1. 24
  1.  

  2. 2

    Time to pop these into Nvidia’s DPU’s, and make a ridiculous compute cluster, expanding this concept