|
I currently have a 7700K which I'm testing the waters in overclocking it @ 4.8GHz. Currently, I have my Vcore set in the BIOS to 1.25V (adaptive voltage). With a medium LLC setting, I am getting a Vcore of 1.264V @ full load when running stress testing programs like RealBench and the custom x264 stability test. I'd like to note that I passed the x264 stability test overnight with these settings. I tried RealBench for a little bit and it was able to pass an hour (didn't have time to do more). My question though, is that with Vdroop, I'm expecting my full load voltages to be LOWER, but this isn't the case since my Vcore is at 1.264V during stress testing. But during gaming, what I'm finding is that when I'm playing games like BF1 and Overwatch, my Vcore is actually hovering at around 1.232V. Why would this Vcore be lower when it's less intensive than RealBench/x264? Does anyone know why this is the case? What I'm afraid of is that I've tested stability @ 1.264V and that seemed OK. But when I actually use my PC for it's primary purpose (gaming), it's running at 1.232V, which is far under what I stress tested the overclock for. This is potentially unstable, but I won't know for sure until I keep playing games. Any advice on why this is happening and how I can bring the 2 Vcore values closer? Mobo is an Asus Maximus IX Hero.
|
# ¿ Apr 24, 2017 07:24 |
|
|
# ¿ Apr 25, 2024 00:48 |
|
That's actually not quite the issue. I'm familiar with Vdroop and load-line calibration settings. There seems to be an issue with adaptive voltage settings. It looks like whenever I run a stress-testing/benchmarking application that has AVX2 instructions my Vcore gets increased by about 0.03V. I've changed my LLC settings so that now when I set 1.25V in my BIOS, I get about 1.25V at full load when gaming. However, when I run a stress test/benchmark that has AVX2 instructions like x264 or RealBench, my voltage now goes up to 1.28V (increase of ~0.03V as before). The only thing I've seen of this online is a similar thing happening in Haswell, but not for Skylake or Kaby. Very weird, I'm searching online and haven't found much discussion over it.
|
# ¿ Apr 25, 2017 02:44 |
|
GRINDCORE MEGGIDO posted:Does the VID change when running avx? I'll have to log Vcore and VID tonight using HWInfo while using 2 different cases, gaming (BF1) and AVX stress tests (RealBench) and let you know. From my understanding though, once you're at that turbo ratio and when using adaptive voltage settings, it shouldn't care about VID and has to use your settings? I've tried a higher LLC setting and get the following now: BIOS Vcore setting: 1.25V Gaming Vcore: 1.25V (yay) AVX Vcore: 1.28V (nay) Using any AVX workload seems to bump my voltages up by ~0.03V, which I've seen to be consistent on all LLC settings. I'm reading a lot of conflicting reports if this is expected or not. My goal is to max out at 1.25V Vcore regardless of what type of workload at full load but I can't seem to be able to do it with adaptive voltage settings.
|
# ¿ Apr 25, 2017 20:16 |
|
I just ran some tests to compare VID and Vcore. 30 minutes of BF1: 1.248V Vcore, 1.255V VID. Both averages. 1 run of x246: 1.28V Vcore, 1.299V VID. Both averages also. Yeah, I just don't know. The extra 0.03V isn't too bad, I guess, but it's making me run a lot hotter for AVX workloads so I'm looking to find something to reduce that without sacrificing going lower than 1.25V in gaming scenarios.
|
# ¿ Apr 26, 2017 00:58 |
|
GRINDCORE MEGGIDO posted:Pretty sure AVX just gets toasty hot, it's a super toasty aggressive workload. Maybe it's that rather then the extra vcore it's dumping in? Yeah, it's that too, but I still want to minimize voltages, you know? I do video editing as well so I suspect I'll be using AVX-type workloads. I'll have to check again but when I dropped my LLC a bit yesterday I was getting 1.264V when running AVX stuff (but 1.232V when running gaming stuff which I don't want). It ran a bit cooler with that voltage (which makes sense) so I'm just really hoping I can find a way to have both be consistently at 1.25V somehow. Seems like there's plenty of people online who are not having this issue. I suppose I can set an AVX ratio to downclock it, but I want to figure this one out. Then again, most overclocking forums where I've asked this question don't even get my question to begin with and start giving me roundabout answers.
|
# ¿ Apr 26, 2017 01:17 |
|
Yeah, I don't quite know how to approach this. It's my first time dealing with it since I'm coming from Ivy Bridge. The premise of adaptive voltage seemed like a perfect feature at first. Ideally, I'd like everything to top out at a given voltage (1.25V in this case) at full load regardless of the type of load. Since my gaming voltage is lower than the AVX voltage, it's hard to put in an offset during adaptive mode. If I put a negative offset to bring down the AVX voltage, then my "gaming" voltage goes even lower than what I'd want. Seems like most people are doing OK on adaptive, I just don't know how they're doing it.
|
# ¿ Apr 26, 2017 02:17 |
|
|
# ¿ Apr 25, 2024 00:48 |
|
Maybe due to vdroop, your voltage at high load is actually lower than idle. And for some reason, your system hates the higher voltage even at no load? I dunno, hopefully you have no more issues though.
|
# ¿ Sep 5, 2017 19:49 |