Cross-Layer and Multi-Objective Programming Approach for Next Generation Heterogeneous Parallel Computer Systems
Funding programme H2020
Coordinator The Open Group
Partners Easy global market, GMV, University of Stuttgart, University of York, Unparallel, Wings
Descriptions The overall objective of the PHANTOM project is to deliver an integrated cross-layer (hardware and system software/programming environment), multi-objective and cross-application approach that will enable next generation heterogeneous, parallel and low-power computing systems, while hiding the complexity of computing hardware from the programmer, thus fostering productivity in programming.
More specifically, the PHANTOM system comprises a hardware-agnostic software platform over reconfigurable multi-core and heterogeneous (GPU, FPGA, CPU) hardware platforms. On the one hand, the software platform will offer the means for multi-dimensional optimization. Hence, a multi-objective scheduler decides - dynamically and cost-effectively - where in the computing continuum (e.g. cloud, embedded systems, mobile devices, desktops, data centres), at which cross-layer system level (analog, digital, hybrid analog-digital, software) and on which heterogeneous technology (GPU, FPGA, CPU) to execute each part of an applications, as well as orchestrate dynamically the hardware and software components of reconfigurable hardware platforms.
Security and runtime monitoring mechanisms support the cross-layer system operation, while programmer- and productivity-oriented software tools are provided. APIs per application class are proposed facilitating programmers along with a parallelization toolset with emphasis on performance analytics and data locality to increase performance and reduce the power cost of data movement and ubiquitous Big Data towards the so called “Near-Data Computing”.
The heterogeneous hardware platforms will offer the required abstraction and configuration mechanisms to enable the seamless migration and execution of application components, while having energy efficiency, performance guarantees and cost effectiveness as intrinsic features.
Specific industrial case studies will be utilised to validate the effectiveness of the project technologies within the Telecom, Surveillance and Higher Performance Computing domains.
Intecs The Brainware company