The PC powers down for 4 seconds before restarting each time it boots with XMP activated.
The PC powers down for 4 seconds before restarting each time it boots with XMP activated.
Hello, I've been using an ASRock B550 Pro4 with a 3600x and a 2x8 KHX3200C16D4/8GX chipset rated at 3200CL16 for four years without any problems. Recently, about 15 days ago, I added another 2x8 kit, the CMK16GX4M2B3200C16, and set the XMP to 3000CL15 because the C3200CL16 wouldn't boot. I attempted manual overclocking but didn’t succeed, so I stuck with 3000CL15. I know there are timing differences between the two kits, but the XMP profile has been running smoothly in Windows for 15 days without any errors or crashes.
My current problem is that every time I restart the PC (only when I turn it off to avoid instability issues), it powers on for three seconds without a successful POST or reaching the BIOS screen, then shuts off for four seconds before booting normally. This happens only after cutting power to the PC, not during normal operation. It doesn’t occur during any restarts where power remains on, even though all such restarts result in the same four-second shutdown.
I’m curious—what might be causing this behavior? Is it safe for the system? And if it poses a risk, what steps can I take to fix or prevent it?
Thanks.
The BIOS has already been updated to the latest version. However, I noticed that different testing methods revealed errors I hadn't encountered during initial tests. These issues appeared after the BIOS update, which is why I plan to explore manual OC further. JEDEC speeds are proving difficult for my workflow. Any suggestions would be helpful.
It makes sense that the performance drops at 3200MT/s. My system has a 2600X that wouldn’t handle 3000MT/s until I adjusted the CL/CAS values of the two Kingston DIMMs from CL=16 to CL=18. I might have managed with CL=17, but I didn’t run another full MemTest86 because it’s just a spare PC. The 2600X is rated up to 2933MT/s, and anything beyond that depends on the specific silicon and your skill in tweaking RAM timings. Consider adjusting the CL on both DIMMs and thoroughly test. While boosting CL can increase bandwidth, it also reduces stability; you might achieve 3200MT/s if you raise the supply voltage slightly—just be careful not to exceed safe limits. The guide notes that default DDR4 runs at 1.20V at JEDEC speed and about 1.35V at XMP, but pushing it to 1.40V could help stability. Be cautious not to misinterpret a small 50mV increase as safe.
Thanks for your responses, I think I resolved it by last night and ran several tests today to achieve stability.
Here’s a summary of what I did:
Earlier, the RAM settings were A1 HX, A2 CORSAIR, B1 HX, B2 CORSAIR. After following online suggestions, I changed their positions. Although Corsair offers better B-die Micron chips than HyperX's SK Hynix CJR, Corsair had weaker timings. By placing them as the closest options, I enabled their XMP profile, which successfully worked with the HyperX units and eliminated the restarting behavior at the default XMP 3000 setting. I’m satisfied with the stability now and won’t make further changes since I don’t want to exceed 1.35V on the CJR dies, as they seem sensitive to higher voltages.
The only adjustment I made was to VSOC, increasing it slightly from the default of 1.025 to 1.05. This resolved an error in the Y-cruncher and confirmed better IMC compatibility with XMP, allowing a small voltage increase.
I hope this helps someone facing a similar problem in the future.