Overview
Watch a 32-minute conference talk exploring the groundbreaking integration of open-source hardware acceleration support in Presto and Velox data analytics platforms. Delve into how hardware acceleration enhances data analytics efficiency, with Krishna Maheshwari from NeuroBlade demonstrating potential performance improvements exceeding 30x per dollar without requiring data Lakehouse rearchitecture. Learn about CPU and ALS architectures, data formats, logical flow, and code implementation through a detailed walkthrough of the Velox integration process. Gain valuable insights into SQL optimization, performance metrics, and the collaborative efforts with the Velox community to revolutionize data analytics capabilities through open-source hardware acceleration support.
Syllabus
Introduction
Why do we exist
SQL
CPU Architecture
ALS Architecture
How Do We Fit In
Data Formats
Performance
Logical Flow
Code Walkthrough
Veloc Integration
Questions
Taught by
Presto Foundation