Waterloo Data has released a simple but elegant solution for dramatically lowering the barrier to entry for high performance computing. This solution boot-straps an Ubuntu server distribution with Docker, CUDA, Nvidia Container Toolkit, Kubernetes, and Kubeflow through a custom Ubuntu repository. Our solution addresses the common pain points faced by teams when setting up AI/ML ops that are dependent on GPU acceleration.
Solution
GpuGlue enables enterprises to quickly begin using GPU-accelerated computing for AI model creation, training, and deployment – all at the lowest possible cost for available compute power. This best-of-breed MLOps toolkit has been designed to automate the creation of GPU-accelerated AI environments, and has been carefully curated and extensively tested on GPU core systems in Waterloo Data’s advanced data science lab and across Waterloo Data’s private virtual mesh cloud.
Request access to our private beta today.
Check out the Press Release
