This week I continued setting up my local system and cloud systems. Fortunately, I have my own local GPUs to develop with. So, I started to identify and code some simple benchmarks. It was entirely in Python (i.e. no HPCC) and TensorFlow. This is mainly to have some sort of basis for how fast certain known models can train with. I happen to have a very fast GPU, so the results will need to be analyzed with this in mind.
There is a new direction HPCC is taking, it is toward the cloud. Through this, they plan to use Kubernetes to containerize the system. I will hopefully be able to use this, but it is still in the very early stages. I am going to get early access, just waiting on some approvals.