Thumbnail: dask

Talk: Intro to distributed computing on GPUs with Dask in Python

by on under talks
1 minute read

This week I presented a talk on using Dask with RAPIDS as part of a BlazingSQL webinar series.

Abstract

RAPIDS is an end to end data science stack built entirely for CUDA GPUs. Faster analytics, at scale, for lower total cost of ownership. Dask natively scales Python and the RAPIDS ecosystem stack onto multiple servers and GPUs, supporting unprecedented scale. BlazingSQL is a distributed SQL engine built in Python. It performs incredibly fast SQL queries on the RAPIDS DataFrame and ensures optimal usage of GPU primitives.

BlazingSQL is built using the same libraries underpinning the RAPIDS ecosystem. As RAPIDS improves, so does BlazingSQL.

This workshop runs through the basics of using Dask alongside RAPIDS to perform out-of-core distributed GPU computations in Python. We cover high level APIs such as DataFrames and Arrays and then dive under the covers to explore delayed functions and distributed futures.

Video

Notebook

Python, RAPIDS, GPUs, Dask, BlazingSQL, BlazingDB, Talk, Public Speaking, Slides
Spotted a mistake in this article? Why not suggest an edit!
comments powered by Disqus