Skip to yearly menu bar Skip to main content


Talk
in
Competition: Practical Vector Search (Big ANN) Challenge 2023

[Keynote] Accelerating vector search on the GPU with RAPIDS RAFT

Corey Nolet


Abstract:

Artificial intelligence, which often depends on general-purpose GPU (GPGPU) computing, continues to disrupt nearly every industry–from automobiles and healthcare to finance and the arts. Nvidia offers accelerated algorithms for vector search via RAPIDS RAFT, an open source library for applications and library developers running on GPUs. In this talk, I present the benefits of using GPU-accelerated vector search algorithms, along with some of the challenges we overcame while building them. I will also show how adopting RAFT helps developers accelerate other key components in their AI workflows.

Chat is not available.