Proven Solutions Aren't Always Ideal... (Board-Rated RAM vs CPU-Rated RAM)
Proven Solutions Aren't Always Ideal... (Board-Rated RAM vs CPU-Rated RAM)
I've noticed many have seen my recommendations for Crucial Ballistix with Ryzen, especially the 3200 series. It really has performed well in my earlier 1700 build. But with the 5900X and the Asus TUF B550 PLUS board, the 2x16 3200 might not be ideal. I've been sharing this board and RAM with others, and it seems I might have caused some trouble. There were a few times I saw a blank screen when starting up—trying to determine if it was the GPU or something else. After some research, I learned Asus boards can have memory compatibility issues. I checked their site for supported RAM, but only 4000 MHz models were listed, which didn't match my 5900X's 3200 MHz. Eventually, I found that certain configurations did support Crucial, but not the 2x16 I had. It looks like I might have missed something. Looking into Patriot options gave more choices, though none offered 32GB. Maybe it's worth trying a different stick—since it does show 16GB support—to see if it fixes the issue. Before ordering new RAM, I wonder if unsupported or mismatched RAM could be causing this random video problem. So far, I haven't had crashes or other issues, just occasional no-video starts. Everything seems normal otherwise. Updated July 8, 2022 by An0maly_76
Question it. The memory layout on ASUS boards for B550/X570 is quite basic, but it still handles most 3200MHz configurations well. Considering the memory training methods built into the AGESA and what vendors are allowed to modify, the main differences come from how the system behaves near its limits. Unless you're pushing the RAM beyond 3600MHz, the choice of board doesn't really matter—it still functions. Also, keep in mind that memory QVLs have lost much of their value over time. They were crucial for compatibility in early Ryzen designs when PCB design played a big role, but often get left behind after releases and aren't updated. This leads to outdated kits when new ones appear, making them less useful. Since RAM manufacturers frequently switch ICs on the board for cost reasons (excluding Micron, which is owned by a different company), the memory you end up with might not match what you expect. That said, faulty RAM can still cause GPU problems, though they’re usually sporadic rather than outright display failures. If memory were the problem, you’d see more consistent issues than occasional glitches. This seems odd because many reported problems stem from BIOS quirks—especially on AMD boards, where revisions sometimes break. Updating or downgrading the BIOS might help. But 3200MHz is the standard target; in practice, chips can reach 4800MHz without issues on solid boards. Staying in 1:1 mode is safer and reaches up to about 3800MHz.
You haven't been sharing misleading details before, which is positive. I saw some others facing a comparable problem and adjusting the SOC voltage helped—especially with a Ryzen 3950X. I wasn’t sure it would work, but it did make a difference. Since your stock voltage was at 1.000, I raised it to 1.05, staying within safe limits as suggested. Other suggestions?
The maximum safe voltage for long-term use on Ryzen 3000/5000 series chips is around 1.2V, though it often reaches its optimal performance range (typically between 1.1 and 1.2V) for non-APUs. This could help a bit since the SOC manages the PCIe bus and other functions, but since it usually works fine most of the time, it probably won’t solve any issues.
It’s possible the PCIe3 M.2 could affect the GPU during startup.
Hold the phone... I just discovered this, actually... It's quite intriguing, especially with my USB 3.2 PCIe riser having six 4-amp and two 2-amp ports, all populated. Maybe I should replace it with an external powered hub? Another consideration came up: my front panel I/O employs two cables—one for a 3.2 port and another for a standard 3.0 header. My approach was to install a splitter on the board's 3.0 header, including a 3.2 adapter for the additional cable. Based on the insights in the attached image, could this be playing a role? Edited July 8, 2022 by An0maly_76 Revised, more info
Regarding the problem that led to this effort, I reviewed the situation and removed the USB PCIe riser. I added a splitter or adapter for connecting the front panel USB 3.2. During the "Windows Repair" process it froze, so I had to step away for the night. Upon closer inspection of the GPU today, it seemed to have four DPs and one HDMI, but now it looks like there are two HDMI ports and two DPs, with one HDMI hidden inside the DPs—this could be important. It seems the display connection port I used might have been a secondary one rather than the main one, which might explain the difference. I wonder why they would place an HDMI in a group of DisplayPorts instead of having it as the primary port. Any insights would be appreciated. Edited July 8, 2022 by An0maly_76 Updated with more details
It looks like Asus claims the board can handle 16GB of RAM at 3200 MHz, but not 32GB. For 32GB it only supports 2666 MHz (DOCP/XMP turned off). I turned that setting off. So far, everything has worked smoothly with twelve consecutive starts, shutdowns, and reboots.