GPUs built on CUDA have been used for deep-learning and other applications for a long time. But, when you look at data scientists and they work they’re doing, CUDA doesn’t really fit well into their workflow. Today’s scientists want quick exploration, quick results and to be able to shift gears without interrupting their train of thought. They want to think at the speed of data.
In this webinar, Bartley Richardson, PhD, a former fellow of The Data Incubator and a senior data scientist at Nvidia, addresses this issue. Richardson shares Nvidia RAPIDS project, an open-source suite of data processing and machine learning libraries that enables GPU acceleration for data science workflows. It also delivers a 50- to 100-times improvement over traditional GPU processing, but still using the same code and following the APIs that data scientists are familiar with (e.g., Pandas, SciKit).