Browse Prior Art Database

Method and apparatus for seemless GPU enablement on Stream Computing Platform

IP.com Disclosure Number: IPCOM000245119D
Publication Date: 2016-Feb-10
Document File: 4 page(s) / 253K

Publishing Venue

The IP.com Prior Art Database

Abstract

The essence of the invention is a practical method for seamless integration between Stream Computing and GPGPU platforms to benefit from their powerful capabilities. The method proposed by the invention offers unprecedented application performance by offloading compute-intensive portions of the application to the GPU, while the remainder of the code still runs on the CPU within high-performance Stream Computing platform. In addition, the method proposed by the invention hides GPGPU integration complexity from Streams developer, providing generality, enhanced reusability and eliminates need in GPGPU domain expertise.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 53% of the total text.

Page 01 of 4

Method and apparatus for seemless GPU enablement on Stream Computing Platform

Architecturally, the CPU is composed of a few cores with lots of cache memory that can handle a few software threads at a time. In contrast, a GPU is composed of hundreds of cores that can handle thousands of threads simultaneously. The combination of a CPU with a GPU can deliver the best value of system performance, price, and power.

Stream Computing continuously integrates and analyzes data in motion to deliver real-time analytics. Stream Computing platforms consist of both a development environment and a high speed runtime architecture.

Machine Learning real-time applications often requires to performs heavy computation in a sequential nature such as matrices manipulations on data with thousands of features thus becoming a performance bottleneck of the overall solution.

The method described in the disclosure is generic and might be applied to any Stream Computing platform. The current presentation is going to be based on IBM InfoSphere Streams* as a leading Stream computing solution.

InfoSphere Streams* is a high-performance streaming platform that can run your algorithm in a very efficient way - one of the aspects of streaming is breaking the monolithic application into small chunk of code - called operators and easily parallelize some part of the application by using multiple threads/processes/hosts.

Relevant use-cases list:

1. Use Case Abstract: Early warning earthquake system

Early warning ea...