Intel packages CPUs, FPGAs as GO to target automated driving systems

January 05, 2017 // By Graham Prophet
Intel has created the Intel GO branding for automated driving products, including the Intel GO In-Vehicle Development Platform for Automated Driving with Intel [formerly Altera] FPGAs sitting alongside either an Intel Atom or Intel Xeon processor. Automotive system supplier Denso has been a collaborator in assembling the package of CPUs, FPGAs and software.

GO is intended to be a scalable, car-to-cloud system as a step towards a ‘driverless future’ (pictured; development system). Information available at time of writing implies a system configuration that continues to employ discrete Intel CPU chips alongside [former Altera] FPGAs. The latter now include an Arria 10 chip available as a automotive-specification part. Intel says this will be the highest performance processing automotive FPGA, offering one speed grade performance advantage over competing devices. There seems, for present at least, no intention to integrate these elements on a single die.

Specifically, the new automotive-grade Arria 10GX mid-range FPGA for autonomous driving is up to 40% lower power than previous generation FPGAs and features hard floating-point digital signal processing blocks. The Arria 10 automotive-grade FPGA offers more than 320 Gbps of DDR4 bandwidth and more than 200 Gbps of transceiver bandwidth. It also supports 12G serial protocols, 1.5K variable-precision DSP blocks and up to 24 transceivers.

 

Comprising hardware and software development kits, the full Intel GO system includes:

  • Two versions of Intel GO In-Vehicle Development Platforms for Automated Driving , delivering unmatched scalability and power-performance optimization. Scaling from next-gen Intel Atom processors to high-performance Intel Xeon processors plus Arria 10 FPGAs , these two platforms provide the computing resource to perform a range of automated driving functions including perception, fusion and decision-making.

  • The Intel GO Automotive 5G Platform offers a 5G-ready platform for the automotive segment allowing car makers to develop and test a wide range of use cases and applications ahead of the expected rollout of 5G in 2020.

  • The Intel GO Automotive SDK provides several key tools specific to the automated driving industry – including deep learning tool kits – and promises a consistent development experience to help engineers maximize hardware capabilities while speeding the pace of design.

Intel further comments that, “The importance of 5G to our self-driving future cannot be overstated. Automated vehicles will both generate