Overclocking CPU and Memory Independently, But Not Simultaneously
Overclocking CPU and Memory Independently, But Not Simultaneously
Here is the revised version:
I have the following configuration for my setup.
Case: Phanteks Evolv mATX
Motherboard: Asus ROG STRIX Z-370G (with latest BIOS update)
Processor: Intel i7 8700K
GPU: Nvidia RTX 2080 Founders Edition
Memory: Corsair Vengeance RGB Pro 2x8GB 3000Mhz C15
CPU Cooler: Corsair H60 (120mm fan)
Power Supply: Corsair TX650M 80+ Gold Certified
Storage: Samsung 970 EVO 250GB NVMe M.2
Ventilation:
Intake: 2x140mm Phanteks (included in the case)
Exhaust: 1x140mm Corsair ML140 Premium Magnetic Levitation
Regarding my issue, when I overclock the CPU at 4.8Ghz with 1.30V and performed a stress test, temperatures remained stable. Similarly, overclocking memory using the XMP profile (15-17-17-35 at 1.35V) also stayed within safe limits. However, after running a blend test in Prime95 for about 20 minutes, temperatures rose into the mid-80s to the high 90s.
What steps can I take to reduce these temperatures? Should I adjust the RAM speed? If I change the RAM speed from 3000Mhz to a lower setting, what timing and voltage specifications should I follow?
Could it be that my CPU overclock is restricting how quickly the RAM can be overclocked?
Any guidance would be greatly appreciated.
If this were my situation, I'd begin by adjusting the XMP/memory settings as you've done. Keep the CPU fully stocked and conduct stress tests. After success, revert to the original XMP/memory configuration and then gradually increase the CPU while continuing stress testing (this will take a while, so be patient).
CPU and memory aren't completely separate when it comes to stability. You might get one clock with memory and another with CPU, but they don't function well together—this is quite typical.
Your CPU OC operates at a relatively high voltage. Many units are reaching around 4.8 with a core voltage of about 1.26 volts. This is likely the source of the extra heat, possibly leading to thermal throttling or unstable performance tuning. You might consider adjusting the timings and voltage settings directly through DIMMS. The XMP profile could be the issue—testing slower speeds such as 2933 or 2800 might help. Have you performed a CMOS reset after the BIOS update? If not, giving it a try could resolve lingering microcode remnants and improve stability.
The Corsair H60 and similar 120mm liquid AIO CPU coolers are too small to effectively cool an overclocked i7 8700K. A 220/240 mm radiator AIO cooler is necessary to lower temperatures.
Your CPU OC operates at a relatively high voltage. Most units reach around 4.8 with a core voltage of about 1.26 volts. This is likely the source of your extra heat, possibly leading to thermal throttling or unstable overclocking. You might consider adjusting the timings and voltage settings manually through DIMMS. The XMP profile could be the issue—try using a lower speed such as 2933 or 2800 to see if performance improves. Have you performed a CMOS reset after your BIOS upgrade? If not, give it a try; old microcode remnants can sometimes cause problems, and a CMOS reset might resolve them. Thanks for the advice.
If this were my situation, I'd begin by adjusting the XMP/memory settings as you suggested. Keep the CPU fully stocked and conduct stress tests afterward. After success, revert to the original XMP configuration and gradually increase the CPU speed while continuing stress testing (it will take some time, so be patient).
CPU and memory aren't completely separate when it comes to stability. You might get one clock with memory and another with CPU, but they don't function well together—this is quite common.
Thanks for the advice. I plan to try it, likely starting at around 3.8Ghz with standard voltage and increasing from there.
Lopez Mr.22:
Roland Of Gilead:
Your CPU overclocking setup shows a notably high voltage level. Most units are reaching around 4.8 volts at approximately 1.26 vcore. This is likely the source of your extra heat, which may be leading to thermal throttling or unstable overclocking. You might consider adjusting the timings and voltage settings manually via DIMMS. The XMP profile could be the culprit—perhaps testing a slower speed, such as 2933 or 2800, will yield better results.
Have you performed a CMOS reset after your BIOS upgrade? If not, give it a try. Sometimes residual old microcode can interfere, and a CMOS reset might resolve the issue.
Thanks for your response.
I concur that the voltage is higher than what others report. Initially, I set it at 1.25 and everything worked fine. But after modifying the memory and attempting to overclock the CPU, hardware errors appeared in Prime95. Eventually, I increased the voltage until the problems disappeared.
Regarding RAM timings, I’m uncertain about the correct settings for something like 2800Mhz. It would be helpful if you could assist me with this.
Install or run CPUZ and examine the memory tab for a screenshot, then the SPD tab. This will display the recommended timings and voltage for specific RAM speeds. Record these values.
After rebooting, navigate to BIOS and disable XMP, then apply the same timings and voltage you recorded. Observe the results. This approach is safe and could resolve the issue.
If it fails, consider using XMP instead.
Calvin7’s observation is valid. With a 1.3 vcore, your cooling capacity may be limiting performance. While not critically low, it’s important to maintain temperatures below 80°C and ensure your cooler isn’t operating at maximum power to keep things stable.
You might also experiment by slightly increasing the SOC voltage. This could improve stability during overclocking.
If this were my situation, I'd begin by configuring the XMP/memory as you suggested. Keep the CPU fully stocked and perform stress tests afterward. After success, revert the XMP settings back to their original state and gradually increase the CPU while continuing stress testing (it will take some time, so be patient).
CPU and memory aren't completely separate when it comes to stability. You might get one clock speed with memory and another with CPU, but they won't function together effectively. This is quite common.
+1 I can't disagree with the approach at all. Set the memory to stock levels first, then adjust the CPU during testing. Many people just fine-tune settings without realizing what's happening. Just remember, patience leads to better outcomes.
Roland Of Gilead :
If this were my situation, I'd first configure the XMP/memory as you suggested. Keep the CPU fully stocked and perform stress tests. After success, revert XMP settings and gradually increase CPU load while continuing stress tests (it will take some time, be patient).
CPU and memory aren't completely separate when it comes to stability. You might get one clock with memory and another with CPU, but they don't function well together. This is quite common.
+1^ I fully agree with the approach. Set memory at default, tweak CPU settings during testing, and be patient. This ensures better results.
I suggest at least two hours using Prime95 and/or RealBench. Honestly, I prefer RealBench since it mimics a more realistic environment (encoding videos, handling multiple tasks, etc.). It’s best to let it run all night—remember the patience we talked about. 😉
During testing, pay attention to how quickly your temperatures stabilize, meaning they stop rising. If they keep increasing slowly and take a long time to settle into a steady maximum temperature, it could indicate cooling issues need addressing.