drops in voltage when faced with a load, not in an idle state (correct me if I'm wrong).
So, my LLC needs to be set based on the drop I have; when I'm stress testing - however - that doesn't answer my initial question as to why my vcore is different in the BIOS vs CPUZ in the idle state (let alone when I'm stress testing); I've been digging since posting this question and have found some enlighting answers; but I'm not sure this directly corresponds to what I'm seeing. Any ideas?
as to why its reading differently, are you sure its the newest CPUZ? have you tried any other HW monitoring tools? CPUZ is mostly a validation tool. It's sensors are meh.
as to manual OC. I tried OC'ing my 9700k just like I would've my 4930k, & it didn't go well. set voltage & go just didn't seem to work as well as it used to. Instead, I used offset to tune it, and left LLC on "medium" or setting 3 of 5. Some folks set it even lower.
is it the best OC in the world? probably not. Does it do better than my MSI's AI OC? by a few degrees and 300mhz? yes.
here's a 9900k OC done via offset with LLC low.
running the highest LLC isn't advisable because of overshoot, the temperature is irrelevant. vdroop makes overshoot a non-issue, if you set 1.25v and are getting 1.15v under 100% load then even an unrealistic level of overshoot isn't pushing your CPU voltage that'd be harmful.
overshoot isn't detectable in software, just because you're not seeing spikes to high vCore doesn't mean it isn't.
lengthy but good information, watch it at 1.5x speed or something because he takes forever to go through some of it, lol.
most of the boards that don't have a VR VOUT sensor have accurate vCore, but if you have VR VOUT then you should go by that. Otherwise just go by why HWInfo is saying. My board shows ~0.15v higher than what I'm actually using compared to VR VOUT.
A friend and I both recently purchased the exact same set-up. Gene XI M/board, CPU 9900ks and ram (as in my sig) both with custom loops.
Mine runs 5.2GHz no problem and might be able to go higher but his struggles to get past 5.0GHz. He has been having similar issues as you with the blue screens etc if he tries to pass 5.0, even with pumping the volts and trying higher llc. We settled on 6 for llc and raising the idle voltage, as higher llc just seemed to make it worse. I even tried saving a .cmo file to usb with "safe" settings on my pc and sent it to him, but that didn't work either.
I get a higher reading in HWinfo than cpu-z also but not much.
What @skupples is alluding to makes good sense and I might look into it for my mate
Maybe with the advent of the 9900ks chips it's just harder to get a decent 9900k?
I can't lower my LLC as it'll only create a larger VDroop :O
And yes, latest CPUZ and HW
I'm at 7/8 LLC and I'm still off my VCore...in IDLE! At this rate, I feel a fresh BIOS reset might be the only solution - as I can't seem to find any logic behind this all.
FYI: BIOS is on the latest provided by Asus.
As we speak my 4.8GHz 'OC' has been running for hrs straight; no negatives so far - so it looks like I'm near stability on 4.8GHz...at 2.31 Vcore. If that's the case, there's no way I'm pushing past 4.8Ghz as I'll get thermal throttling and/or too much of a VDroop.
I must have gotten unlucky or done something wrong in the BIOS; really can't put my finger on it.
Maybe setting a higher idle voltage with llc 6, or even 5 will create smaller transients and not drop into blue screen territory when load comes on. The higher idle voltage won't be creating too much heat without the load/current draw?
Correct me if I'm wrong, I'm trying to work this out for my mate also :)
BTW on these boards the spikes arent the issue contrary to what some are telling you except at very high bios voltages like >1.45v. The spikes happen at load release (low current) and won't become an issue unless they exceed 1.5v on an oscilloscope capture. Its the dips that are the problem.
|All times are GMT -7. The time now is 12:17 AM.|
Powered by vBulletin® Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.
User Alert System provided by Advanced User Tagging (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.
vBulletin Security provided by vBSecurity (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.