This document summarizes highlights from OpenACC in May 2017, including:
- OpenACC provides a directives-based approach to parallel programming for performance and portability on CPUs and GPUs.
- A molecular simulation saw an 11.7x speedup using a GPU compared to a CPU.
- Adding simple OpenACC directives to serial code enabled parallel execution on different hardware with minimal code changes.
- Upcoming OpenACC events in June 2017 include a user group meeting and workshops at the ISC conference.
2. OpenACC is a directives-
based programming
approach to parallel
computing designed for
performance and portability
on CPUs and GPUs.
3. PERFORMANCE PORTABLE PROGRAMMING MODEL FOR HPC
Powerful
LSDALTON
Simulation of molecular energies
1.0x
11.7x
CPU GPU
Big Performance
CCSD(T) Module, Alanine-3
Titan System: AMD CPU vs Tesla K20X
SpeedupvsCPU
Simple Portable
main()
{
<serial code>
#pragma acc kernels
{
<parallel code>
}
}
Add Simple Compiler Hint ARM
PEZY
POWER
Sunway
x86 CPU
x86 Xeon Phi
NVIDIA GPU
Quicker Development
Lines of Code Modified
<100 Lines
# of Weeks Required
1 Week
5. OPENACC AT ISC 2017 COMPLETE SCHEDULE
Featured Event Link
OpenACC User Group Meeting Read More
BoF 05: Performance Portability & MPI+X: The State of X Read More
Performance Portability with OpenACC & PGI Compilers Read More
Second International Workshop on Performance Portable
Programming Models for Accelerators (P^3MA)
Read More
User Group Meeting Talks Workshop Members
6. OPENACC COMMUNITY AT ISC17
REGISTER NOW
3rd OpenACC User Group Meeting
Have a Great Time!
Discuss Training
Collaborate on OpenACC-related research
and activities
Share Experiences and Best Practices
Provide Feedback on the Specification
7. NEW PGI COMMUNITY EDITION
First release to support Microsoft Windows
Comprehensive support for the latest NVIDIA Pascal GPUs
CUDA Unified Memory support in both CUDA Fortran and OpenACC
Full OpenACC 2.5 support; improved cache directive support
Available for Download. Now on Windows!
WHATS NEW DOWNLOAD
8. GTC 2017 AT A GLANCE
COMPLETE LIST OF TALKS
Featured Talk Link
OpenACC Best Practices: Accelerating the C++
NUMECA FINE/Open CFD Solver
Watch Now
Achieving Portable Performance for GTC-P with
OpenACC on GPU, Multi-Core CPU, and Sunway
Many-Core Processor
Watch Now
GPU Acceleration of the Hirad Computational Fluid
Dynamics Code with Mixed OpenACC and CUDA
Fortran
Watch Now
The Future of GPU Data Management Watch Now
OpenACC had a great impact at GTC this year
9. RESOURCES
Paper: Performance portability in reverse time
migration and seismic modelling via OpenACC
“A performance enhancement of ~10× was obtained, when the acoustic
model was ported to a single GPU, compared with a 1.7× speedup obtained
using the isotropic model. Our MPI+OpenACC implementation of reverse
time migration and seismic modelling shows promising scaling when multiple
GPUs were used.”
READ NOW
Tutorial: Michael Wolfe’s newest tutorial video
Introduction to Parallel Programming with OpenACC Part 2
WATCH NOW
10. CALL FOR PAPERS
EVENT DUE DATE LINK
CSCS GPU Hackathon June 9, 2017
https://www.olcf.ornl.gov/training-
event/2017-gpu-hackathons/
COMPLETE LIST OF EVENTS
11. UPCOMING EVENTS
EVENT & LOCATION DATE LINK
XSEDE HPC Summer
Bootcamp
June 6 - 9, 2017
https://www.psc.edu/136-
users/training/2561-xsede-hpc-workshop-
june-6-9-2017-summer-boot-camp
Brookhaven National
Laboratory, Upton, NY
June 5 - 9, 2017 https://www.bnl.gov/gpuhackathon/.
OpenACC User Group
Meeting, ISC, Frankfurt,
Germany
June 20, 2017
http://www.openacc.org/events/openacc-user-
group-meeting-isc
2nd P3MA Workshop
ISC, Frankfurt, Germany
June 22, 2017
http://www.csm.ornl.gov/workshops/p3ma
2017/
NASA GPU Hackathon Aug 21 - 25, 2017
https://www.olcf.ornl.gov/training-
event/2017-gpu-hackathons
COMPLETE LIST OF EVENTS