In this paper we describe the problem of achieving composability of independently developed real-time subsystems to be executed on a multi-core platform, and we provide a solution to tackle it. We evaluate existing work for achieving real-time predictability on multi-cores and illustrate their lack with respect to composability. To address composability we present a multi-resource server-based scheduling technique to provide predictable performance when composing multiple subsystems on a shared multi-core platform. To achieve composability on multi-core platforms, we propose to add memory bandwidth as an additional server resource. Tasks within our multi-resource servers are guaranteed both CPU- and memory bandwidth; thus the performance of a server will become independent of resource usage by tasks in other servers. We are currently implementing multi-resource servers for the Enea OSE operating system for a Freescale P4080 8-core processor, to be tested with software for a 3G-basestation.
No takes yet. Share an insight, caveat, or question.
Behnam et al. (2013) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: