Myself and my colleague, Mahesh Pancholi, are fresh back from NVIDIA’s European GPU Technology Conference held in Munich. Jensen Huang, Founder, President and CEO of NVIDIA, started the event by delivering the keynote which set the tone for the three days, the main course being:
- The Volta V100 GPU
- The rise and rise of AI and machine learning
- Autonomous vehicles
- The Holodeck
The first two points go hand in hand. The V100 has the Tensor processors, or “Tensor Cores” specific for deep learning applications, said to provide a 12x speedup over P100. But of course, the V100 is said to be up to 3x faster than a P100 – but this depends on who you ask of course!
Present throughout the talks and exhibition area was the omnipresence of autonomous vehicles, both use of and technology for. Dotted through the exhibition area were various cars from Volvo, Audi and Mercedes to name a few. Did you know that an autonomous car will have at least 5 neural networks working in parallel when they hit the streets? A win-win for NVIDIA.
The holodeck was a concept that launched last year but productised this year. It is a virtual reality collaborative environment that allows users (potentially from different geographic locations) to collaborate in the same photorealistic virtual environment.
Highlights of the show, including watching one of our own customers present on how they are using AI in their work, Dr. Robert Esnouf, from the Wellcome Trust Centre for Human Genetics, spoke about the current application of Machine Learning (ML) in some very promising research in Oxford. We had some very interesting conversations with a lot of people around our capabilities to provide the High Performance Compute technologies required to run the training required on large or complex datasets for ML. This can range from an NVIDIA DGX-1 server with NVIDIA’s bespoke software suite specifically designed to lower the barrier to entry for non-data scientists or IBM’s Power Systems S822LC ‘Minsky’ server which provides speed ups using NVLink technology and an alternative CPU technology with POWER to a “traditional” HPC cluster loaded with GPU cards and a high speed parallel file system to ensure access to data is not a limiting factor. There’s a configuration to suit everyone’s needs and budgets!
Overall there was a great buzz around the show and a tangible feeling that this is just the beginning of a very exciting time in the application of game-changing tools – Machine Learning and AI… watch this space! If you attended the show too, we’d love to hear your thoughts. Please leave a comment below.