Doing a Frankencard Driver for RTX 3050 Ti Mobile desktop card with DEV_2523

[seadersn]

Hi

I was trying to find a driver for this and failed; I then tried to use nvcleanstall and failed even more as it looks... The github frankendrivers aren't working too, no matter if tried alone or with nvcleanstall.

This card came with a chinese "gaming PC" (no laptop but desktop) and has a chip named GN20-P1-B-KB-A1 on it (I removed the cooler, see attachment). This PC had IoT Enterprise windoze on it with kind of working drivers (as my friend to whom it belongs told me there were problems with fortnite, which didn't even start; but I unfortunately don't know if it was EAC related, I can remember he told me something about a DirectX error). He then decided to install win 11 on it (a server board with a Xeon 2xxx V3 on it, X99 chipset) and there the problems began. No driver works, so he gave this card to me for testing and hopefully finding one.

What I did try already was:
- using every available 3050 Ti Mobile ID from the template (2523, 2563, 25A0, 25A2, 25E0) in nvcleanstall and a RTX 3060 mobile and nonmobile ID
- used different driver version numbers and DCH / Studio / datacenter tesla / rtx quadro versions
- flashed some other bios versions; every other ID than 2523 kills the card, but an asus 2523 bios gave a pic but completely broken ("pixel salad", see attachment). May be due to too high frequencies, I don't know.

Every time the card wasn't able to start, respectively windows is not able to load the driver and tells me that there are additional installs needed. "device stopped code 43", together with problemcode 0000002B (which seems to indicate some kind of problem with USB ports; the card doesn't have one) and the message "drivers not migrated from previous OS install due to partial or ambigous device match".

I'm out of ideas... The seller isn't answering anymore and is not even available anymore on ebay. He pointed to latest nvidias chinese website studio driver, which obviously didn't work too...

What anymore can one do in such a situation? I don't know for sure if this card is based on a GA106 or GA107.

I've uploaded the bios to the database, it shows "0MB RAM" there despite having 8GB (a config which shouldn't exist?) and nvflash --version shows

G:\Users\seade\OneDrive\Desktop\nv>OMGVflash.exe --version org.rom
NVIDIA Firmware Update Utility (Version 5.780.0)
Copyright (C) 1993-2022, NVIDIA Corporation. Veii Extreme Edition

Sign-On Message       : E4735 SKU 110 VGA BIOS
Build GUID            : 66F436FEC3514C6F88B572FCAB78C7FE
Build Number          : 29613652
IFR Subsystem ID      : 10DE-0000
Subsystem Vendor ID   : 0x10DE
Subsystem ID          : 0x0000
Version               : 94.06.1E.00.01
Image Hash            : N/A
Product Name          : GPU Board
Device Name(s)        : Graphics Device
Board ID              : 0x02C7
Vendor ID             : 0x10DE
Device ID             : 0x2523
Hierarchy ID          : Normal Board
Chip SKU              : 715-0
Project               : 4735-0110-P
Build Date            : 02/12/21
Modification Date     : 02/12/21
UEFI Version          : 0x6000C ( x64 )
UEFI Variant ID       : 0x000000000000000A ( GA1xx )
UEFI Signer(s)        : Microsoft Corporation UEFI CA 2011
XUSB-FW Version ID    : N/A
XUSB-FW Build Time    : N/A
InfoROM Version       : G001.0000.03.03
InfoROM Backup        : Present
License Placeholder   : Present
GPU Mode              : N/A
CEC OTA-signed Blob   : Not Present

 

[seadersn]

taken from https://git.proxmox.com/?p=systemd....s;hb=1690452fb03fedb541ca85979afdc05e20e6b5bb


2523 GA106M [GeForce RTX 3050 Ti Mobile / Max-Q]

 

[seadersn]

no way just no way... something is missing or wrong in general; wrong I couldn't imagine because that would mean three (four) diff pcs are "wrong"... or it's some leftovers from old drivers which are killing a new install with completely diff architecture...? I don't know, I give up