Question for a beginner about i7 9700k
Question for a beginner about i7 9700k
Hi!
I've been attempting to achieve a stable 4.9ghz OC on my Intel i7 9700K and was seeking advice. I chose 4.9 because that's what came out when I enabled enhanced turbo boost in the BIOS, and I thought a lower core voltage would suffice.
I began by adjusting the core voltage to 1.310 with LLC 4, while keeping the manual C-states and speedshift enabled on my MSI Z390-A Pro. According to HWINFO, this setup reached up to 1.320 under maximum load, but at full load with Lower LLC or LLC5 it dropped below 1.300 during stress tests. LLC3 would push the voltage even lower.
This configuration caused Realbench to stop early due to instability. I increased the core voltage in the BIOS to 1.320, which then stabilized at 1.328 under full load during the stresstest. The test ran for 41 minutes and displayed "Result Hash Match!" but also showed a "Result mismatch. System Unstable. HALT!" message once. After about 12 minutes, both monitors briefly went black before coming back on—possibly an indicator of instability.
My recorded temperatures during the stress test were:
Core 0 – Max 85°C
Core 1 – Max 84°C
Core 2 – Max 89°C
Core 3 – Max 87°C
Core 4 – Max 88°C
Core 5 – Max 86°C
Core 6 – Max 84°C
Core 7 – Max 84°C
CPU Package temperature reached a maximum of 89°C.
The average across all cores was around 82°C, with Core 2 being the highest.
I have a Noctua NH-U9S and I don’t know my room temperature precisely, but it seems to be about 20°C, which makes these readings feel elevated, especially since I set the Noctua to 80% speed at 80°C.
Some users recommend letting the stresstest run for several hours or even days, but given these temperatures and the error messages, I’m hesitant to leave the system unattended for long periods.
Could you clarify what the realbench warnings mean? The "Result mismatch. System Unstable. HALT!" message is clear, but I’m puzzled as to why Realbench sometimes halts the test automatically and at other times gives me that same alert.
I also tried CoD Warzone with core voltages at 1.310 (up to 1.312) and 1.320, both appearing normal in practice mode with fewer players. However, since it was in single-player mode, the load on the CPU was much lower than during typical gaming sessions.
I’m still not fully comfortable with vdroop. Given my voltage drops by 0.008v from idle to full load, am I correct in assuming that’s vdroop?
I mainly use my PC for gaming and occasional video editing, usually for short sessions of 8–12 hours or so. As long as the temperatures remain normal (around 60–70°C with brief dips into the low 70s) and I don’t encounter any BSODs, would it be safe to keep things as they are? Other signs of instability would include a regular screen freeze, wouldn’t they? If that happens, I’d need to either raise the core voltage or lower the multiplier.
Thanks in advance!
Erik
I lowered my OC to 4.8ghz for easier temperature control.
After experimenting with core voltage starting at 1.250, I reached this setting:
VCORE bios 1.280V Max VCORE recorded by HWINFO 1.288V VCORE and remained stable under load during the stress test, staying at 1.280V.
During testing in realbench, I received several "Result match!" messages and two "Result mismatch. System unstable. HALT!" alerts.
Stress test completed successfully.
Timer: 30 minutes.
Core temperatures during the test:
0 Max 79 AVG 72
1 Max 78 AVG 71
2 Max 82 AVG 75
3 Max 80 AVG 73
4 Max 80 AVG 75
5 Max 80 AVG 73
6 Max 77 AVG 71
7 Max 77 AVG 72
CPU Package power Max 122.506 W AVG 110.500 W
The two "Result mismatch. System unstable. HALT!" messages are concerning, even though the test finished normally.
I’m wondering if this setup is reliable for everyday use and should I only increase the vcore when I see a BSOD or freezing?
I’d also like my vcore to adapt automatically—seeing suggestions to keep CPU core voltage in "override" mode and switch Windows Energy plan to "balanced" helps. But it doesn’t seem to lower idle voltage.
If I switch to adaptive voltage mode, could it cause the vcore to jump significantly on an i7 9700K? Would that require an offset to maintain my target setting? Could this offset affect stability when the CPU is idle?
Appreciate your help!
Erik