GPI-2

Right on time for ISC’14, a new version of GPI-2 (v1.1.0) is released. The new version includes the following new features: Non-blocking collectives Threads integration Integrated MPI Interoperability Asynchronous topology building Focus on  large scale execution GPU (CUDA) integration Load Leveler support In this post, we go through some of these new features, why we think…

The Fraunhofer ITWM will be present at the ISC’14 in Leipzig, Germany. This is a good opportunity to know more about GPI-2 and its features. The GPI-2 team would like to invite you to come over to booth 860 (Fraunhofer booth). Here is the floor plan of the ISC’14 exhibition to know how can you…

The slides used in the GASPI tutorial are available here. You can also use the menu bar and under Documentation, a sub-menu Tutorial is now available.

One more tutorial will be held at the LRZ (Leibiz Supercomputing Centre) on Thursday, June 5 2014, 9:00-13:00. More information here.  

A tutorial on GASPI will be held at HLRS, on the 30. January 2014. This is a good opportunity to get to know the GASPI specification and get some hands-on experience with GPI-2. Abstract In this tutorial we present an asynchronous dataflow programming model for Partitioned Global Address Spaces (PGAS) as an alternative to the…

A small update has been done to GPI-2. This version (v1.0.1) includes minor fixes and a dynamic look-up of devices. You can find this version in the Download page.

Last June (2013), Mellanox released a white paper (link) where the GPI-2 support for heterogeneous architectures, namely Intel’s Xeon Phi, is highlighted: “Over the last decade, specialized heterogeneous hardware designs ranging from Cell over GPGPU to Intel Xeon Phi have become a viable option in High Performance Computing mostly due to the fact that these heterogeneous…