Overclock test, elevated clock speed, reduced voltage core
Overclock test, elevated clock speed, reduced voltage core
Hi everyone,
Welcome again, first time here.
Here are the details:
Thermaltake Chaser MK-1 case
i7-5820k (33~36)
Thermaltake Riing Silent 12 Pro 120mm CPU cooler
X99 Deluxe mobo
40gb DDR4 2400mhz (operating at 2400mhz, no XMP)
Corsair HX1000i PSU
GTX 1080 GPU
Two 120mm front intakes
Two 200mm top intakes
140mm rear exhaust
I’m planning an overclock check... I usually kept Vcore low, avoiding anything above 1.25v (I’m careful with my savings).
Recently I replaced the thermal paste (Arctic Alumina) and thought it was a good idea.
For a while I was using 45,45,44,43,42,42 @ 1.22v (hexi-core), stable enough. Then I moved to 46 @ 1.30v (VCCIN stayed auto at 1.920-1.936). I learned that 45 at 1.3v was fine.
Then I tried 46 @ 1.290 – still stable.
47 @ 1.280v worked too.
48 @ 1.275 didn’t boot, so I’m near the limit, but not gaming and don’t see a need to push back to ~1.3v for higher speeds. At 1.3v temps settled around 80°C, which is my self-set cap.
Other notes:
ASUS MCE turned off
C-states auto, EIST enabled. Windows set to balanced power mode.
PLL, LLC auto.
VCCIN auto
No BIOS changes beyond that.
Now here’s what I’m about to say – a bit of feedback. I’ve tested stability before but it doesn’t sit well with me. I play games occasionally, otherwise I spend time on websites and basic tasks; nothing too intense. So I stopped. I’ve used RealBench, AIDA, P95, IBT, etc., but I don’t like the strain they put on my system when I know I won’t reach those levels. Still, I run AIDA and Prime to monitor temps.
I noticed something with AIDA: during stock tests (36 @ 1.07v), the timer would count four numbers then skip one (like 1,2,3,4, pause, then 6,7,8,9, pause). Every time it happened consistently. So I’m using that as a reference. If I push overclock too much, the timer slows down. Even though I don’t stress my CPU for long with tests, opening AIDA shows temps stabilizing and the timer ticking normally (4 seconds then slight delay), I let GTAV and FH3 run for a few hours – that’s good, unless I notice instability.
Have I found a problem here? Or am I just too unstable to reach those limits? My PC is fast, loads instantly, max temps around 75°C under stress, gaming at 65°C. I think I could lower Vcore slightly more while keeping stability.
And about AIDA – is my theory correct? If the CPU can keep the AIDA timer running at stock speed when pushed hard, it’s a positive sign of stability?
Thanks for reading and for any thoughts. Just checking in on where I stand with this overclocking.
Cheers!
Sound good, voltage is great and temperatures appear acceptable.
A few considerations:
Have you confirmed the voltage during a stress test? You might set it at 1.275, but the motherboard (for various reasons) could push higher.
You may manage with lower voltages for some components like vccin and system agent. Auto tends to be very conservative; trying it could lead to reduced temperature and power use.
Regarding stability testing, it depends on you. You can run the PC normally unless you encounter BSOD or unusual behavior (reviewing system events is a good idea). If that happens, it’s likely stable for your needs.
Overall, you should check if you’re...
Sound good, voltage is great and temps looks fine too
Just a few points:
Have you verified the voltage when running a stress test? You might set it it at 1.275 but the mobo (for various reasons) pushes higher voltage
You might be able to get by with lower voltages on some of the other components like vccin and system agent, auto is always very conservative and if you are willing to try it might result in a lower temp and power consumption
As far as stability testing goes it is really up to you, you can just use the PC normally if you dont get a bsod or notice some unexpected behavior (checking system events is not a bad idea) then you are stable for your usage and that is mostly fine.
With that said you want to see if you are unstable on demand and not wait for some issues to surface and for that reason you might want to rest for stability a bit, you don't need to run prime95 for 24 hours if you don't want, in my experience running realbench stress test for about 30 min and passing is an excellent indication of stability.
i actually think AIDA and prime95 are not that great for stability testing only for max temps, I had unstable OCs passing prime95 for 24 hours but crashing immediately in games or realbench
Hope it help!
Here’s a revised version of your message:
The response is helpful, but the stability isn’t what I expected. It started up properly this morning, but after the first restart it triggered a WHEA error. I adjusted the clock and slightly changed the Vcore, now at 46 at 1.265, and everything seems to be working without issues so far.
Useful points: I’ve been watching temperatures but not voltages under load. After testing, I confirm the maximum voltages stay within 0.016v of what’s set in BIOS, which is normal for LLC/PLL behavior—so it should be fine.
I use HWInfo for monitoring; I’ll note that. The VCCIN value has been stable between 1.8 and 1.94, so I’m considering setting it to 1.94 automatically and leaving it as is. This ensures I won’t exceed the maximum VCCIN during load, which should be safe.
Regarding SA: I have a quad channel setup with 8GB x 2 and 4GB x 6. All components were purchased together, so I’m not sure what the optimal range is for SA or how it will impact my RAM. Any advice would be appreciated. From experience, SA tends to stay just under 1v.
I keep an eye on the event viewer and haven’t noticed anything unusual. I’ll continue doing that. I prefer Realbench for testing, though I’m a bit concerned about stressing my GPU during tests—just as with any test. I run half-hour sessions and check the results.
Thanks again for your time—it really helps!