RTX 4090 experiencing persistent BSOD due to DPC_WATCHDOG violations
RTX 4090 experiencing persistent BSOD due to DPC_WATCHDOG violations
ZOTAC hasn't replied to my support ticket in two weeks, and when I tried RMA through the seller, they simply sent the card back. The PC repair shop suggested it likely just plugged it in without proper testing and advised me to follow my current approach. I'm trying to process an RMA directly with the manufacturer or the seller again. It's frustrating because my expensive workstation PC remains useless until ZOTAC responds—or I have to wait and hope the seller fixes it this time. Maybe I should call ZOTAC support next week?
Well, even with an older setup, I've got an i9 9900k with a 2080 Ti in my system. My PSU is 1600W, but I'd need something around 1200W or more for the 4090. Both 3090 and 4090 models tend to show spikes, and if it lasted three months, it probably means the PSU wasn't handling the surges well. It seems they can't keep up anymore.
I believe gamersnexus demonstrated the card only reaching a spike of 40% for just 1ms, which fits within the capacity of a 1000W power supply even at that level. Most importantly, I've tested the card at an 80% output while keeping the CPU in its 140W Eco Mode for most of its lifespan, ensuring it never hit such high levels.
All the dump files reference the Nvidia driver. This crash makes it difficult for the OS to determine which driver is responsible, so I reviewed the other pending DPCs in the dump file but didn’t find any clear candidates. The PPM (Processor Power Management) appeared frequently, which is normal. You might consider updating the Chipset driver, though I’m not optimistic about that. There’s a new BIOS available—you could attempt to flash it if you haven’t already. Additionally, using DDU to remove the old driver before installing a new one may help.
It seems the connector might melt due to a failing PSU after three months. Have you checked other GPUs using the same power supply? Few managed to handle the same demands as a 4090. If the manufacturer doesn’t offer an RMA for your GPU, consider returning the PSU.
Thanks for checking out the dump files. Updating drivers, BIOS, or DDU hasn't been easy—especially after going through a lot of troubleshooting, including installing the newest GPU VBIOS and reformatting Windows twice while keeping the BIOS updated every few months. Still, it's worth a try to see if it helps.
It was mainly because I didn’t need the card to deliver that much power most of the time. I’d have liked having more parts to replace the GPU or PSU, but after moving to the UK and selling my old PC, I ended up with a lot of unused components. I’m not keen on spending a lot on replacement parts just to fix something that’s working fine otherwise. At that stage, it seemed best to bring the PC to a nearby repair shop for them to check, but they suggested returning the card as the problem was likely the same.
I’d consider a GPU too, as described. There are many parts involved, such as the PSU, which might not be the real issue unless it’s under heavy strain from a powerful GPU. It’s like throwing pasta at the wall and checking what sticks. If they can’t match it in their build, we’re left with little information about their configuration. The PSU seems to be the most likely bottleneck, even with a large power supply, but it’s also strange that artifacts appear during crashes—it usually just shuts down completely if the PSU fails. You’d need to test various solutions in troubleshooting, and turning off the expo\_xmp might help too. All other components between A and Z...
I can't perform physical actions or access hardware components. However, I can guide you through the process if you'd like. Let me know how you'd like to proceed!