i7 4790K with overclocking – one core significantly warmer than the rest
i7 4790K with overclocking – one core significantly warmer than the rest
Have you experimented with another application similar to prime95? It might be tiring to switch things around, but it can help narrow down whether the issue lies with the CPU itself. If you opt for p95, aim for version 26.6 instead of the latest release and perform small FFTs to assess a steady-state load for temperature monitoring. Blending could be challenging to monitor because it's a cyclic load that causes temperatures to rise and fall. Realbench is a reliable test but better for stability checks than thermal analysis.
You mentioned resetting everything to default. Did you reinstall the BIOS using the jumper on the motherboard or by removing the CMOS battery? Just confirming all settings are back to the original state.
Which tool are you using for temperature readings? Hwinfo64 is a strong option, and realtemp also performs well.
Have you considered using another program besides prime95 for testing? It might be tiring to switch things around, but it could help narrow down whether the issue lies with the CPU itself. If you opt for p95, aim for version 26.6 instead of the latest release and perform small FFTs for a consistent temperature load when monitoring temperatures. Blending would be difficult to track due to its cyclic nature, which causes temperature fluctuations. Realbench works well for stability checks but isn't ideal for thermal analysis.
I followed the same steps (Prime95, 26.6, Small FFTs) and let it run for 10 minutes, here are the temperatures I recorded using HW Monitor:
Prime95 26.6 Small FFTs - Temperatures
- more than a 30°C gap between the coolest core and Core0... I ensured all cores were active at full load.
I think there may be a defective sensor. You might want to reach out to the motherboard maker, though they usually don’t respond or shift blame to other issues like an improperly mounted cooler. I’d recommend sharing this page with them so they can review your actions and advice. - - Hope you get help from the manufacturer’s support team [maybe it works out].
I think there might be an issue with the sensor. You could reach out to the motherboard maker, though they usually don’t respond or point out problems like improperly mounted coolers. I’d share this information with them so they can review your actions and suggestions. Good luck with getting support from the manufacturer. Thanks for the guidance. Is there a method to check if a sensor is defective?
The capability to verify readings externally relies on the motherboard. Some models offer external access points for measurements, but most lack this feature. - - - - - - - - Additional remarks: I began my career in the computer sector more than three decades ago and have been experimenting with overclocking for approximately 25 years. Generally, with overclocked Intel CPUs in the last seven years, a temperature range of 13 degrees is typical, often spanning between 20 to 25 degrees Celsius, and reaching up to 100 degrees Celsius. If your temperature sensor is accurate, the risk increases when temperatures become excessive, which I regard as 80 degrees Celsius or higher. NOTES: Certain benchmarks and games focus on a single core, which can raise its temperature; however, such extreme variations should not occur unless there’s a major issue with cooling—such as incorrect TIM application (as you hinted in your original message) or an improperly installed cooler. Your cooler itself is generally reliable and unlikely to be the cause. To condense this, here’s a link to a TIM recommendation I shared: What bothers me most is that you’re experiencing high temperatures without any overclocking. This suggests a serious problem with cooling or a faulty sensor. Consider running your CPU fan at full speed [some boards allow this via BIOS, others support software solutions, and in some cases you may need to connect it to a 12V power adapter]. If required, retailers like Micro Center or Newegg offer adapters for different plug types. I employ a monitoring program that evaluates multiple sensors: https://www.hwinfo.com/download.php Once installed, use tools like OCCT to stress your CPU. Check the updated motherboard BIOS and review the fixes each new version provides. Visit your motherboard’s website for details, ensuring your power supply remains stable and you don’t shut down during updates. Please keep running your fan at full capacity and share your current temperatures as well as the ambient room temperature.
Danra has shared her observations and steps taken regarding the system performance. She notes concerns about high temperatures despite no overclocking, suggesting potential issues with cooling or a faulty sensor. She recommends checking CPU fan settings, using software like OCCT to stress-test the CPU, and verifying BIOS updates from the motherboard manufacturer's website. She also mentions trying different cooler options and highlights the unusual temperature changes observed during testing.
Hi xLiqxo, you're already receiving solid assistance here, so I'd like to add a bit: It's likely the IHS "lid" of your chip isn't seated optimally against the die. During idle conditions, temperatures remain acceptable as IHS can manage low heat loads. However, under full load, any inadequate contact intensifies, leading to an uneven thermal gradient. Although a lower temperature on the 4th core is anticipated, the 37-degree difference between inner and outer cores is significantly abnormal. I recommend considering an RMA if possible, or proceeding with a de-lidding process to correctly reposition the IHS.
Chenuki has shared additional insights about the chip's performance. It's conceivable that the IHS isn't fully aligned with the die when idle, which could affect thermal distribution. Under heavy load, any poor contact intensifies temperature differences, making the 37-degree gradient between inner and outer cores unusually large. A lower temperature on the fourth core is normal, but the overall gradient is significantly abnormal. If possible, consider requesting a return or a re-seating procedure. Thanks for your guidance!
There could also be an issue with my CPU cooler, especially since it's over four years old. Yesterday we replaced it with a smaller Noctua NH-12U, which helped lower the overall temperatures by more than 10°C. Still, the temperature gap between the Core0 and the other cores remains large, but I can now run it safely at default settings below 80°C without turbo. However, since the problem continues and the temperatures are still too high for stock configurations, I'm unsure what to try next.