Maro97

Don’t Use V2 use V3 of ssd nvme adapter However, I believe V3.1 has been released, but I’m not sure what the differences are between it and the V3 I currently have.

No way to upgrade rtx 2080 chipset to rtx 3080 chipset … RTX 2080 Mobile (TU104, the N18E-G3 variant): the chip uses the GB4B-256 package with 2228 solder balls, measuring 37.5 × 37.5 mm. This figure is confirmed by NVIDIA's official document (NPN-09178). RTX 3080 Mobile (GA104): I couldn't find a confirmed number in the sources I searched. The commonly cited figure for GA104's package is around 1752 balls, but I'm not sure of it, so don't rely on it when buying a stencil or reballing parts. What is certain is that the ball count and package size differ between the two chips, so a single stencil won't work for both. You also can't swap one chip for the other on the same board.

The Max bandwidth is 128Gbps

This is an ADTLINK PCle 5.0 Oculink to PCle X16 Extension Cable EGPU Adapter For NUC / ITX / STX / Notebook PC Mini-PC M.2 NVMe Slot . I just found this on AliExpress: https://a.aliexpress.com/_mMYzyFv working with this MXM Adapter

I didn’t expect this. I saw this video a few hours ago and replayed it several times to make sure that it was actually possible to run a desktop GPU externally on MXM-based laptops using this adapter. The adapter has an internal NVMe connection and a DP port that connects to the graphics card to output the image on the laptop’s internal display. And as you can see, it actually works successfully. https://b23.tv/ENKfQ4S

【老哥闲来无事，琢磨着无核显笔记本外接台式机显卡并点亮内屏终于成功了!笔记本外接RTX4090不再是梦-哔哩哔哩】 https://b23.tv/ENKfQ4S

I saw it featured in a video on Bilibili.

This tested with clevo barebone P751DM2, using an MXM to PCle adapter to externally connect a desktop graphics card, and connecting the graphics card's video cable back to the MXM card, finally managed to light up the internal screen.

Has anyone tried this adapter to run an external desktop GPU?

And for me, the fun is in the journey of learning and gaining experience.

I bought a GPU, but it turned out to have a problem, and I don’t know what caused it. So I decided to return it and save myself the headache. Now I’m trying to repair the damaged GPU that was working perfectly and was known to be reliable.

The two VRAM chips that are most likely faulty are C0 and D1, because they were close to the damaged area. The remaining low error counts are probably just signal interference, meaning “error ghosting.” I think there is still a good chance the card can be repaired.

I'm asking in case anyone knows the answer: should I replace VRAM C0 and D1 , or is there something else I should check?

The 1.8V VRAM supply rail had a problem because the PCB layers made contact with each other, which caused a short circuit. As a result, the two VRAM chips closest to the crack were damaged.

And now the short has been removed and the PCB crack has been repaired. Display data appeared on the screen, and I ran a MATS test to find out which VRAM chip is faulty and causing the artifacts on the screen.