MWRC 2011 - Ruby Supercomputing: Using The GPU For Massive Performance Speedup

Channel:

Confreaks

Subscribers:

42,400

Published on February 17, 2015 7:34:56 PM ● Video Link: https://www.youtube.com/watch?v=C6y4YkpaT_I

Duration: 35:35

273 views

2

By, Preston Lee
Applications typically measure processing throughput as a function of CPU-bound algorithm performance. Most modern production systems will contain 2-16 processor cores, but highly concurrent, compute-heavy algorithms may still reach hard limits. Within recent years, vendors such as Nvidia and Apple have formalized specifications and APIs that allow every developer to run potentially 1,000s of concurrent threads using a piece of hardware already present in the machine: the GPU.

Help us caption & translate this video!

http://amara.org/v/GIj4/