Title: Configuring the Linux Scheduler to Meet the Latency Requirements of ADAS/AD Workloads
Proposal: There is a perception in the automotive industry that Linux is not an "RTOS" and can therefore not meet the latency requirements of ADAS and autonomous driving high-performance compute workloads. The goal of the initiative is to define best practices and scheduler configurations that allow Linux to meet these requirements.
How it supports ELISA: ELISA working groups will benefit from a prescriptive approach for leveraging the Linux OS with challenging latency requirements
Tangible Deliverables: A reference configuration and experimental data to prove that Linux can meet the demanding latency requirements of ADAS/AD workloads.
What's preventing progress: Lack of contributors, Lack of technical expertise, Missing infrastructure
Being worked on by other project/community: OSADL has been working on the problem of Linux latency for years, but their approach is more academic and not rooted in real industry use-cases.
Funding: Minimal (1-2 FTE)
If not pursued, what opportunities/benefits/capabilities will be lost: Automotive (and maybe others) OEMs will continue to use proprietary operating systems and ignore Linux for use in their products.
Title: Configuring the Linux Scheduler to Meet the Latency Requirements of ADAS/AD Workloads
Proposal: There is a perception in the automotive industry that Linux is not an "RTOS" and can therefore not meet the latency requirements of ADAS and autonomous driving high-performance compute workloads. The goal of the initiative is to define best practices and scheduler configurations that allow Linux to meet these requirements.
How it supports ELISA: ELISA working groups will benefit from a prescriptive approach for leveraging the Linux OS with challenging latency requirements
Tangible Deliverables: A reference configuration and experimental data to prove that Linux can meet the demanding latency requirements of ADAS/AD workloads.
What's preventing progress: Lack of contributors, Lack of technical expertise, Missing infrastructure
Being worked on by other project/community: OSADL has been working on the problem of Linux latency for years, but their approach is more academic and not rooted in real industry use-cases.
Funding: Minimal (1-2 FTE)
If not pursued, what opportunities/benefits/capabilities will be lost: Automotive (and maybe others) OEMs will continue to use proprietary operating systems and ignore Linux for use in their products.