From Markus Levy
President

February 2016
Un-subscribe.

The Multicore Association®

NEWSLETTER


Topics

  • New Working Group to Manage and Expand OpenAMP Framework
  • Video: Technology and Application Trends Driving the OpenAMP Standard
  • OpenAMP Demonstrations at Embedded World
  • Mentor Graphics Upgrades from Working Group Member to Executive Board Membership
  • Welcome Express Logic and Micrium as New Working Group Members
  • Multicore Activities from University of Houston and University of Delaware

New Working Group to Manage and Expand OpenAMP Framework

Join the Multicore Association OpenAMP working group, chaired by Tomas Evensen (CTO Embedded Software at Xilinx). It’s targeted at the expansion and standardization of the Open Asymmetric Multi Processing Framework (OpenAMP). This framework manages systems with multiple operating systems and compute elements - allowing lifecycle control of heterogeneous cores in an SoC - the boot, shut down, loading, and software execution. The original OpenAMP project is released as Open Source and is available under BSD license. This framework, with first-generation implementations available in the Linux distribution, is supported by MCA members Express Logic, Mentor Graphics, Micrium, NXP, Wind River, and Xilinx. According to Sven Brehmer, (MCA CTO), “OpenAMP provides a complement to the MCA’s current specifications, helping to ensure platform coverage from low- to high-end embedded systems.”

Check out articles on the subject:

Video: Technology and Application Trends Driving the OpenAMP Standard

Adopting heterogeneous multicore SoCs across the industry is challenging traditional approaches to embedded software development, however The Multicore Association has established the OpenAMP working group to standardize a framework to address heterogeneous multicore development. This panel of Multicore Association, Mentor Graphics, and Xilinx experts discuss the role of the OpenAMP standard to support applications, markets and technology trends addressing disruptive industry changes and requirements for secure and safe applications in systems. http://go.mentor.com/4s4ld

OpenAMP Demonstrations at Embedded World

Join Mentor Graphics (booth 4-422) and Xilinx (booth 1-205) for live demonstrations of OpenAMP in action. No appointments necessary.
Mentor Graphics Upgrades from Working Group Member to Executive Board Membership
Mentor Graphics has been an MCA working group member since 2006; recently they have upgraded to the Executive board member level to help set the consortium’s strategic direction. The Mentor Embedded Multicore Framework, utilizing the MCA OpenAMP standard, allows developers to configure and deploy multiple operating systems and applications across homogeneous or heterogeneous multicore processors. The solution is available today on the latest SoCs from vendors including Altera, NXP, TI, and Xilinx.

Welcome Express Logic and Micrium as New Working Group Members

The Multicore Association welcomes both Express Logic and Micrium as working group members, specifically to participate in the new Open Asymmetric Multi Processing (OpenAMP) working group. With its ThreadX/SMP, Express Logic already has been utilizing the Multicore Association’s MCAPI with interprocessor communications support from PolyCore Software. Micrium has already implemented an initial version of OpenAMP, which enables Micrium's product portfolio to target high-end multicore designs combining a variety of high-level operating systems and RTOSes using this portable cross-platform API over a standardized protocol.

Multicore Activities from University of Houston and University of Delaware

University of Houston and University of Delaware have been developing an open-source software stack targeting embedded platforms. Software portability is a primary concern with embedded systems and most solutions are proprietary requiring programmers to change the source code every time the target platform changes. Consequently, maintaining a single source code is almost impossible. This university project aims to provide suitable solutions to this persisting challenge.

Traditional C/C++ code can be parallelized using OpenMP. An OpenMP compiler translates the parallel code to multithreaded code consisting of function calls to a customized runtime library. The project’s strategy translates an OpenMP code to MCA libraries, thus exploiting the underlying resources using MRAPI, communicating between cores using MCAPI, and executing tasks using MTAPI. This enables portability and scalability, yet maintaining a single code base. This implementation was evaluated using MiBench, Rodinia, NAS, and BOTS benchmark applications, and it was observed that results were either closer to or better than proprietary customized runtime libraries. Platforms used for this work include traditional x86, Freescale (NXP) PowerPC platforms (8-24 cores), and NVIDIA’s TK1 (ARM + DSP).

For further info and published papers: https://github.com/MCAPro2015/OpenMP_MCA_Project. The universities are looking forward to extending the work to explore load balancing and scheduling tasks on heterogeneous systems using an OpenMP-task based model and MTAPI.


If you do not wish to receive e-mail from The Multicore Association®, you can un-subscribe. MCA sends no more than one e-mail per month to registered users at www.multicore-association.org. Continuing your subscription ensures you'll be notified when events and other important announcements are available.