Overclocked 9700K performance loss?
Overclocked 9700K performance loss?
TL;DR: I used a strong voltage boost to my 9700K (around 1.41V @ 4.8GHz) with a custom loop in spring/early summer. It worked well until mid-July when the PC would randomly crash after Steam attempted to write to my secondary NVMe drive. This happened on both motherboard slots (Z390-E). I removed and returned the P1, which was RMA'd. After that, the PC would crash during normal gameplay with only the NVMe drive installed. I checked both slots, got the same issue. It seems the problem is likely with the motherboard or GPU, not just the drive. I’ve tested the 9700K on a used Z390+8086K, but it still crashes under stress. I ran multiple tests—no issues with games, memory, or multitasking. The RAM passed all checks. My conclusion: the stock voltage and high current are damaging the board. Underclocking or undervolting would fix it, but that would make it slower than the 8086K. I’m considering giving it another try, but I’m worried about making a mistake. Have I made a huge error? Should I wait a day and do more tests? Long story—hope this helps!
Configure 4g 1.2v static frequency on the CPU and begin testing once more. It might be linked to an IMC issue; if memory tests fail during memtest, a severely bad CPU setting could be the culprit. The stuttering concern also points to possible RAM or IMC problems. A voltage of 1.4V is quite low for degradation testing, especially at 7nm processes—Intel typically uses higher voltages (around 1.52V) and would not list such a setting. After all, the 14nm architecture usually operates closer to 1.52V, making extreme voltages unlikely to trigger excessive wear.
I noticed the issue again—switching the CPU and re-running tests tomorrow. I plan to execute the same test suite plus two Memtest86 runs to diagnose the problem. I’m aiming to fix it and get the 9700K back up. I’ll share the results soon!
Updated testing sequence has been applied. The evaluation followed a specific order: OCCT CPU Stress, Linpack, Memory, Power, Unigine 1080p Extreme, 4K Optimized, and 3D Mark TimeSpy F1 on the Miami track under wet conditions for five laps. Memtest86 Standard was used in its free version, with the entire process repeated twice. All recordings were made using OBS, while monitoring voltages and temperatures via OCCT. A Twitch tab remained active during testing, occasionally stuttering or freezing but never crashing. I observed several points: the 8086K voltage and frequency auto-adjusted, XMP disabled, and no stress test failures. The Twitch video experienced lag and occasional freezes, sometimes resuming quickly; no tab crashes occurred, and OCCT reported zero errors. Memtest86 passed both runs without issues. Temperatures stayed around 65°C max, with Vcore averaging about 1.110V and peak frequency near 4.3GHz. OCCT consistently logged errors on the physical core, starting early and increasing toward the end. At approximately 4:15 minutes, these errors became more frequent. The Twitch tab also suffered from multiple crashes and stuttering, with chat failing to load until cache was cleared. This suggests potential memory issues. No hard lockups or full crashes appeared. Besides the benchmark suite, I ran Intel’s CPU diagnostic tool, which reported no faults. XTU testing succeeded, confirming a 5.0GHz core and 5.0GHz cache clock. The Vcore was set to 1.4V for stress verification. Overall, the system remained stable, though thermal paste application was adjusted—corner areas now had adequate coating. I plan to run more OCCT checks tonight and verify stability with automatic settings. Current observations remain unclear, but no critical failures occurred beyond browser instability and cache problems. I’m still trying to identify the root cause. If you have additional insights, please share! Edit: The OCCT Linpack test ran for 6.5 hours without errors on the 9700K, maintaining an average Vcore of 1.35V. Should I re-post if issues recur? Thank you for your attention.**