hmm not sure if it’s Nvidia. It might be tools like LM Studio and Ollama (which use llama.cpp) which could make it simpler. Someone who has multiple GPUs can comment on how easy and how well such tools make use of multiple GPUs. I’m worried your second 6000 Pro itself might not give you the benefits for a while.
Not sure if you’re aware of https://www.reddit.com/r/LocalLLaMA It’s a good compliment to this forum. If you want to run a quant of Deepseek, there’s stuff like DeepSeek-R1-0528: How to Run Locally | Unsloth Documentation and DeepSeek Deep Dive R1 at Home!
One goal would be to build/compile llama.cpp/ik_llama.cpp. After that trying different commands that use multiple GPUs is easier.