1 / 10

PaPP ̶ Developing Future Proof Parallel Software

PaPP ̶ Developing Future Proof Parallel Software. Mats Brorsson, Professor, Swedish Institute of Computer Science. Oct 26, Paris, France. ARTEMIS Joint Undertaking The public private partnership for R&D actors in embedded systems. This is PaPP. Predictable performance, for

evelyn
Download Presentation

PaPP ̶ Developing Future Proof Parallel Software

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. PaPP ̶ Developing Future Proof Parallel Software Mats Brorsson, Professor, Swedish Institute of Computer Science Oct 26, Paris, France ARTEMIS Joint UndertakingThe public private partnership for R&D actors in embedded systems

  2. This is PaPP • Predictable performance, for • Parallel applications, on • Heterogeneous embedded manycores

  3. PaPP -- Consortium

  4. Technological Innovation • Multiple application domains • Task-centric parallelism • Separation of concerns • Enables better resource management • Multi-level, co-operative scheduling and resource management • Hypervisor • OS(s) • RTS:s • Demonstrator of feasibility for multiple domains and architectures Domain A Domain B Domain C DSL:s Framework Standard languages Run-time systems Run-time systems OS1 OS2 Hypervisor Arch 1 Arch 2 Arch 3

  5. Approach • Revisit the entire software stack: • Programming model • Performance modelling tools • System software (RTS, OS) • Hardware abstraction • With the purpose of enabling performance predictability and portability

  6. Overall objective • Achieve predictable performance portability of software when moved from current platforms to future parallel platforms Tle-GX100 Nvidia CARMA

  7. Sub-objectives • Efficient and accurate performance models • Demonstrated performance portability across platforms • Demonstrated portability of the software stack across application domains • Vastly improved resource usage and management • Reusable parallel software components • Improved software developer productivity

  8. Market Innovation • For the space application domain • Verify ongoing ESA features • For the mobile communication domain • Improve the power/performance trade-offs • Cutting time of (re)certification • In the multimedia domain • Reuse of media processing IP:s on wider set of heterogeneous processors

  9. Next steps • Yearly integration of tool chains • Workshops and tutorials • Demonstrators • Contribution to ARTEMIS Tools platforms

  10. Thank you for your attention! Mats Brorsson, matsbror@sics.se ARTEMIS Joint Undertaking

More Related