Just lately we’ve been playing around with IBM PowerAI in order to ensure our customers can leverage it in large-scale on-premise training. PowerAI in itself is IBM’s solution for deep learning consisting of software and hardware to help you quickly train deep learning models. Today we’re happy to announce that Valohai fully supports PowerAI and our customers can start using it!
"We’re super excited to see platforms such as Valohai adopt PowerAI in doing large-scale deep learning in enterprises. PowerAI offers a unique set of software and hardware that lets companies train their models and combined with Valohai lets customers get to market fast." – Sumit Gupta, VP, AI, Machine Learning, & HPC. IBM Cognitive Systems
The main selling point for using PowerAI hardware is the speed of training. Where traditional data centers have been built with GPUs connected to each other over PCIe, with connection speeds in the 16GB/s (PCIe v3) to 63GB/s (PCIe v5) range, NVLink delivers a 300BG/s throughput. This means a 5X speed in not only GPU to GPU but also to memory and network adapter making it possible to do super fast hyperparameter search and training of models. The latest hardware setup that we’ve built Valohai for consists of 2 ultra-fast Power9 CPUs and 6 nVidia V100 GPUs which frankly blew our mind.
If you’re serious about on-premise deep learning or want access to IBM cloud this is definitely the way to go.
Using Valohai in collaboration with PowerAI lets you start up new training sessions lightning fast and automatically run down the Docker container once done without any manual work. Valohai thus takes care of the machine orchestration for you, as if you had your own PowerAI cluster on localhost. Valohai also transparently brings version control to machine learning, as every experiment is reproducible with the exact code, training data and training runtime (Docker container, hardware, and dependencies).
PowerAI and Valohai as a combination is the ultimate turbo boost to deep learning, helping you get access to machines in seconds that finish even the most complex hyperparameter sweeps in a fraction of the time of the competition.
Ready to get started with Valohai? Create a free account and get started now!