3 ms·
Is it feasible to run LLM inference comparably without CUDA or Rocm? How much of the cost performance goes away?
by mistercheese 6mo ago
Is it feasible to run LLM inference comparably without CUDA or Rocm? How much of the cost performance goes away?