I haven't encountered any problems with your PC.
I haven't encountered any problems with your PC.
I’ve been experiencing some problems with my PC lately. It began after I switched back to Windows 11 in December 2023. While watching Netflix, the interface wouldn’t go fullscreen and would freeze completely until I powered off the machine. The issue also appeared on YouTube. Now, any video playback causes frequent freezes. Recently, I noticed that my peripherals and displays aren’t working even though the PC is running. Inside the case, the CPU LED was lit. I’m unsure which part is faulty. My thoughts are: the GPU could be the problem due to video issues and some Wattman crashes; the CPU might be affected by a changed thermal paste from a year ago; the PSU seems dusty; or the RAM could be mismatched because of differing frequencies. Please help me figure out the CPU.
Start by thoroughly cleaning everything. Next, reinitialize the CMOS and update the BIOS using a single RAM stick—your current setup may be incorrect. Then, attempt to use another GPU to see if crashes originate from your own hardware.
Ensure you have all the necessary codecs for Netflix installed, including HVEC and H.264.
You're probably seeing the new AV1 codec on Netflix and Twitch because they use it. Your 5500xt struggles to decode it, so it forces the decoding onto your CPU. This became noticeable when I watched on an older i5 9600k/1070ti with all six cores heavily used and no other apps running. Adjusting the video quality or switching resolutions didn't help. Make sure your CPU temperatures are normal while watching a YouTube video—dust buildup can affect performance.
This component is likely nearing its end, following a common practice for noname units: the real power delivered by all rails is usually half the advertised value. That means you’re working with about 350 watts total. Of that, roughly 100 should go to the +5V rail—something typical in older designs like Rev 1.0 or P4 with altered wiring. They often include EPS and larger connectors (6 or 8-pin), which is unusual. The remaining 250 watts split into around 50 for the 3.3V rail, leaving about 200 watts for most of the system. Modern parts rarely rely heavily on 5V, making this setup tricky; you need to balance the rails carefully, especially if the unit is meant for 12V applications. If it fails during startup, the power supply is probably the issue. I recommend OCCT 4.5.1 or a similar version. If unavailable, I have the .zip file with it.
It's dry and I only clean it occasionally. The temperatures look normal.
I attempted most options. I’ll ask my friend for a GPU just to confirm I don’t have to pay a lot for one.
Size: 8136432 bytes (7945 KiB) SHA256: 8558d1429ef11e49690f5ec6c0e2417bec519fce7224508509889148d0d3c1fc This is still the demo version, and because it's a bit outdated your clock speeds may not display correctly, but the power supply test is what matters. It's still better than the new demo which includes many ads and restrictions. Configure it as: Test type: Automatic Duration: 30m Idle Periods: 0 and 0 DirectX: 11 Graphics Card: RX 5500 (otherwise it operates in software mode) Resolution: whatever your monitor's native resolution is. Fullscreen: Yes 64 Bits Linpack: Yes The remaining option is active by default. If the system crashes immediately, verify temperatures. GPU should stay below 85°C and CPU below 90°C. It won't clean itself. occtpt4.5.1.zip occtpt4.5.1.zip.sha256
The system ran smoothly for the first five minutes before I stopped the test. I’m unable to monitor temperatures during crashes because of the Radeon software. I’m also unsure if this will assist in troubleshooting, but I received an error from the AMD tool while many tabs were open. It says “AMD software detected a driver timeout.”