Events - Daily
|Tuesday, April 25|
Research Computing General Weekly Help Session
Looking for some help getting started? Can’t get your code to run?
We offer an open help session every Tuesday morning from 10:00-11:00 in 347 Lewis Science Library.
This is an opportunity to meet with research computing staff for one-on-one help with all things cluster related. Topics include, but are certainly not limited to:
• Getting started on the cluster(s)
• Navigating the file systems
• Understanding and troubleshooting error messages
• Transferring and storing data
• Installing and compiling software
• Writing SLURM submission scripts
• Improving performance
• Programming strategies
• And many more…
Think of this as CSES office hours - no appointment necessary. We are also available to meet outside of these hours; please email email@example.com to schedule an appointment.
Room 347, Visualization Lab · 10:00 a.m.–11:00 a.m.
Introduction to GPU programming with OpenACC, 4/25/17
Introduced in 1999, Graphical Processing Units, or GPU, were originally designed to render ever more realistic images and fast animations on computer displays. This specialized hardware gave rise to the highly successful and lucrative gaming industry and led to a rapid technological development. A decade ago, the Nvidia company introduced the Compute Unified Device Architecture (CUDA) and its implementation, the "Tesla" GPU hardware for high performance computing, along with a C language extension to program it. Used as a co-processor to a standard CPU host, the Tesla GPU allowed scientific codes to take advantage of its high compute power and reach a new level of performance. Many of the top supercomputers are now using GPUs to deliver record-setting performance at high energy efficiency. One example is the Titan computer at the Oak Ridge Leadership Computing Facility.
While being successful, GPU programming with CUDA often requires significant code refactoring, leading to non-portability between CPU and GPU. To solve this problem, a consortium of compiler vendors and GPU programmers introduced "OpenACC", which is a set of compiler directives that allow a CPU code to run on GPU via simple instructions to the compiler inserted in relevant sections of the source code. These instructions can be ignored when compiling for CPU.
This mini-course will introduce the attendees to GPU programming with OpenACC directives through examples of increasing complexity. Profiling will also be discussed.
Dr. Stephane Ethier is a Computational Physicist in the Computational Plasma Physics Group at the Princeton Plasma Physics Laboratory (PPPL). He received a Ph.D. from the Department of Energy and Materials of the Institut National de la Recherche Scientifique (INRS) in Montreal, Canada.
Please register online at the training website, www.princeton.edu/training or contact Andrea Rubinstein at firstname.lastname@example.org /258-1397.
Room 347, Visualization Lab · 2:00 p.m.– 4:00 p.m.