TEXT: Towards EXaflop applicaTions
Massive computing systems will be essential to Europe's future competitiveness in science, engineering and business.
Top systems are now reaching the Petaflop barrier, and the new challenge is to understand how to implement applications to meet the Exaflop target. Multicore chips are already here, but over the next decade they will grow to several hundreds of cores, and future exascale systems will consist of hundreds of thousands of nodes based on these manycore chips.
The TEXT project built on the belief that the programming model is the key component to support high productivity and efficient use of an Exaflop system. A hybrid model consisting of MPI + SMPSs can be demonstrated today and can show the way to Exascale. The SMPSs model, which uses directives to describe tasks and their data dependencies, provides the necessary support for asynchrony and heterogeneity as well as enabling incremental parallelisation, modularity and portability of applications. By integrating it within MPI we can propagate SMPSs' characteristics to the global application level. This is also a way to leverage and provide a smooth migration path for the huge number of applications today written in MPI.
The focus of the TEXT project was to install the MPI + SMPSs environment at the HPC facilities of several partners and demonstrate how it can be used to improve seven real and relevant applications/libraries. The codes were selected for their importance to each of their scientific communities and fall in the areas of: basic linear algebra libraries; geophysics; plasma physics; engineering and molecular dynamics.
EPCC provided support for external users and developed synthetic benchmarks to be used to ensure the quality of MPI+SMPSs implementations. Other activities included improving the support for the MPI+SMPSs environment, developing debugging and profiling tools, and working on asynchronous algorithms. TEXT also engaged external HPC users by providing access to MPI+SMPSs installations, together with in-depth support and training.
- Deploying the MPI + SMPSs programming model infrastructure on production HPC systems
- Porting applications to MPI + SMPSs
- Providing support and consultancy to external users
- Developing the MPI + SMPSs programming environment, including debugging and performance analysis
- Improving the SMPSs runtime
- Investigating asynchronous algorithms and load balancing techniques
- Promoting the SMPSs model and supporting standardization efforts
- Training and dissemination
For further information visit the TEXT project website