Section: New Results

Mastering Heterogeneous Platforms

Participants : Cédric Augonnet, Olivier Aumage, Ludovic Courtès, Nathalie Furmento, Andra Hugo, Raymond Namyst, Samuel Thibault, Pierre-André Wacrenier.

• We continued our work on extending StarPU to master exploitation of Heterogeneous Platforms.

• We have extended the StarPU scheduler into managing parallel tasks which permit a better exploitation of CPUs and load balancing with GPUs.

• We have designed over StarPU a lightweight DSM over MPI, which permits to seamlessly execute StarPU applications over an MPI cluster of GPU-enhanced nodes.

• We have been developing a GCC plug-in which extends the C language with pragmas and attributes that make writing StarPU applications a lot easier.

• We have brought to StarPU support for automatically converting data between CPU and GPU formats (typically arrays of structures vs structures of arrays). We are now optimizing it.

• We have added an OpenCL interface to StarPU , SOCL [42] , which permits to execute unmodified OpenCL applications over StarPU .

• We have introduced in StarPU theoretical bound support [27] : from a record of the set of tasks submitted by the application, StarPU uses linear programming to give the execution time of an ideal scheduling, which can then be compared with the actual results.

• We have continued collaboration with the University of Tennessee, Knoxville for StarPU support in the state-of-the art dense linera algebra library, Magma, in particular LU [26] and QR [27] factorizations. We have also collaborated with the University of Mons [41] and Linköping [32] .

• Cédric Augonnet defended his PhD on StarPU  [17] .