AI 导读
突发新闻🚨：NVIDIA 的解码专用软件基本只照顾 Blackwell Ultra 和 Rubin，对 Hopper 或 Blackwell 上的“GPU 穷人”几乎不闻不问 😭 好在 @mgoin_ 也心系无产阶级，让他的智能体集群写了 10,000 行极致融合的 Hopper 内核，解码速度最高提升 38%！
正文
BREAKING NEWS🚨: NVIDIA has mainly only shown love to Blackwell Ultra & Rubin with its decode-specialized software, but has shown mostly none to the GPU Poors on Hopper or Blackwell 😭 Luckily, @mgoin_ cares about the proletarians too and got his agent swarm to write 10,000 LoC of ultra-fused Hopper kernels that improve decode speeds by up to 38%!
来源：SemiAnalysis · x.com