Last month, in the heart of Washington, D.C., hundreds of industry leaders, policy makers, developers, researchers, and technologists came together at NVIDIA’s Conference: GTC DC to take a deep dive into all things artificial intelligence (AI) and deep learning.
The event was full of interesting applicable announcements – even to those who don’t follow NVIDIA’s products and solutions. Attendees learned that the U.S. Postal Service is using AI to speed up mail handling, that AI is improving at the edge, and that NVIDIA has bested the benchmark for ML inference performance, in addition to its performance in the training benchmark.
Pure was a Platinum sponsor of the event and the booth was a great success.
Pure’s Brian Carpenter, Senior Director, FlashBlade Technology Strategy, took the stage to discuss data loading and unpacked the next frontier in scale-out deep learning. Attendees learned why it’s important to have a deep level of granularity of performance metrics for training jobs, how the format of the input data can drastically impact training throughput, and that if disaggregated storage is performing worse than local storage during deep learning training, it’s important to be curious and investigate the cause. For a closer look at the deep learning pipeline powered by FlashBlade, explore our Real World AI: Deep Learning Pipeline Powered by FlashBlade whitepaper.
Resources
See how Pure and NVIDIA enable AI and advanced analytics solutions and help usher in the next era of human advancement.
