Apache: Big Data North America 2017 will be held at the Intercontinental Miami in Miami, Florida. 

Register now for the event taking place May 16-18, 2017. 
Back To Schedule
Wednesday, May 17 • 5:40pm - 6:30pm
Construct a Sharable GPU Farm for Data Scientists - Layne Peng, EMC

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

Feedback form is now closed.
With the development of machine learning algorithms, GPUs are winning the favor of data scientists. But the high cost of GPU devices and low utilization caused by statically allocation are heavy burdens both in financial and management aspects in introducing GPUs to the data science team. In this presentation, we will introduce our latest research topic of how we enable GPU virtualization, chaining GPUs into one shared logical instance based on an intelligent queue model. In this model, the logical server can present a GPU service to one or more clients that represents GPUs local to the data center, GPUs in the cloud or some hybrid combination of local and remote GPUs executing the client application. The allocation of GPU resources is intelligent controlled based on attributes of the task, running concurrently where possible on a GPU or pre-empted to manage higher priority activity.


Layne Peng

Principal Technologist, Architect in EMC. Leading Cloud Management & Orchestration, Converged Infrastructure initiatives in EMC Office of CTO China. Thirteen patents related to cloud, SDDC and big data. One of the author of book Big Data Strategy, Technology and Application.

Wednesday May 17, 2017 5:40pm - 6:30pm EDT
  Big Data