Abstract:Coarse-grained reconfigurable architectures (CGRAs) are increasingly employed as domain-specific accelerators due to their efficiency and flexibility. A CGRA typically relies on compilers to perform task scheduling. The longstanding problem of static scheduling is that it suffers from insufficient parallelism in handling irregularities due to over-serialization and workload imbalance, which leads to severe resource underutilization and performance loss. To counteract the limitations of static scheduling in CGRAs, it is essential to exploit dynamic parallelism automatically and manage hardware resources adaptively. However, existing dynamic scheduling mechanisms, e.g., work stealing, often reschedule aggressively for instant performance but sacrifice efficiency, which is unfavorable to CGRAs that emphasize efficiency and fewer reconfigurations. This article proposes an elastic task scheduling scheme that enables lightweight dynamic scheduling in CGRAs. Tasks are rescheduled at runtime according to the classic tagged-token dataflow paradigm to enable dynamic task-level parallelism. Meanwhile, tasks are dynamically resized according to run-time throughputs via duplication, combination, and substitution operators for balanced multitask execution. We implement the elastic task scheduling scheme on a well-known reconfigurable architecture - triggered instruction architecture (TIA). Evaluation on the MachSuite benchmarks shows that the proposed scheme is effective in improving performance and energy efficiency. The average speedup is 2× over the baseline. Also, our design attains a 57 percent improvement in the area-normalized performance and a 49 percent better energy efficiency. Compared with a state-of-the-art dynamic scheduling method, our scheme achieves 1.6× speedup and 1.6× energy efficiency than work-stealing mechanism on the same substrate.

The construction and management of template library for coarse-grained reconfigurable system

Compilation approach for dynamically reconfigurable architecture based on grouped tasks

A Template Mapping Algorithm For Reconfigurable Computing

Temporal task partitioning algorithm for dynamically reconfigurable systems

An Elastic Task Scheduling Scheme on Coarse-Grained Reconfigurable Architectures

Template-based Compilation for Coarse-Grained Reconfigurable Processor

Algorithms of Resource Management for Reconfigurable Systems Based on Hardware Task Vertexes

TRAM: an Open-Source Template-based Reconfigurable Architecture Modeling Framework.

Fast On-Line Real-Time Scheduling Algorithm for Reconfigurable Computing.

Mapping Optimization Of Affine Loop Nests For Reconfigurable Computing Architecture

THRAM: A Template-based Heterogeneous CGRA Modeling Framework Supporting Fast DSE.

Temporal Partitioning Algorithm for a Coarse-Grained Reconfigurable Computing Architecture

On-line Scheduling of Real-Time Tasks for Reconfigurable Computing System

Combining Memory Partitioning and Subtask Generation for Parallel Data Access on CGRAs

A Graph Covering Method for Template Based System Partition

A Modeling and Mapping Method for Coarse/fine Mixed-Grained Reconfigurable Architecture

A method to build reconfigurable architectures by extracting common subgraphs

Optimizing Spatial Mapping of Nested Loop for Coarse-Grained Reconfigurable Architectures

Affine Transformations for Communication and Reconfiguration Optimization of Loops on Cgras

Trigger-Centric Loop Mapping on CGRAs

Battery-Aware Task Mapping for Coarse-Grained Reconfigurable Architecture