S3 to GPU at 60Gbps: Why Parquet is Bottlenecking Your Data

Why watch

Onur Satici presents Vortex, an open-source columnar file format that achieves 60 Gbps S3-to-GPU streaming by eliminating Parquet's CPU decompression bottlenecks. The video covers Vortex's decoupled logical/physical types, cascading encodings enabling compute on compressed data, and zero-copy integration with Apache Arrow, DataFusion, DuckDB, and Polars. Targeted at data engineers and ML platform engineers optimizing GPU analytics pipelines.