Unusual actions on your latest computer are happening.
Unusual actions on your latest computer are happening.
I recently assembled a new setup for my brother featuring these components: an i3 13100F ASUS PRIME H610I-PLUS with a fractal design, a 450W PSU (though I know it’s not ideal), MSI GeForce RTX 4060 VENTUS 2X BLACK, 8GB DDR4 3200MHz, a Noctua NH-L9i-1700, Samsung 850 EVO 2.5", a 1TB M.2 SSD, and a Kingston SSD. Right away, I faced frequent freezes that forced a hard shutdown. There were no error messages, temperatures stayed normal, and performance was decent—Cinebench ran for about five minutes, Fortnite loaded with extremely low frames, and it would freeze instantly when exiting the game.
I built the PC without accounting for the power demands of the RTX 4060, and after realizing this, I upgraded to a 750W PSU. The results were identical. I checked thermal paste, connections, RAM placement, drivers, BIOS updates, fresh OS install, and other potential issues. Still, the GPU seemed faulty. I swapped it with his older 1650 model, and everything functioned perfectly with updated drivers.
Then came the unexpected part. I was planning to return the RTX 4060 for an RMA, just to test it in my own machine. But it performed flawlessly—playing Battlefield 2042, Warzone, and other games ran smoothly without any problems. What should I do next? Am I overlooking something? My goal is to get the 4060 working on my brother’s PC using a 600W PSU instead of the 450W he currently has. Feel free to ask if you need more details.
It seems there might be an unusual driver issue at play. If swapping out the GPU resolves the problem, the cause likely lies elsewhere. Moving the faulty GPU to another system might indicate a software-related fix. Consider adjusting settings like enabling high-performance mode, using DDU, reinstalling drivers, updating motherboard and BIOS, and ensuring Windows is up to date.
We checked using Cinebench 2024 with GPU only, skipping CPU checks since games are the main concern. The results were consistent across all PCIE versions and power configurations. TL;DR: Need to verify CPU and RAM stability separately. Explanation at the end. Have you run a memory test with MemTest86? With 16GB RAM, it will still take a long time (hours), but not as long as those with bigger RAM. Assuming your RAM is stable, does CoreCycler flag any errors? You should be able to use it with standard settings given the stability problems. Are you running any overclocks or XMP profiles? Some boards auto-overload the CPU. If you skip MemTest86 and CoreCycler, OCCT could work instead. But wait—didn’t you mention testing with a GTX 1650? Everything worked fine. Ah, but your post said 3D apps. Those are GPU-limited, not CPU-limited (I checked userbenchmark). They show the RTX 4060 is roughly twice as fast as the GTX 1650. This suggests any test should cover both GPU and CPU. If one consistently struggles, you’ll miss the other. It seems every run tests both together. The RTX 4060 will demand much more from the CPU, RAM, and chipset. Therefore, testing CPU and RAM in isolation is essential. Let us know your findings so we can pinpoint the issue.