Sounds like an excellent setup for machine learning / big-data workloads. People have already moved data-base computation to the GPU, providing low-latency NVM

Sounds like an excellent setup for machine learning / big-data workloads. People have already moved data-base computation to the GPU, providing low-latency NVM access to such a set-up would really accelerate things.