26/11/2014 - NVIDIA GPUs and Intel Xeon Phi Coprocessors

Event Information: NVIDIA GPUs and Xeon Phi / RSMAP

Registration is required to join this event. If you have not registered, please do so now.

Event status:Not started (Register)
Date and time:Wednesday, November 26, 2014 4:00 am 
Eastern Standard Time (New York, GMT-05:00) 
Change time zone
Wednesday, November 26, 2014 9:00 am 
GMT Time (London, GMT)
Wednesday, November 26, 2014 5:00 pm
China Time (Beijing, GMT+08:00)
Wednesday, November 26, 2014 6:00 pm
Japan Time (Tokyo, GMT+09:00)
Program:
Univa
Duration:1 hour
Description:
Coprocessors can work in tandem with a CPU to accelerate CPU intensive applications. Processes executed by the coprocessors may include floating point arithmetic, graphics, signal processing, string processing, and encryption.

In this webinar we will explain how coprocessors like NVIDIA GPUs or Intel Xeon Phis are integrated and used in a Univa Grid Engine compute cluster.

Agenda:
* Introduction
* Why Coprocessors Matters
* How Coprocessors are Configured in Univa Grid Engine
* Submitting and Monitoring Jobs Using Coprocessors
* Working with Coprocessors' Usage Metrics
* Optimizing the Configuration for NUMA Machines and Mixed Workloads
* Summary and Q&A