Automatic Generation of Parallelized Multicore Code
Use eMBP® to generate parallel code for multi/manycore systems, working with models from MathWorks® Simulink® and/or C source code. eMBP is optimized for the latest processor technologies that offer excellent performance and power efficiency.
Its Model-Based Development (MBD) approach maximizes design quality and final system reliability: particularly valuable when creating high performant and real-time applications.
Freedom to Choose Hardware
Based on open standards, and with no vendor locks, eMBP lets you freely choose the optimum multi-/manycore processor. After analysing the application, it estimates the performance based on your chosen processor’s architecture and performance data.
Performance estimates use the SHIM standard, which was recently formalized as IEEE Std 2804-2019.
Transforming the Multicore Programming
eMBP delivers superior efficiency compared to other model-based parallelizers. Working with the Simulink block diagram, at a higher level of abstraction than the C source code, eMBP ensures the resulting parallelism in the control algorithm closely meets the designer's intentions. Techniques include critical-path analysis, block structure extraction, and grouping and mapping blocks to cores. An intuitive display tool lets expert users visualize the grouped blocks to control the output.
Learn more
Help to Enter the Parallel Universe
Simulink is used in automotive for real-time control systems such as engine-, steering- and brake control.
When migrating from single core to multicore systems, eMBP can statically assign Simulink function blocks to individual cores and generate parallel code.
The eMBP reference application based on Renesas RH850 multicore processors enables performance estimation on a processor-in-the-loop simulation (PILS). The following multicore RH850 series are already supported: C1, F, P1, E2.