My system has a Ryzen 9 7900X, an ASUS TUF Gaming B650-PLUS WIFI motherboard, a G.Skill Trident Z5 Neo RGB 64GB kit (2x32GB DDR5-6000 CL30), an RTX 4060, a Corsair RM750e power supply, and Windows 11 24H2. The memory kit was purchased new in 2023 and had been working normally.
After repeated game crashes that seemed related to a recent Windows update and RGB software, I uninstalled the update. The computer appeared fine afterward, but later it refused to POST after waking from sleep. The fans and lighting came on, but there was no display and the motherboard's amber DRAM diagnostic LED stayed lit.
I drained the power, cleared CMOS, confirmed the modules were in A2 and B2, and allowed plenty of time for memory training. The system still would not start with both modules installed. Testing them individually showed that one module boots normally in slot A2, while the other produces the amber DRAM light and never posts. The suspect module's RGB still illuminates, so its lighting controller is receiving power, but the memory itself apparently cannot train. I have not yet tested it by itself in B2, and I plan to test the known-good module there as well.
For now, I am using the working 32GB module at default DDR5-4800 settings. I have already started an RMA with G.Skill and still have the original invoice. Since this is my only work computer and I use it for video editing, I would prefer not to send away the working module for several weeks if the company requires the entire kit.
Does the working RGB combined with a failure to train point to a dead memory module or PMIC? Is there anything else worth trying before sending it in? Will G.Skill sometimes replace just the failed module, or do they normally require both sticks from a matched kit? Could the timing be related to the Windows update or RGB-driver issue, or is that almost certainly just coincidence? Also, I accidentally bent the known-good module slightly while reinstalling it during troubleshooting, although it still boots and runs normally. Could that cause problems later?
3 Answers
Your testing already points very strongly to a failed memory module rather than a motherboard slot or Windows problem. RGB only proves that the module's lighting circuitry is receiving power; it does not prove that the DRAM chips, memory controller interface, or PMIC are healthy. A module that consistently fails training by itself is a reasonable RMA case.
Before shipping anything, test the known-good stick alone in B2 to confirm both slots can boot. You can also try the suspect stick once in B2, but avoid repeatedly forcing the modules into place. If the good stick works in both slots and the bad one fails in both, the diagnosis is about as clear as it can be. The update and RGB-driver issue are very unlikely to prevent a machine from reaching POST, so the timing is probably coincidental.
For future peace of mind, run a proper memory test with EXPO enabled after the replacement arrives. Also mention to G.Skill that this is a matched kit and ask whether they can cross-ship or replace only the failed module, but be prepared for them to request the complete kit for compatibility testing.
If the good module posts in B2, I would stop troubleshooting and proceed with the RMA. A dead PMIC is one possibility, but you do not need to identify the exact failed component for the warranty claim. The fact that the RGB works is not evidence that the memory is usable; those circuits can remain functional while the actual memory path is dead.
The slightly bent working stick is worth inspecting closely under good light, especially around the gold contacts and the PCB edge. If it seats normally and passes extended memory testing, it is probably fine, but do not keep flexing it or use excessive pressure. Take photos of both modules before sending them in, and explain the situation to support rather than assuming the whole kit must be unavailable for weeks.
Related Questions
Lenovo Thinkpad Stuck In Update Loop Install FilterDriverU2_Reload