The browser version you are using is not recommended for this site.
Please consider upgrading to the latest version of your browser by clicking one of the following links.

OpenFabrics Enterprise Distribution (OFED*)

The OpenFabrics Alliance develops, tests, licenses, supports, and distributes OFED software, which is the open source software foundation for High Performance Computing (HPC) clusters. Stock OFED forms the core for the enhanced Intel for OFED+* software offering, and Intel provides support for stock OFED-only solutions used with HPC applications.

Building on the rigorously tested industry standard OFED, Intel will provide inter OFED release revisions for bug fix and feature additions that will be rolled into the next OFED release. This enhanced software is fully compatible with ongoing OFED releases and includes tools that accelerate the benefits of HPC investments by simplifying installation and configuration of any size HPC cluster. Performance Scaled Messaging (PSM) which is included in OFED brings a purpose-designed, lightweight, connectionless architecture that was specifically created to optimize HPC communications. For HPC clusters to reach Exascale performance, using a general-purpose InfiniBand* driver layer is an unnecessary performance impediment. PSM is well positioned for the future evolution of HPC networks.

The PSM library uses an HPC-optimized lightweight software (SW) driver layer, roughly 10% of the footprint of standard InfiniBand SW drivers. In addition, the Intel® True Scale Fabric suite also supports standard I/O-focused protocols via the standard verbs layer.

View or download the product brief ›

PSM highlights

  • Low end-to-end latency
  • High message passing interface (MPI) message rate, effective bandwidth, and collectives performance
  • Accelerated MPI performance for the most widely available Intel® MPI Libraries
  • Included in OFED software releases
  • Increased application performance and scalability
  • Coprocessor/GPU support (Intel® Xeon Phi™ product family, NVIDIA Tesla*)
  • Congestion control architecture that reduces the effects of fabric congestion caused by low credit availability that can result in head of line blocking
  • Full native vFabric Suite* support is for all Intel MPI Libraries supported by PSM

Related Content

Related Videos