This event has been cancelled.
The goal of the hackathon is for research groups to work with HPC mentors to optimize their software to take full advantage of petascale class HPC systems. The research groups selected to participate will send teams of at least 2 developers along with an existing application that can benefit from taking full advantage of petascale class computing systems to accelerate their research productivity.
There will be intensive mentoring during this 5-day hands-on event, with the goal that the teams leave with applications running more efficiently and at greater scale, or with a clear roadmap of how to enhance their code performance. Each team will be assigned mentors who have extensive experience in petascale class programming.
The hackathon is intended for teams of developers wanting to:
- adapt their codes to run on a petascale class system using CPUs and/or GPUs,
- optimize applications that already run on petascale class CPUs or GPUs,
- accelerate machine learning and/or deep learning applications, or
- Implement other innovative uses of petascale class computing systems.
Research teams must apply to participate. The application should include a description of the research code(s) they will bring to the hackathon including the programming languages used, the size of the codes, system software needed, any dependencies such as libraries that are needed, and the names of the attendees along with a bio (max 1 page) for each attendee that primarily includes their programming experience, any previous HPC experience, and their research experience.
The codes are expected to be working so that scaling tests may be conducted, as described below. The codes are not expected to be in production.
Each team that applies will be provided with a training account on the Blue Waters system at least a month ahead of the event to allow time to become familiar with the Blue Waters system. The teams are expected to use the training account to conduct scaling tests on their codes before and during the event.
The scaling tests will help the research teams and the mentors to identify the portions of the codes that require the most attention during the hackathon, to achieve the best outcomes.
A conference call will be conducted after all of the teams have received their training account to talk about how to prepare for the event.
All teams are asked to apply at: https://www.openacc.org/hackathons.
Teams selected to participate will be notified within two weeks of applying.
All participants will have access to the Blue Waters computing system.
For research teams wishing to utilize GPUs and who are new to using GPUs, we recommend starting with OpenACC, but we are open to other GPU programming paradigms (e.g. CUDA, OpenMP4+) for those that have applications that already have some partial GPU port.
The partners in this venture include Blue Waters, the National Center for Supercomputing Applications (NCSA); NVIDIA; and OpenACC.
NCSA will be the host site for the event.
Please send any questions to firstname.lastname@example.org.
Updated: May 1, 2019