The System Virtualization Platform is a set of system services and language interfaces for the exploitation of concurrency on many-core processor chips. SVP can be used with any multicore chip with general-purpose cores, but is oriented towards heterogeneous platforms with support for per-core hardware multithreading.

While SVP also targets current architectures, one of the target hardware platforms for SVP is the Microgrid, an idealized chip architecture with hardware multithreading and dataflow scheduling. The Microgrid has an elastic design which can be used to build systems at any scale, from embedded microcontrollers to entire datacenters.

SVP is a joint work of the CSA group at the University of Amsterdam, UTIA, the University of Hertfordshire and other partners.

What does it mean for you?

As an external IT observer, you might be interested to understand SVP to obtain a better conceptual control over parallel systems of the future: hybrid chips containing various components and programming models.

As a computer scientist, you might want to understand how the replacement of traditional Von Neumann computing with SVP-like Hybrid Dataflow has an impact at all levels of system design, and how to deal with it practically.

As a student in computer science, you might be interested to try out our simulation environment and programming tools to learn about dataflow computing and fine-grained concurrency management.