• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

8700k throttling under realbench at only 4.6G

Associate
Joined
11 Dec 2013
Posts
257
hi,

just trying to get a very mild o/c going, basically MCE without the crazy voltages. i'm using a Maximus X Hero and an 8700K with 16GB of 8 Pack 3600C16. The problem I'm having is that when stress-testing RealBench the cores drop to 4.6 and HwInfo reports that the core power limit has been reached. The CPU Package Power in HwInfo also reports 95W only (or thereabouts), though it shows a max of 103W. Temps aren't a problem, max that it hits is around 60C. The VCore in HwInfo when it's running at 4.6 is also only 1.186V or thereabouts.

BIOS Settings below:

XMP
MCE disabled
Sync All Cores to 47
Adaptive Voltage 1.325V
VCCSA 1.20
VCCIO 1.15
IA AC Load Line 0.01
Ia DC Load Line 0.01
LLC 5
CPU Current Cap: 140%
 
Last edited:
It's fine at stock, but obviously only runs at 4.3 because that's how stock turbo works. It's a standard retail CPU so has not been delidded, but I don't think it's temp-related, it's in a custom loop by itself and is only running at 56 deg when it is throttled, max seen was 63.

I haven't tried 48 yet, was trying to start with plain multi-core turbo. AVX is 0.

I should say that the cores do run at 4.7, just not during realbench stress test. They also start at 4.7 during the stress test, just after a while they drop to 4.6.
 
it seemed to be unstable just turning MCE on and keeping the above settings. as in, not even realbench, just at the desktop.
 
as a follow up, and after looking on the ROG forums, it turned out to be the Long Duration Package Power Limit. Apparently if you enable XMP and then select No, the 'Auto' values use the default processor TDP which is 95W: https://rog.asus.com/forum/showthre...-required-ASAP&p=681302&viewfull=1#post681302 The post is in reference to the Strix motherboard but the same issue/whatever must affect the Maximus X Hero as well. It seems the only way to 'fix' the Auto values is to Reset to Defaults and then say Yes when asked. Or you could just set the values manually.

My CPU uses 95W at 4.6 on 6 cores in RealBench, I saw it peaking at 102-103 in HwInfo, so I just manually set the Long Duration Package Power Limit as I did to 105. Problem solved. Has passed 15m of RealBench at 4.7 on all cores using only 1.168V as reported by HwInfo, which seems pretty low to me! Might do a longer run to really check stability, or I may just let it throttle to 4.6 under huge loads as > 10% power increase for 100Mhz is not very efficient. Equally I may set it back to stock as it only uses 70-75W to do 6 cores at 4.3-4.4, so > 30% power increase for 300Mhz :S.
 
Hmm seems to be a decent chip, done 2 hours of realbench stress at 4.9 with adaptive on 1.28 and LLC4 which results in 1.232 Vcore on hwinfo. efficiency out the window though, it uses about 125-130W vs 100-105 for 4.7.
 
yes i was going to see what it will take to run at 5.0. 1.232 might not be as low as it can go, i haven't tried any lower. though in all honesty I will probably just run it at 4.7.
 
hum has done an hour of realbench at the same voltage and 5.0. still only 130W.

EDIT: oops spoke to soon, realbench passed but event viewer shows some WHEA errors.
 
Last edited:
i forgot to dumb the uncore down to auto, could be that was a bit high. i suspect you're right, seems like it's marginal at 1.28. tbh i'm quite happy with it at 4.7 and 1.16, don't see the need to push the envelope on it for 5% FPS with a lot more power consumption.

EDIT: actually they were correctable cache hierarchy errors so it probably was the uncore.
 
Last edited:
Back
Top Bottom