The new CPU cooler will arrive tomorrow, assistance is needed for the OC setup.
The new CPU cooler will arrive tomorrow, assistance is needed for the OC setup.
I've been using my i7-4790k for nearly a year now, paired with a CM hyper evo. The original stock voltage was 1.21, and I increased it to 1.25 at 4.6ghz. It has remained stable at both the voltage and frequency over the past year, keeping temperatures well within normal ranges. I haven’t performed a stress test on the CPU for stability; instead, I gradually raised the voltage and everything stayed consistent.
The upcoming cooler is the Noctua NH-D15, which I prefer because it’s water-resistant and provides quiet operation. For this upgrade, I aim to reach 4.8 (possibly 4.9 if Zeus smiles), but I plan to stress test the CPU to determine the lowest stable voltage needed at that frequency. I’m seeking guidance on how to perform this voltage stability test. Some people warn against using certain programs with the 4790k due to excessive heat generation...
Here's a good read to get you started:
http://www.overclock.net/t/1411077/haswe...statistics
and
http://www.overclock.net/t/1490324/the-i...wners-club
Nice choice of CPU cooler.
Your overclocking potential depends on the luck you get in the bin lottery. Stress test tools such as IBT and Prime95 run a limited set of instructions that create heat. I prefer OCCT because it uses more typical instructions. The maximum overclock level is constrained by the voltage you can safely handle. For continuous operation, keep the boost above 1.3v. Once you hit your ceiling, apply speedstep and adaptive voltage to lower the multiplier and voltage when the processor is idle. Be ready—your NH-D15 is quite powerful. Using one of the included low-noise adapters can help reduce fan noise to near silence.
Hello, I recently installed a new CPU cooler for my i5 4690k (Master Master Air Maker 8) and noticed a noticeable improvement. To check voltage and system stability, I use Intel Extreme Tuning Utility and let it run for about 4 to 5 hours. For temperature verification, I also run Prime95 (REMEMBER: DO NOT USE ANY OLDER THAN VERSION 26.6 – here’s the link for version 26.6 http://www.mersenneforum.org/showthread.php?t=15504)).
timeconsumer :
Here are some useful resources to begin:
http://www.overclock.net/t/1411077/haswe...statistics
and
http://www.overclock.net/t/1490324/the-i...wners-club
A great selection of CPU coolers is available.
geofelt :
Your overclock potential depends largely on the luck you get in the bin lottery.
Stress testing tools such as IBT and Prime95 rely on a limited set of instructions that produce heat.
I prefer OCCT, which uses more typical instructions.
The maximum overclock level should stay within the voltage you can handle.
For continuous use, keep the voltage under 1.3v.
Once you hit your limit, apply speedstep and adaptive voltage control.
This will lower the multiplier and voltage when the processor is idle.
Be ready—your NH-D15 is quite large. You might want to use one of the included low-noise adapters to reduce fan noise even further.
I’ll aim to stay at 1.3v or below. How will I know when I reach my limit? What should I look for if stability tests fail? Isn’t offset voltage preferable to adaptive? I plan to switch to a different fan connector so I can control the fans via PWM myself. I’ve also verified compatibility, so everything should work well.
Luke2468 :
Hello, I just purchased a new CPU cooler for my i5 4690k (Master Master Air Maker 8) and it makes a noticeable difference.
To check voltage and system stability, I use Intel Extreme Tuning Utility and run it for about 4-5 hours.
If you want to verify your temperatures, I also use Prime95 (REMEMBER: DO NOT USE ANY VERSION OVER 26.6 IF YOU’RE USING HASWELL—here’s a link to get version 26.6: http://www.mersenneforum.org/showthread.php?t=15504)
No version higher than 26.6 has worked for me, thank you.
OCCT is expected to graph vcore during operation. I question whether this is correct for vcore. The system will terminate testing if temperatures hit 85°C. I also use cpu-Z, which I believe accurately reflects vcore. Adjusting the motherboard settings to reduce the multiplier and vcore when idle varies by manufacturer. The labels appear to change depending on the same controls. It's been a while since I configured it, so I'm not sure about the exact steps. With faster RAM, higher voltages are required if you enable XMP settings. I remember that using higher voltage hurt my ability to increase the multiplier. Faster RAM is balanced by more expensive capacitors, resulting in minimal performance difference.
Geofelt shared some observations about OCCT's behavior. They questioned whether the vcore measurement is correct during operation. They mentioned that OCCT might stop testing if temperatures hit 85°C. The user also notes running cpu-Z as a reliable indicator of vcore. Differences in how brands adjust the multiplier and vcore were pointed out, along with confusion over varying terminology for similar settings. Since they haven't configured their system recently, they're unsure about the exact process. They suggest that using fast RAM may require higher voltages with XMP settings, which could affect the multiplier adjustment negatively. Faster RAM paired with higher CAS numbers seems to balance performance little differently. A link was provided for further reading on memory scaling.
Vikerules :
geofelt :
OCCT claims to graph vcore during operation.
I question whether this is correct for vcore.
OCCT will terminate the test if temperatures hit 85c.
I also check cpu-Z, which I believe accurately reflects vcore.
The way to adjust the motherboard’s multiplier and vcore when idle varies by brand.
The language used appears to change for the same controls.
It has been a while since I configured my system, so the steps aren’t clear to me.
If you have fast RAM, higher voltages may be needed with xmp settings.
I remember that using a higher voltage hurt my ability to increase the multiplier.
Faster RAM is balanced by higher CAS numbers, resulting in minimal performance difference.
http://www.anandtech.com/show/7364/memor...on-haswell
I’m testing offset vcore with Windows Power Plan set to balanced; vcore drops to 0.7 and 800ghz, or will adaptive behave differently?
You’re right, the terms I used match what I remember being required to achieve the multiplier and vcore reduction when idle.
The cooler is very simple to install, though I had to take out my GPU to fit the fans but that wasn’t a big issue. During stress testing with the trial version AIDA64, Windows booted at 1.25v 4.7ghz and everything ran smoothly for 20 minutes. It didn’t boot at 4.8ghz, 1.25v, or 1.27v, but did start at 1.29v. I’m currently stress testing at 4.8ghz 1.29v with temperatures between 67-69°C. If the OC passes the AIDA64 stability test, should I proceed confidently? Or would it be better to try another stress tool? Would achieving stable performance around 4.9ghz at about 1.32v be worthwhile, or should I stick with 4.8?