Join the Multicore revolution!


[Background] [Kista Multicore Center] [Swedish Multicore Iniative] [Projects] [Concrete degree projects] [Acknowledgments]

Background

Semiconductor technology limitations prohibit a continued increase in performance of single-threaded microprocessors. Instead, performance improvements will mainly come in the form of multicore processors, or chip-multiprocessors (CMP) where several processors, or cores, cooperate.

The ability to efficiently utilize multicore processors in products wil be critical for software intensive industries. The multicore technology makes possible higher performance, new valuable functionalities and lower power consumption. According to the semiconductor industry roadmap, the number of cores on a chip is expected to grow with 40% per year. In 2008 we have commercial chips with 4-64 cores so the prediction leads to 20-350 cores by 2013 and 100-1850 cores by 2018, only ten years from now. Assuming that each core remains as powerful in the next generation as in the current, the performance increase is a factor of 5x by 2013 and 25x by 2018.

Multicore processors require parallel software. Future proof software has to automatically utilize a growing number of cores. This is a paradigm shift for all software development. There is a risk that legacy software will stop working on multicore platforms even though they might be written with concurrent threads since in many cases these threads are written to assume exclusive access to the processor when it is executing and context switches occur only at very well defined places in the code. Furthermore, parallel software development is difficult and error-prone. Current state-of-the-art development methodologies for parallel software are two to three times as expensive as those for sequential programming. Cost-effective development of parallel software for the competitive products of the future is a gigantic challenge requiring new methods and tools.

Kista Multicore Center

SICS, KTH, and Uppsala University are forming a joint Multicore Computing Research Center in Kista. The center will provide world-leading solutions to, and expertise on, the multicore computing challenges that face the software-intensive systems industry.

In a five year perspective, by 2013, the center will demonstrate scalable, safe, and engineering efficient solutions to the challenge of leveraging future (2013-2018) multicore technology for the software-intensive system products of center partners. This strategic research program is further detailed in the SSF proposal Scalability and Programmability in the Manycore Era.

Strategic research projects will be complemented with projects addressing near term challenges of industry partners. The center will work as a seamless extension of partner R&D efforts, exploiting the unique expertise of center researchers in commissioned projects. Near term and strategic research are highly synergistic. Near term projects guide strategic projects by tracking and reassessing industry needs and requirements. Strategic projects guide near term projects by setting necessary targets for strategic technology decisions.

The center researchers have exceptionally strong backgrounds in parallel computing. Prof. Brorsson led the first open source implementation of OpenMP, the industry standard for shared-memory parallel computing, and has recently developed energy saving methods for multicore. Prof. Hagersten has a background as chief architect of high-end parallel servers at Sun Microsystems, and has recently developed tools for performance optimization of multicore memory use. Prof. Haridi has led award-winning research at SICS on parallel programming systems and parallel computer architectures, and has recently developed event-based software architectures for multicore.

Together, the center researchers achieve the critical mass to:

The center is active in international research collaborations. The center researchers are partners in the European Network of Excellence HiPEAC, the premier European network for multicore research, and will explore opportunities offered by the EU research program.

To its industry partners, the center will be able to offer:

Read more about the center at: http://www.sics.se/groups/multicore

Swedish Multicore Initiative

The Swedish Multicore Initiative is a concerted effort to address the engineering and strategic issues related to multicore processor technology for the software intensive systems industry in Sweden. The Initiative ties together all parties interested in forwarding this technology with the main objective of drastically reducing the cost of software production for multicores, ideally making it as easy as for single-core systems.

The vision of the Initiative is to make multi/many-core microprocessor technology as easy to use for the Swedish software intensive industry as single-core microprocessors.

Objectives of the Initiative therefore include:

We beleive that this can only be done by a focussed collaboration between industrial and academic organisations.  To facilitate this, the Initiative will form a virtual center which acts as a one-stop shop for the highest competence in utilizing multi/many-core technology.  This center could be seen as  a Swedish counterpart to international industrial/academic partnerships such as the one at UC Berkely/University of Illinois UC (Microsoft, Intel US $20 M) and at Stanford University (AMD, HP, Intel, NVidia and Sun US $6 M). The center naturally connects to international competence networks through its members. One example is the link to the HiPEAC Network of Excellence supported by EU FP7.

Activities

The Swedish Multicore Initiative will launch a number of activities to meet its objectives:

Advisor and steering committees

The Swedish multicore initiative is coordinated by a steering committee consisting of representatives of the founding members (BTH, Chalmers, KTH, Mälardalen university, SICS, and Uppsala university). The steering committee seeks advice from an international advisory committee with members from important multicore research communities.

Read more about the Swedish Multicore Initiative here: http://www.multicore.se

Projects

Here is a list of currently ongoing projects related to multicore activities:

Previous related projects:

Concrete degree projects

See  this page for examples of concrete degree projects. Contact Mats Brorsson if you want more information.

Some degree projects, including some industrial projects, may also be announced here: http://www.sics.se/groups/multicore.


Acknowledgment

The work in these projects is graciously supported by the following organizations:

Updated 2009-02-13