Scalable platform meets demand for high computing performance

December 17, 2020 //By Christoph Hammerschmidt
Scalable platform meets demand for high computing performance
Processing the massive amounts of data required for highly automated driving at SAE levels 2+ to 5 requires a steep increase in computing power in vehicle computers. The advent of centralised computing platforms also calls for very highly integrated, extremely complex multi-core chip architectures. With its new R-Car V3U SoC, Renesas aims to address these requirements.

The R-Car V3U is the first product based on Renesas'R-Car architecture Gen 4 of the open and flexible Renesas autonomy platform for ADAS and AD. With this SoC, the platform offers scalability from NCAP entry-level applications to highly automated driving systems. For this purpose, the R-Car V3U is equipped with various hardware accelerators for connectivity, security and functional safety. Hardware accelerators also boost the performance of deep learning algorithms to a breathtaking 60 TOPS (Tera Operations per Second). This is sufficient even for complex tasks such as obstacle detection and classification.

A DSP for processing radar signals, a multi-threading computer vision engine for traditional computer vision algorithms, image signal processing to improve image quality and additional hardware accelerators for important algorithms such as dense optical flow, stereo disparity and object classification round off the equipment and ensure a computing power of up to 96. 000 DMIPS - sufficient reserves for demanding tasks, even for those that will only come to the computer platform in the future, because Renesas assumes that these platforms will be commissioned with additional tasks via OTA updates over their lifetime.


Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.