Many-core @ DATE: what if the number of cores is too small?
Enschede – March 26, 2014 – In an interactive talk at DATE 2014 Timon ter Braak zooms in on Recore System’s solution to optimally distribute tasks in a many-core system when the limitations on the available resources may have a negative impact on the behavior and performance of the applications. The so-called ‘guided local search’ run-time solution compares favorably to state-of–the art commercial solvers: it is on par for small platforms, scales better with the size of the platform, and has lower memory usage.
In an ideal many-core world, programmers can easily parallelize various computational tasks in an application such that the total workload is nicely balanced over the available cores. In addition, they can distribute the tasks over the cores in such a way that exchanging data between the cores is immediate. We all know that the real world is not ideal. In many-core systems, some tasks take longer than others, and it takes time for data to travel from one core to the other. For predictable execution of any application, an embedded many-core system must ensure in advance that it provides sufficient resources to process tasks, to exchange data, and to control peripherals.
The 'guided local search' algorithm provides an application with a predictable execution environment. The algorithm partitions the available resources of processing elements (cores) and the bandwidth of the interconnect into multiple virtual platforms. The algorithm allows for a hybrid strategy where configurations computed at design-time may be used as references to lower the computational overhead at run-time, and to adapt the configuration to the state of the platform. It compares favorably to state-of–the art commercial solvers: it is on par for small platforms, scales better with the size of the platform, and has lower memory usage.
Timon ter Braak will pitch his presentation during session 6.6 ‘Model-Based Design and Hardware/Software Interfaces’ at DATE on Wednesday, March 26, 2014, 11:00. His interactive presentation is on Wednesday, March 26, 2014 between 16:00 - 16:30. See the DATE website for more information...
More on DATE 2014
The DATE conference addresses all aspects of research into technologies for electronic and embedded system engineering. It covers the design process, test, and automation tools for electronics ranging from integrated circuits to distributed embedded systems and addresses both hardware and embedded software design issues. The conference scope also covers design requirements and new architectures for challenging application fields such as telecoms, wireless communications, multimedia, healthcare, smart energy and automotive systems. http://www.date-conference.com/
About Recore Systems
Recore Systems is a Dutch fabless semiconductor company which develops reliable and fault-tolerant heterogeneous many-core subsystems. Our mission is to make the many-core life easy for both hardware and software developers. We implement our mission in hardware and software by delivering both many-core embedded RTOS and middleware, as well as many-core hardware subsystem IP tailored to our customer’s application needs. Recore Systems is headquartered in Enschede, The Netherlands. To learn more, please visit www.recoresystems.com.
|
Related News
Breaking News
- Frontgrade Gaisler Unveils GR716B, a New Standard in Space-Grade Microcontrollers
- Blueshift Memory launches BlueFive processor, accelerating computation by up to 50 times and saving up to 65% energy
- Eliyan Ports Industry's Highest Performing PHY to Samsung Foundry SF4X Process Node, Achieving up to 40 Gbps Bandwidth at Unprecedented Power Levels with UCIe-Compliant Chiplet Interconnect Technology
- CXL Fabless Startup Panmnesia Secures Over $60M in Series A Funding, Aiming to Lead the CXL Switch Silicon Chip and CXL IP
- Cadence Unveils Arm-Based System Chiplet
Most Popular
- Cadence Unveils Arm-Based System Chiplet
- CXL Fabless Startup Panmnesia Secures Over $60M in Series A Funding, Aiming to Lead the CXL Switch Silicon Chip and CXL IP
- Esperanto Technologies and NEC Cooperate on Initiative to Advance Next Generation RISC-V Chips and Software Solutions for HPC
- Eliyan Ports Industry's Highest Performing PHY to Samsung Foundry SF4X Process Node, Achieving up to 40 Gbps Bandwidth at Unprecedented Power Levels with UCIe-Compliant Chiplet Interconnect Technology
- Arteris Selected by GigaDevice for Development in Next-Generation Automotive SoC With Enhanced FuSa Standards
E-mail This Article | Printer-Friendly Page |