Several issues are occurring with the PC.
Several issues are occurring with the PC.
I came back home a few days after the Christmas break and everything seemed normal when I turned on the PC. Something odd happened yesterday when I opened it up. The case fans were completely off despite the CPU being pressed down. I checked the BIOS and confirmed the CPU was overheating at 110°C within seconds, even though all fans were running. After replacing everything except the pump plugged in (my be quiet pure loop 360), the pump didn’t make any sound. I then tried re-plugging it and still got nothing. Later, I connected it to the CPU’s optional pins, but nothing changed. Hours passed, so I went back to the original connection point. Surprisingly, it started working after about half an hour. After 30 minutes, the GPU fans began spinning wildly just when Discord loaded. I checked hwinfo and saw only two fans instead of the three the GPU actually has (Aorus 3070 Ti Master). It reported 0 RPM on both fans. When I restarted, the GPU stopped making noise and the app showed the correct RPM. I thought it was a one-time issue and gave up. About 20 minutes later, when I opened a game, the GPU fans jumped to 1500 RPM right after clicking it. After closing the game, they stayed at that high speed. I’m really confused—what’s going on? Does this relate to the pump problem from earlier?
This reminds me of some unexpected problems I faced with my 5900X, which turned out to be faulty RAM that the motherboard and/or CPU struggled with. I experienced repeated crashes or freezing after cold restarts, often showing no video output for 8-12 minutes while sleeping. It seemed the RAM wasn't listed in the manufacturer's quality version, causing similar issues on another board and processor. Eventually, I purchased a compatible kit from the board maker, and everything stabilized. After learning that sleep mode can stress components, I turned it off and switched to shutdown mode, eliminating further issues. Here’s the memory QVL for that board with this processor... You can look up your RAM's part number in most browsers by pressing Ctrl-F and entering it in the search bar. Much simpler than scrolling through individual columns.
I'm not sure about the pump issue, but my GPU and fans were spinning fast without video. NOTED: @qboIodp Consider reseating the GPU and RAM—it might help temporarily. I had to do this every 8-10 reboots until I found the RAM problem. Updated Jan 15, 2023 by An0maly_76 Added more details
It's possible. I noticed Ryzen and their compatible boards handling RAM differently. My problems appeared on two different boards and processors using the same RAM, GPU, and storage. I think this might have contributed to similar issues on my older 1700 rig too. Don't forget, others here have faced the same problem and resolved it in the same way. As a tip, try re-seating the RAM and GPU if the issue persists—it was one of the main signs I saw. It usually fixes things temporarily, but once memory stabilized again, you'd need to repeat the process. Edited January 15, 2023 by An0maly_76 Revised, more info
I swapped the RAM, GPU, and everything checked with MemTest, still no issues. What should I try next?
My theory is, the memory zigs when it's supposed to zag and throws the motherboard for a loop. At that point, the GPU still has power, but no data for fan control, so it defaults to 100%. It's possible this is a dual-rank / single-rank thing. I understand that to be a thing with Ryzen also. I was trying not to get into all the gritty details, but all I can tell you is this. My system ran fine for about 30 days from build. Then one night in a particularly heavy BeamNG.drive session, it was like I pulled the plug... but not. Screen and RGB went dead, fans went wide open all around. I had to hard-press the power button to shut it down. After that, no-video POST. Hauled it down from its perch (particularly troublesome due to exercise-induced asthma and a hypersensitive throat disorder), inspected and found nothing wrong, but reseated GPU and RAM at another member's suggestion. Video is back. Sweet, we're in business. After that, it refused to wake from sleep mode and did that no-video POST every 8-12 cold boots, requiring another reseat of the GPU and RAM. I was doing this twice a week for four months and I was about ready to throw it in a lake. I was suspecting the motherboard and sent it off for RMA, then @BiotechBen mentioned something about memory training every so many cycles. So I bought a cheap B450 board and 3600X, threw all the other components on it and it had issues from the start with only one DIMM. Put the other one in, BOOM. No-video POST. Same results with another kit of the same SKU from my 1700 rig (which has also been neurotic), and memtest passed both kits. Asus board returned "could not duplicate issue", replaced RAM with kits from each board's respective QVL list. Aside from a couple isolated incidents that seem to be sleep-mode related, no more issues. Another member here tells me that he remembers using the Ballistix I first used in an Intel Coffee Lake build. So seeing as certain kits are supposedly optimized for Intel or for AMD, it would appear this is a thing. So that's two systems on which QVL'd RAM was an issue, possibly a third. And there are others here who have found that swapping the RAM for a recommended kit fixed their issues. I can't promise it will fix yours, but I can say there does seem to be a pattern with Ryzens being fickle about RAM, and your issues seem to align with how mine started. Edited January 15, 2023 by An0maly_76 Revised, more info