Choosing a cloud GPU machine and batch size for CNN training is difficult without knowing how these choices affect training time and cost. We developed a web application to help users compare the options before starting a long training run.

Horovod is a software framework for distributed training of neural networks. This is a short report on my setup for using a number of NVIDIA Jetson Developer Kits for distributed training with Tensorflow.

Interactive charts to compare cloud GPU providers by calculation time for a fixed amount of calculations.

Interactive charts for comparing cloud GPU offers by cost and performance.

How to use template files in bash.

A chart with a structure of Docker-related tools organised by their functionality.

Here is a way to set up networking in VirtualBox VMs so, that VMs can see each other and also the Internet.

Here are some results of testing performance of different Docker Inter Container Communication (ICC) techniques.