2 points | by TechPreacher 9 hours ago
3 comments
I have 1 DGX Spark and running models with vLLM to, out of curiosity why not using Llama.cpp / TensorRT-LLM or any other alternatives?
Oh thank god. Finally a man of the people who can show us how to optimize 10k worth of equipment.
Because we all have at least two of these. Shout out to OP!!
[flagged]
I have 1 DGX Spark and running models with vLLM to, out of curiosity why not using Llama.cpp / TensorRT-LLM or any other alternatives?
Oh thank god. Finally a man of the people who can show us how to optimize 10k worth of equipment.
Because we all have at least two of these. Shout out to OP!!
[flagged]