i7-4790k cache voltage
i7-4790k cache voltage
Hi,
I just bought a 4790k today and plan to use OC to 4.6 with about 1.25Vcore.
Regarding cache voltage, I’m thinking of keeping the cache ratio at the default (40x).
Some guides recommend 1.15, others suggest 1.85 or 1.2, and some say just leave it on auto.
I tried setting BIOS to default but can’t see the default cache voltage.
I checked CPU-Z, HWINFO64, Bios, and even Intel Extreme Utility — all say “cache voltage: auto, or default, no proper value.”
Do you have a reliable source for the default value?
If you keep the cache ratio unchanged, I’ll also maintain the cache voltage at its standard level. I’d only adjust voltages for the cache, ring, system agent, etc., if you alter other parameters such as the ratio for a specific setting. Leave everything else untouched unless you’re adjusting related values. Anything outside your control, don’t interfere with it. LOL.
It’s really not essential. This is just a very light overclock, and the advantages from the cache are minimal. Focus on stability after modifying core voltage and multiplier first. Then move to memory. While you’re overclocking, and until you’ve confirmed full thermal and stability compliance, stick with the default non-XMP memory settings. Work on XMP, timing adjustments, or further tweaks...
If you keep the cache ratio unchanged, then the cache voltage should also remain at its default level. I would only adjust voltages related to the cache, ring, system agent, etc., if you alter other parameters such as the ratio for a specific setting. Avoid making changes to settings unrelated to the cache; stick to what you’re not altering. LOL.
In reality, it’s not essential. This is essentially a very light overclock, and the advantages from the cache are minimal. Focus your attention on stability after modifying core voltage and multiplier settings. Then concentrate on memory adjustments. While you’re overclocking, and until you confirm full thermal and stability compliance, keep memory at its standard non-XMP configuration. Only proceed with XMP, further overclocking, or tighter timing adjustments once the CPU is fully stable.
If you wish to revert after basic stability is achieved, do so—but refrain from adjusting CPU settings during the initial phase.
A brief summary of the overclocking and stability check process:
Set CPU multiplier and voltage to your desired values in BIOS. Avoid using presets or automatic tools, as they may overcompensate for core and other voltages. It’s preferable to manually configure most core parameters and leave the rest set automatically if you plan to revisit settings later, such as cache frequency, system agent voltage, VCCIO, and memory speeds or timings after the CPU is fully stable.
Save your BIOS settings (creating a new profile if supported) and exit.
Launch the Windows desktop environment and install Prime95 version 26.6.
Install HWinfo or CoreTemp.
Use HWinfo to run "Sensors only" or open CoreTemp.
Execute Prime95 (version 26.6 only) with the "Small FFT test option," running it for 15 minutes while monitoring core/package temperatures to ensure you stay within CPU thermal limits (typically around 80°C for Intel and current Ryzen models). For older AMD FX and Phenom chips, use a thermal monitor with "Distance to TJmax" options and aim for no drop below 10°C. Anything exceeding this is outside the safe thermal range.
Once your CPU passes the thermal test, proceed to stability validation.
Download and install Realbench. Select the Stress test option and choose a memory size roughly half of your installed RAM (e.g., 8GB for 16GB, 4GB for 8GB). Start the test for 8 hours without using the system for anything else. This will simulate realistic AVX and handbrake workloads; success indicates strong stability.
For extended testing, run 12–24 hours in Prime95 Blend mode or Small FFT. You don’t need to run HWinfo or CoreTemp simultaneously during Realbench—this test already verifies thermal compliance. If you do, monitor HWinfo regularly to ensure no core/thread usage falls below 100%. If any core shows less than 100% utilization, the test has failed and you should stop.
If you encounter errors in any stability test (including Realbench or Prime95), adjust BIOS settings—either lower the multiplier or raise the voltage while keeping the multiplier constant. Any change requires restarting and rechecking thermal compliance.
For a more detailed but beginner-friendly guide, refer to this resource:
*CPU overclocking guide for beginners*
Thanks for the update. I've adjusted some settings and completed a 10-minute Intel stress test, which succeeded. The performance was around 4.6ghz with a -1.287 margin. With 40x cache, it's -1.16, though I usually set it to 1.15 manually since it doesn't display a value on auto. It seems tricky to keep track of this number.
Temperatures during the test reached a peak of 87°C on core 2, while core 3 was at 77°C—no difference in CPU age or clock speeds. The temperature gap is about 13°C. This was tested on an Aircooler with one fan.
What are your thoughts?
There was an issue, the system is unstable regardless of the cause. Even if it's just a game, your game files will become corrupted over time with each save or level. If you're not willing to perform the full testing needed to confirm stability and thermal compliance—especially since you're already 7°C above the recommended temperature for any Intel Core-i processor—then you should reset everything and avoid overclocking. Don't attempt it without proper work, or risk making a poor decision.
What you meant is that the crash in the game occurred before too, involving a locked i7-4790 I had earlier, which might be connected to the game. Either way, I'm proceeding with the tests you recommended now.
I've added an exhaust fan at the top, and my temperatures dropped by 15 degrees Celsius; Intel stress test now caps around 72 package temperature.
I'll download Realbench and run it for 8 hours tomorrow to ensure full stability.
Thank you very much.
I understand your concern regarding the game crash, and it is indeed quite plausible. Not every crash stems from instability, but it's always wise to verify that first. If stability seems intact otherwise, consider checking the GPU driver or the game itself. It’s a sensible practice to address these matters.
Here are the initial steps to follow when tackling hardware issues like this. Should you have already attempted these actions exactly as described, we can proceed to more advanced fixes.
First, confirm that your motherboard runs the latest BIOS version. If not, update it. This resolves many problems, even if the newer release doesn’t mention graphics card or hardware compatibility improvements. They rarely list every modification when they publish a new BIOS.
Next, visit the official product page for your motherboard on the manufacturer’s site. Download and install the most recent drivers for the chipset, storage controllers, audio, and network adapters. Don’t overlook updating drivers just because you think they’re unrelated to your issue. A single driver update can impact all connected devices, and an unreliable release might lead to system instability. Drivers aren’t released randomly for entertainment.
If other hardware is connected that isn’t covered by the motherboard drivers, check the support page for that component. Look for newer driver versions and install them if available.
Lastly, focus on the graphics card drivers. Even if you believe you have the latest drivers, performing a clean reinstall often yields better results than simply updating or using the Display Driver Uninstaller tool. This method has a high success rate and is worth trying.
If you’ve used both Nvidia and AMD cards during your system’s operation, run the DDU twice: once for the old driver set (Nvidia or AMD) and again for the current graphics card drivers (AMD or Nvidia). This ensures both previous and present configurations are addressed.