ECOSCALE : A Novel Framework for Utilising Multi-FPGAs in HPC Systems

Abstract

ECOSCALE implements a scalable programming environment and architecture, aiming to substantially reduce energy consumption as well as data traffic and latency. ECOSCALE introduces a novel heterogeneous energy-efficient hierarchical architecture, as well as a hybrid programming environ-ment and runtime system. The ECOSCALE approach is hierarchical and it scales well by partitioning the physical system into multiple independent Work-ers (i.e. compute nodes). Workers are interconnected in a tree-like fashion and define a contiguous global address space that can be viewed either as a set of partitions in a Partitioned Global Address Space (PGAS), To further increase energy efficiency, as well as to provide resilience, the Workers employ recon-figurable accelerators mapped into the virtual address space utilizing a dual stage System Memory Management Unit with coherent memory access. The implemented UNILOGIC architecture supports shared partitioned reconfigura-ble resources accessed by any Worker in a PGAS partition, as well as automat-ed hardware synthesis of these resources from an OpenCL-based programming model.