OC Settings Guide 13700K
OC Settings Guide 13700K
I struggle to understand this OC setup so I need a review of my configurations. There must be something off or a mistake I’m making. Please check the attached file! My system runs a z790 ASUS Apex with an 13700K and a G.Skill 7800. I’ve pushed this CPU hard in attempts to stabilize it. Initially, I overclocked my RAM to reach 8000 MHz. I aimed for stock timings at 8000 MHz: 5.5 for all P cores, 43 for all E cores, and 48 for the rest. I began with RAM overclocking, then moved to CPU testing using Y-Cruncher in Benchmate. If that succeeded, I tried TM5, ran 1usmus_v3 and Extreme1@anta777, and if those passed, I continued with a 2-hour VST test—double the recommended time. After that, I completed a 24-hour Karhu stress test, which should be adequate for RAM OC. Then I moved to the CPU testing phase. My confidence is wavering. I kept idle voltage under 1.4 volts, but on a 13700K it seems risky. Here’s what I attempted: I managed to pass OCCT Small without issues. When I tried OCCT Large, errors appeared. The OCCT Manual settings showed voltage ranges like 5.375F, 5.415F, etc., peaking around 1.36F. I’m concerned about stability if idle voltage stays high. I’m open to re-evaluating everything and starting fresh with RAM overclocking, testing incrementally. I’d appreciate someone more experienced helping me navigate this. I’ll keep updating this thread until it works properly! help_setting.txt
Idle performance on that chip works well at 1.45V, so no major concerns there. The P and E core configurations are quite cautious, while the uncore setting is quite bold. Without the uncore you need a lot of voltage to maintain stability, and hitting 4.8GHz is likely pushing the limits for the 13th generation processors I've tested. It seems the voltage you're targeting is probably too high for reliable operation at that frequency. From past experiences with a Z690 APEX (assuming similar VRM), stick with an LLC 4 setup unless you're aiming for per-core overclocking—then level 2 would be ideal—or focus on frequency validation modes where level 6 or 7 works best. I'd suggest setting around 1.43V on level 4, lowering it to 4.6–4.7 and seeing if stability holds.
I'm facing issues recreating my previous RAM OC. It keeps failing during y-cruncher tests. I suspect focusing only on the RAM and stress-testing without overloading the memory controller might help. Others have adjusted the RAM to 8000 MHz, increased voltages to 1.5V/1.5Vddq, and found success without further changes. I recall being advised to first tune the RAM, then the CPU, and finally adjust the MC voltage until stable. This approach seems logical. I'm about to test Extreme1@anta777 next to avoid affecting the MC.
It ran smoothly on the older 8000 MHz settings. I plan to stick with the tighter 7800 MHz but increase the frequency to 8000. I found some voltage options in BIOS that HWiNFO64 shows are fluctuating a bit. VPP is bouncing between 1.8 and 1.815, while SA stays steady at 1.312 even though it should be 1.29 in the BIOS. IMC readings range from 1.403 to 1.365 under load. Could any of these factors be affecting my consistent issues in TM5?
Progress made so far. CL36-48-48-48-125-2N @ 8000 mhz successfully passed the first test with everything running normally except VDD and VDDQ, which needed 1.510v. The second test is now underway; searching online indicates it may take several hours. I’m hoping it won’t fail after just 30 minutes of completion.
I'm struggling a lot here. I'm constantly adding voltage to the RAM but it's not making a difference. I previously scored 1.52/1.52 on VDD and VDDQ during testing on the Exterme1@Anta777. I also had issues at that test with VDD set to 1.52/1.52 while MCv was at 1.25 and 1.5 in BIOS settings. It makes me think maybe the CPU settings are stuck in auto mode. I'm going to try a different method now. I reset MCv to auto. Then I adjusted the core voltages—reduced MCLK to 40, ECLK to 30, and Uncore to 35. Also set LLC to 4 as recommended by ASUS in BIOS. I'll lower the RAM voltages to 1.5 and test at 1.49 to avoid the warning. Y Cruncher passed and TM5 is running smoothly at 1usmus_v3.
Numerous RAM configuration adjustments were made. I'm currently near the final RAM test before revisiting the CPU and testing your suggestions @RONOTHAN##. I gave up on achieving stable stock timings with 8000 MHz on my 7800 board. It seemed close, but it might have been too challenging to fine-tune. I’m questioning the reliability of other users’ RAM overclock settings that reach 8000+ MHz. Instead of using a 36-46-46-46-125-2N at 8000 MHz, I’m now using a 38-48-48-48-128-2N at 8000 MHz with precise voltage settings (1.49/1.49), while the rest left to auto. This configuration passed Benchmate’s quick Y-Cruncher stress test. I then tried @1usmus_v3 and Extreme1@anta777. It was time to attempt VST. I understand VST can be tough on memory controllers. I completed 9 attempts without errors before it crashed. I ran it again at 1.4250V on the MCv, but it failed after about 61 minutes. Then I tried 1.43750V on the MCv, which also failed within under 10 minutes. Currently, I’m running it at 1.45V and it’s still working. It just failed again in 3 minutes. I’m planning to try 1.5V on the MCv next. Some people report success with 1.58 to 1.6V in air—good luck!
The voltage also fits perfectly in this case. On my Unify-X, I never perform better than 1.65V; anything higher causes failures and lower speeds lead to VST errors within ten minutes at over 7000MT/s. Of course, some boards thrive at 1.4V or even lower, and based on what others have shared, the ideal IMC ranges are around 1.35V to 1.45V. Overall, it doesn’t always mean higher is better—sometimes more can be harmful.