After I already got two Nvidia Tesla P4 GPU cards from my colleague Marc Huppert (VCDX#181) in the beginning of 2021, coincidence struck in October 2021, I met a former colleague of mine again at my second hobby mountain biking, we also talked about my Homelab and there my former colleague surprised me with an Nvidia Tesla M10 GPU that was lying around unused at his place. So the GPU ended up in a free PCIe slot in one of my servers. At this point many greetings to Marcel C. and many thanks 🙂
As you might know, the Nvidia Tesla M10 GPU card is one of those that still need an additional power connector. Specifically one with 8 pins.
Unfortunately, the GPU enablement kits for my HPE DL380 servers are only available at horrendous prices on eBay. So I thought I’d go looking for a single cable that connects from the 10-pin connector on the server’s riser card to the 8-pin connector on the graphics card.
In short, it was a nightmare with a lot of returned cables, as HPE doesn’t seem to follow the ATX standard here, why would they?
After 4 exchanged cables, partly with the same PIN assignment and only with different lengths, the card finally worked.
32 GB vRAM for VMs is already a lot and I now have the possibility to supply each of my VDI VMs with a vGPU, which allows me to simulate a lot of customer use-cases.
Server without GPU Riser Server with installed Nvidia Tesla M10 – short Power Cable was not working 🙁 wrong 8 Pin Connector 🙁 Power Connector of Tesla M10 Power Connectors are not matching 10 Pin Power Connector on HPE Riser Board Installed Nvidia Tesla M10 with long Power Cable – also not working 🙁 M10 Card – Up and Running 🙂
After the Nvidia Tesla M10 GPU card finally ran in my server I was able to get a lucky shot on eBay, but how it goes on you can read in Homelab version 2.3