Changes in behavior resume following memory overclocking.
Changes in behavior resume following memory overclocking.
I recently increased my DDR4 RAM speed from 2400 to 2933. I'm running on a Gigabyte Z270-HD3P-CF board with two Crucial CT8G4DFD824A.C16FBD1 8GB modules. Linux is the operating system. Testing with memtest86 and stressapptest followed by gaming showed good results. The only unusual behavior is during reboots—previously smooth transitions now involve a brief power cut followed by stabilization. I've verified my clocks and voltage using BIOS and dmidecode. My DRAM voltage is set at 1.33 and VCCIO at 1.08. This seems within typical ranges; adjusting training voltage might be worth considering.
It might involve memory exercises with each restart, though I'm not certain if a specific option exists on your motherboard. However, if all your stability checks succeed and there are no further problems, you might be fine.
I also have a memory training feature available. It's currently set to auto mode, mainly because I'm still figuring out the best configuration. Should the memory training voltage match my DRAM voltage or be lower? My side thinks it's okay as is, but another part of me anticipates making adjustments later.
It seems there might be a configuration option tailored for repeated restarts, though I can't verify it with Gigabyte hardware directly. I recall seeing something similar on my ASUS boards. I prefer clean restarts as well. My idea is that the training happens during each restart rather than just after a cold boot, possibly because of some secondary timing issues. You can check the SPD of the XMP profile in BIOS and then look at your timing settings to confirm everything is correct. Some systems automatically adjust tRC values, so reviewing those might help.