vLLM’s DeepSeek Throughput Increase on Rubin Should Be Interpreted as 58×, Not 93.3×

Written by

in

According to the official vLLM blog, the throughput increase of Rubin over GB300 is 58×, based on a rack-to-rack comparison using the AgentX workload at 387 tokens per second per user.

Source:explainx.ai · Read Original

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *