This setup triggers a freeze during power-off or startup events.
This setup triggers a freeze during power-off or startup events.
Hello, I recently enhanced my PC setup by: installing a Ryzen 7 5800x CPU (refer to the attached photo), swapping the GPU to a Gigabyte GeForce RTX™ 3090 GAMING OC 24G (model GV-N3090GAMING OC-24GD) from 32 GB to 64 GB (attached screenshot shows one DIMM spec). All four DIMMs are identical except for Slot #1 and #2 which display different ranks: Dual, while Slot #3 and #4 show Single. I retained the same motherboard: X470 GAMING PLUS MAX (photo attached), and replaced the PSU from 750W to 1500W (though it’s a bit excessive, but I already owned one). The Toughpower Series PS-TPD-1500MPCG – 1500W 80 PLUS® Gold Certified Semi Modular APFC PSU was updated to the latest BIOS (version 7B79vHI1, Beta), which is meant to support Ryzen 7 5000 series. I also tried the 7B79vHB BIOS version (first compatible with those CPUs), but it didn’t work – the OS stayed unchanged: Windows 10 x64 on three monitors (two HDMI, one DisplayPort → HDMI adapter for main monitor).
The issue: when shutting down or restarting via OS or the reset button on my Meshify 2 case, the system freezes. All three monitors go black, the EZ Debug LED stays red at the CPU area, and the BIOS power/reset buttons stop working. After a hard reset, it cycles through the debug LEDs (CPU, DRAM, VGA, BOOT) before booting Windows 10. Note: with all upgrades I don’t see the POST window flash on startup. DIMMs show red at the corners (photo attached). I suspect this is just a feature of CPU-Z software recognizing the 64GB DDR4 RAM, but I’m not sure.
What I’ve done so far:
- Disabled sleep/hibernate and kept PC running 24/7.
- Investigated BIOS C-state adjustment to control power per core.
- Read about compatibility between X470 GAMING PLUS MAX and Ryzen 7 5800x.
- Checked VGA compatibility – RTX 3090 isn’t listed, possibly due to PCIe 4.0 x16 requirement not met by the motherboard.
- RAM wasn’t flagged in any compatibility list.
- Tried adjusting BIOS settings but found no resolution.
My question: is this a problem with the motherboard and CPU compatibility, or something else? I’d appreciate your advice and guidance to fix it without having to reset every time."
Share PassMark RAMMon images? CPU-Z doesn’t perform well at identifying real memory in RAM chips.
I've shared the exported HTML file created by PassMark RAMMon Edit. A screenshot of the outcomes is also attached. For those preferring not to download, you can find RAMMon - SPD Info.html. Updated on July 6, 2023 by Exquisiter.
Test at 1.125-1.15V SOC level. Also consider raising VDDG IOD to 1.05V.
I believe this could be the issue. Thank you! Although I purchased the identical 32GB (2x16GB) from the same Amazon link, it seems two years ago they assigned single ranks and now they provided dual rank DIMMs. That’s quite unusual. I plan to swap the DIMMs into the correct slots tomorrow; I hope removing the second closest to the CPU slot will help, especially with the Noctua NH-d15 radiator potentially causing a tight fit. I’ll follow up with the results. I’m curious whether one RAM channel has dual rank DIMMs while the other has single rank—would that significantly impact performance? Should I configure all four DIMMs as single or dual rank? Initially, everything was set to stock in BIOS: no XMP, no Game Boost (meaning Overclocking wasn’t enabled). I haven’t adjusted any custom CPU or RAM settings yet, but I’ll consider the voltage suggestions from @freeagent next.
Yes, you're using a mix of Samsung A-die and B-die. For the dual rank setting, with four sticks it functions correctly whether the sticks are single or dual rank. However, the RAM configuration differs and may lead to problems. Ensure the sticks are properly matched in the slots.
I managed to adjust the RAM configuration without opening the tower. The setup looked correct: DIMMA1 was on ver 4.49 (M-die), DIMMA2 on ver 4.34 (E-die), DIMMB1 and DIMMB2 switched accordingly. Everything functioned properly, freezing issues disappeared. I can now restart or put the system to sleep without problems. I’m still using XMP BIOS with 3200Mhz, but I’m considering boosting to 3733 or 3866Mhz manually. However, E-die and especially M-die aren’t recommended for overclocking—they tend to cause instability.
Just when I believed my last update fixed the issue, it didn’t. It seemed functional for a week without any problems, but then crashes returned intermittently—especially when the system was lightly used or while playing games like GTA V. The machine would freeze and display a black screen, with the EZ Debug CPU LED illuminating during restarts, forcing a hard reset. I discovered someone else facing identical challenges: they had the same memory and CPU specs and described the same problem. Their forum post linked to a solution for an MSI X470 running Ryzen 7 5800X, noting it doesn’t run properly with stock configurations. They mentioned that the processor behaves differently under BIOS settings, particularly when disabling GAME BOOST in the BIOS, which leads to regular crashes within about 30 minutes after a hard reset. These crashes also occur even without heavy workloads, pointing to low-power states during hibernation and restarts. In his reply, he suggested adjusting settings via AMD Ryzen Master or directly through the BIOS, and hinted that Load-Line Calibration might need tweaking. A related comment suggested checking the Load-Line Calibration as well.
Avoid adjusting the CPU voltage unless you're actively managing the multiplier. Early with AM4 I relied on RM twice. Once I understood I could handle everything through the BIOS, that was my approach.