i7 8700k Delid Performance Temperature Comparisons
i7 8700k Delid Performance Temperature Comparisons
Hi. Here’s the update from my earlier overclocking discussion. I’ve finally achieved a stable i7 8700K overclock.
I’m curious about the temperatures others are achieving around 4.9 to 5.1 GHZ and what cooling solutions they’re employing.
I’m running an Air Cooler by Jonsbo (Jonsbo CR-601), which is sourced from Korea.
At 4.9 GHZ with 1.23 V, idle temperatures sit between 40-41°C and 65-67°C during a 20-minute Aida64 benchmark.
At 5.0 GHZ with 1.29 V, idle temps rise to around 43°C and load temps reach 68-70°C under the same test duration.
Are these readings typical after the delidding process, or are they still quite high?
Thanks!
Just figured out my 8700K with conductonaut on the die and inside the CPU lid—super thin layer. I was really nervous, but it turned out simpler than I imagined.
I’m running the Corsair H150i Pro AIO and Temps at 4.9GHz, 1.278v Overclock on Aida64 for a 20-minute session. The hottest core reached about 59 degrees, with an average of 55 degrees. The drop was huge, especially during Prime95 small FFTs—still in the 60s instead of the 80s.
Idle temps stay between 27 to 29 or 30 degrees C, though I live in the UK where it can dip to 24 degrees in the morning.
Not tried 5GHz yet, but I’ll give it a go soon.
It's really effective cooling for a CPU air cooler. Looks like you got lucky with those speeds at such low voltages. I recommend trying the prime 95 version 26.6. Perform the small FFT test. The blend test isn't a solid proof of stability—it might just be that your system is stable, though probably not. Share your results with pictures.
I've been testing the "Blend" mode and everything has been stable. It seems more reliable to use the Small FFT setting next. I'll attempt it and share the outcomes for both 4.8 and 5.0 OC configurations.
Btw, can you explain the difference between the "Blend" test on Prime95 and the Small FFT test? Also, vs the Aida64 CPU test.
My computer actually immediately blue-screened after trying the Small FFT so I had to up the Voltage at 4.8 to 1.24V. The results were it didn't blue screen but the Cores 4 and 6 actually throttled (CPU temps hit yellow (80's) and CPU Load dropped from 100 to 50's).
It didn't do it when I tested Aida CPU testing though.
Running with Aida64 at 4.8 GHZ and 1.3V for short periods.
After applying your suggestion with Prime95, temperatures rose noticeably across all cores. Some cores became unstable.
Is there a reason this is happening during Smal FFT tests? Should I increase the voltage?
Borrowed from elsewhere....OK so essentially you'd want to use small FTT's to isolate the CPU when trying to identify stability problems, such as insufficient vcore, because it mainly tests the CPU without heavily stressing the rest of the system. During a test like Blend, a lot of RAM is used along with a mix of large and small FTT's to provide more comprehensive stress on both CPU and RAM. If a failure occurs during the Blend test, it might be unclear what specifically caused the issue.
Borrowed from elsewhere.... Essentially, you'd want to use smaller FTTs to isolate the CPU when trying to identify stability problems, such as insufficient vcore due to limited testing scope. This approach helps focus on the CPU without heavily stressing other system components. During tests like Blend, a mix of large and small FTTs is used to more thoroughly stress both CPU and RAM. If a failure occurs during a Blend test, it might be hard to pinpoint the exact cause.
This point isn't fully accurate. The user mentioned experiencing failure during small FFT testing. The Blend test only moderately stresses the CPU, mainly the RAM. The clear reason for system failure is likely insufficient voltage supplied to the CPU. Increasing the voltage gradually while applying LLC (load line calibration) can help maintain stable levels under load.
In short, keep the system running at stock speeds until you're confident. Your current heatsink isn't ideal for overclocking. The Corsair H150i is a good suggestion.
urbancamper :
vMax :
Inspired from other sources... generally, you'd prefer smaller FTTs to isolate the CPU when diagnosing stability problems. This approach focuses mainly on the CPU without heavily taxing the rest of the system. During tests like Blend, a mix of large and small FTTs is used to apply more comprehensive stress. If a failure occurs during a Blend test, it might be hard to pinpoint the exact cause.
This response has some inaccuracies. The user mentioned experiencing issues during small FFT testing. The Blend test mainly affects RAM rather than the CPU. The clear reason for system failure appears to be insufficient voltage supplied to the CPU. Increasing the voltage gradually and using LLC (load line calibration) could help maintain stable levels under load.
In short, keep the system running at stock speeds until you can safely adjust the voltage. Your current heatsink isn't ideal for overclocking. Max is correct about the Corsair H150i. Thanks urbancamper... sometimes things get a bit confusing.