https://medium.com/@ttio2tech_28094/democratize-ai-turn-a-95-chip-into-a-16gb-vram-gpu-beats-most-of-the-discrete-gpus-51a8636a4719
I have tested both training and inferencing for:
A: Stable diffusion, text inversion training.
B: LLM. LlaMA fine tuning (lora).
....
似乎比CPU快些
https://www.youtube.com/watch?v=HPO7fu7Vyw4
有APU的人可以试试,想小玩的人也不用完大钱买NV的显示卡
但是GPU还是较快