4 ms·
> More than twice as fast as Nvidia 4090 for AI. Not in memory bandwidth which is all that matter for LLM inference.
by coolspot 2y ago
> More than twice as fast as Nvidia 4090 for AI.
Not in memory bandwidth which is all that matter for LLM inference.