No output / Increase fan speed on GPU / System failure
No output / Increase fan speed on GPU / System failure
I’m running low on creative options and feeling uncertain about my system’s reliability. The main problems are a black screen or lack of display followed by intense GPU fan activity that halts after about a minute, then either an automatic restart or manual intervention. There’s also another pattern where no signal appears, but input remains and sound plays. These incidents occur mainly during playback, though timing varies. Occasionally, using Firefox with YouTube and Twitch causes a crash, which isn’t directly tied to one app.
I’ve already updated all drivers and reinstalled the correct firmware from Asus directly. The NVMe drive was previously in a laptop, so I swapped it out completely and cleaned everything—testing memory, CPU, GPU, and power supply under stress conditions. Results showed no issues with tools like Memtest86, OCCT, FurMark, etc. PSU cables are secure but not overly slack; I use direct 50s series cables from my PSU to the GPU instead of using adapters. The Armoury CRATE is fully updated.
My suspicion is a faulty CPU—possibly one core is damaged and triggers a crash under stress. There’s also a chance of incompatible RAM or unstable AMD with EXPO enabled, causing random power delivery errors. I noticed a stress test on OCCT sometimes caused the CPU to overheat, leading to crashes similar to the second incident. I saw someone else mention a thermal paste issue and voltage misconfiguration during overclocking attempts, but that doesn’t seem directly related.
I tried overclocking and observed the GPU heating up significantly, which might explain the crash due to high temperatures (around 94–95°C). Another user reported similar issues, but their fix involved incorrect thermal paste and overvoltage settings during overclocking. I can’t even attempt overclocking now because of the instability.
My current setup includes:
- Windows 10 Home x64
- AMD Ryzen 7 9800X3D 4.7 GHz
- 8-core CPU
- ASUS ROG STRIX B650E-E Gaming WIFI ATX AM5
- Crucial Pro 64GB DDR5-6000 CL40 (EXPO)
- M.2 2280 PCIe 4.0 X4 GPU
- Asus TUF GAMING OC GeForce RTX 5070Ti 16GB
I’ve replaced the PSU with a new one and tested it in both 870E-E and similar models. RAM is Crucial Pro 64GB (2x32GB) with EXPO enabled. I’m still cautious about any component that might be unstable under load.
Consider testing RAM at JEDEC speeds while disabling EXPO temporarily. Some AM5 boards remain sensitive to high-speed DDR5 despite passing stress tests. Under load, CPU temperatures and voltages may reach 94–95°C, potentially leading to instability. Verify that the PBO/curve optimizer isn’t applying excessive voltage. You might benefit from setting a manual negative curve or reducing PPT/TDC/EDC limits. BIOS version 3222 is recommended; newer AGESA versions for AM5 offer better stability. If uncertain about paste application, re-seating the cooler with firm pressure can help.
JEDEC speeds? The setup seems to be around double 32Gb storage, aiming for 6000. Are you running it with default settings? Your RGB is plugged in and fans are handling it well—3 case, two CPUs should work fine. The CPU isn’t overclocked yet, but I’m considering adjusting the curves. My mistake was installing an older BIOS; this one (Bios ver. 3278) was updated right away. Initially, the CPU had sufficient paste in the first MB, but after swapping to 870E-E I added extra paste just in case, which likely improved temperatures. Now I think it’s much better, though I’m not sure if tightening it further is necessary. Do you have a second identical CPU for testing?
Under JEDEC speeds the EXPO feature is turned off and RAM operates at the motherboard’s standard safe frequency, typically 4800–5200 MHz for DDR5. This method helps verify stability; if issues disappear, it suggests EXPO/6000 could be stressing the IMC. Your BIOS version 3278 is relatively recent AGESA support, which aids reliability. Regarding the cooler, excessive tightening may occasionally distort the IHS or board. If you suspect over-paste or pressure, re-seating with standard force should suffice. Noctua’s mounting design usually doesn’t require maximum tightness. Since you also have a matching second CPU, testing it can help identify if the problem lies with a defective chip.
Here’s a revised version of your message:
The situation is somewhat clearer now. I realized I had been blocking everything during the testing phase with the 870E-E, so I didn’t enable EXPO and just updated the BIOS without further changes. However, since then both GPUs have experienced crashes—around 20 times for the older one and once for the new one. Based on what I understand, it doesn’t seem EXPO is the cause, nor does RAM. The GPU issue appears to be more likely related to power supply or memory instability, which wouldn’t explain crashes during video playback or gaming. More likely, something stops functioning and then restarts abruptly. I still have media saved in RAM, so audio continues playing.
I’m unsure about the CPU’s role; my diagnostics show no constant problems, but it doesn’t immediately power off. It might be a hardware fault that triggers a restart when it detects failure. I attempted to tweak fan curves and EXPO settings, though I know 6000 is a stable RAM for AMD and the modules are rated for it.
My limited understanding of Windows events suggests possible issues with device lockdown or clock frequency monitoring (SEC_FAULT codes), which could be part of a larger failure chain. It’s also conceivable that a faulty component, like a damaged NIC or GPU, could cause this behavior. I tried adjusting everything, but nothing seems to prevent the crashes. The MX-4 model is still under investigation.
Based on what you explained it doesn’t seem like a massive failure that wiped everything out all at once. Such incidents are really uncommon. More typically, a single faulty RAM module, power supply unit, or CPU mounting/pressure issue leads to widespread problems. Given EXPO/6000’s sensitivity, I’d still test JEDEC speeds for a few days just to confirm. Also, because you have a second identical CPU, swapping it in is one of the most straightforward tests to determine if the problem lies with that chip. If the issue continues with another CPU, then motherboard or power supply could be the culprit. The MX-4 paste won’t usually cause this kind of instability unless the application was extremely uneven, so I’d consider that less likely.
I agree with that, though it’s starting to feel overwhelming... In order, the #1 MB B650E-E had a failing NIC and that contributed to the problems. I replaced it with a #2 X870E-E to avoid other issues, and cleared the NIC from the first model quickly. I also acquired a second #3 B650E-E, which is my preferred backup and currently works fine for internet—each time I restart the system, MSN.com appears unexpectedly. All other components like PSU, CPU, and GPU have remained consistent. I upgraded to a second CPU possibly because of CPU concerns, but I suspect the GPU could be the cause since crashes often stop the image and then reboot. I tried an MSI 5070TI OC about two days ago; by then one crash occurred with the new card, but it had no signal, the GPU fans spiked, and after roughly 30 seconds the system restarted. It seems the problem persists. Either CPU or RAM compatibility might be the issue, as I don’t think the PSU is the root cause—though the 5070 might need different cables instead of the standard ones. The connection method could matter more than a direct cable. I set everything to JEDEC standards and plan to test the new CPU tomorrow morning to see if any crashes appear. Removing one RAM stick might help, but changing too many things at once makes it hard to pinpoint the problem or verify fixes.
Latest update, the computer froze several times yesterday. GPU cleared after cleaning. Swapped CPU and used normal paste application as usual. Performance improves at lower temperatures. (Uncertain if I mentioned, but I ran three paste tests on the final boards to verify.) Memory now in JEDEC format, CPU replaced, GPU new. Started OCCT with power stability test on discrete GPU auto, crashed immediately. Might be a sign of a PC restart after CPU change. After power cycle, I thought maybe unplugging a cable near the motherboard could help, but from PSU side it worked—opened case, touched PSU, then crashed. This feels odd because some crashes occur in under 5% even with slight case contact, yet there are no moving parts. I plan to test RAM one at a time to check for changes. It seems the PSU could be the issue.)
Consider repositioning or exchanging the PSU cables, particularly the 24-pin and EPS connectors. If feasible, use a different reliable PSU for testing. Verify the setup on a separate test bench outside the case to eliminate grounding or short circuit problems. At this point, paste and RAM appear unlikely to be the issue; PSU stability seems more probable.
I set up my PC with a RM1000e Corsair unit, removing the side panel and placing the PSU on top with all cables correctly arranged. No crashes have occurred yet. The only observation is that the 12V (12+4) setup actually shows as 12+2, and there are missing two of those GPU optional sensor cables. I found an interesting article about Corsair 12v pin cables. I tested OCCT power using a discrete GPU and reached a CPU temperature of 93.7°C, with no crashes so far. I played several games extensively and still saw nothing. Tomorrow I’ll run an extended test. I also have the same PSU from a new brand, and later I’ll swap it again to check. Noticing the Asus Loki 1000W model has the 12V 12+4 setup with all four cables—this might be the reason for the crashes. I also saw that one cable from the PSU to the CPU during installation wasn’t ideal, but still functional; possibly it was meant for the GPU and acted as an optional cable that triggered crashes when the GPU was powered on.