What happens after switching the GPU?
What happens after switching the GPU?
I recently switched my GPU from a 3070 to a 5070 and it has been functioning well for a full day. I didn’t play anything demanding, but I ran some quick tests with OCCT and got no errors just to verify performance. However, while browsing the web, I experienced a brief screen freeze lasting about five seconds, followed by a couple of seconds of black screen before everything normalised.
In the Event Viewer, I noticed two errors linked to NVIDIA at that time:
- One instance with the description “The description for Event ID 14 from source nvlddmkm cannot be found.”
- Another with “The description for Event ID 153 from source nvlddmkm cannot be found.”
Both mention installation issues or missing components. If the event happened on another machine, the display details should have been saved with it. The attached information included:
- Device: Video3
- Memory: 32344000 0000b670 00000000 206a7a8a 206a6c4a 206a6db8 206a52ae 206a5aca
- Message resource present but message not found in table
- Four instances of a similar error
I’m unsure what’s causing the problem. Some drivers were uninstalled before replacing the old GPU and then reinstalled after the new one was set up. During stress tests, my GPU temperatures stayed between 30 to 35°C, reaching up to 71°C during the tests.
Here are my system specs:
- Motherboard: MSI Z790 Gaming Plus WiFi
- CPU: Intel i7 14700K
- GPU: Gigabyte GeForce RTX 5070 WindForce OC SF
- RAM: Corsair Vengeance DDR5 6000 (2 x 16GB)
- AIO: MSI MAG Coreliquid M240
- SSD: Crucial P3 Plus 1TB
- PSU: MSI 850W 80+
Anyone have any suggestions on what might be going wrong?
Additional details provided include that drivers were removed before installing a new GPU, then reinstalled after the new one was added.
The entire GPU replacement procedure followed this sequence:
Powered on the PC
Launched installed applications and deleted all NVIDIA-related software (video driver, audio driver, control panel, physx and frameview SDK) version 581.57 from Ocrober 14th
Restarted the computer to finish uninstallation
Powered off the PC
Switched the old 3070 chip for the new 5070
Powered on again
Installed the complete NVIDIA driver package version 591.59, released two days ago with the option "perform a clean installation"
Restarted the PC to finalize driver setup
System information details:
BIOS Version/Date – American Megatrends International, LLC. H.70, 27/09/2024
This appears to match the manufacturer’s website? If yes,
www.corsair.com/it/it/p/memory/cmk32gx5m2b6000c30/vengeance-32gb-2x16gb-ddr5-dram-6000mt-s-c30-memory-kit-black-cmk32gx5m2b6000c30
www.msi.com/Power-Supply/MAG-A850GL-PCIE5
Model release around May 2023, yet I purchased it in January 2024, meaning less than a year of operation.
Additional notes I considered: the GPU is connected to the PSU via a 16-pin to 16-pin cable included with the PSU. The device came with only a 10 cm long 16-pin to 2x8-pin adapter, which I didn’t use. The card appears properly seated and the cable is securely attached as far as I can see.
The BIOS version currently in use features microcode 0x12B. It is recommended to update it to version 7E06vH9, as this microcode helps avoid processor issues unless the device is already outdated. After updating, ensure the CMOS is cleared to confirm the new BIOS version.
I knew there was an issue with the Intel 14th generation processors, concerning overheating and BIOS version, yet I thought the BIOS I installed (the latest one when I purchased the motherboard) had resolved it. The motherboard itself came with a significantly older BIOS version, just sufficient to support the Intel 14th generation chips.
Just a note: if I follow your advice, will the BIOS settings revert to defaults or remain unchanged? (I adjusted the fan curves and applied a slight undervolt just to ensure safe temperatures.)
Will proceed with this as my first step. I didn’t consider switching to a full DDU cleaning because the 3070 and 5070 drivers are essentially the same, so I went ahead with the usual method.
Correct, but here’s a bit more detail:
I purchased the 3070 about two months after its release and used it on my old system until January 2025. In January 2025, I upgraded everything except the GPU (most of what was listed in the first post), except for the GPU itself, which I kept using the original 3070 until now.
Edit:
I performed a complete driver clean with DDU and enabled safe mode.
I’ll keep the BIOS flashing on standby for now since I checked online that changes to BIOS settings could affect stability. I’m also double-checking whether the 0x12B microcode is responsible for the overheating, as it addresses this issue. 0x12F handles another CPU degradation problem that might happen if the PC runs continuously for days—this isn’t my concern, though I’ll wait until I confirm the GPU fix is resolved before making any changes.
I’ll keep monitoring the situation and update this thread if anything changes or if issues persist within a week.
Edit 2:
Unfortunately, this occurred again just now, matching event ID 14 error once and event ID 153 four times. I was browsing at the time, so these might be helpful references.
I’m planning to attempt another DDU clean using slightly older drivers, hoping it resolves any stability concerns.
What other options could I consider?
Edit 3:
Today’s day, I tried playing a bit and everything worked for a few hours before suddenly crashing to the desktop. The only event in the viewer was the game crashing once. No GPU or driver issues were detected. The PC and screen remained responsive during the crash (though mouse movement was possible), while other times everything froze completely for a few seconds. I’m unsure if this is related or just bad luck, but the game hasn’t crashed before in the past month.
I have the identical error logs. ID 153 relates to TDR and ID 14 is similar.
My configuration is entirely different:
Motherboard: ASUS B650EM TX GAMING
CPU: Ryzen 9700X
GPU: Manli RTX 5070
OS: 25H2
However, my system is a fresh install with no changes made, and it isn't connected to other hardware.
I only notice this issue during RDP sessions, and the behavior remains consistent, though I didn’t observe it in other programs.
Update: Performed DDU clean install 591.59 and updated BIOS, but no improvement.
By the way, I’ve also experienced other black screen errors while running CUDA accelerated tasks such as hashcat or LLM applications like whisper; those screens don’t recover, and the fan reaches maximum speed. I have to restart and power on to resolve it.
It has been nearly a week since I switched from version 591.59 to 581.80 (noting there is a version 591.44 in between, but it seems an earlier 591.59, so I chose to skip it). So far, no more errors have appeared—IDs 153 and 14. I’ll keep monitoring, hoping it’s the driver version that had some problems.
If this remains an ongoing issue, initiate a separate discussion thread. Make sure to include all relevant system information for your setup.