Advancements in Heterogeneous Computing

The field of heterogeneous computing is moving towards increasingly efficient and scalable solutions. Researchers are exploring innovative programming models, such as SYCL, to target a wide range of devices, including GPUs, FPGAs, and accelerators. Other efforts focus on optimizing offload overheads in massively parallel architectures and improving performance in molecular dynamics simulations. Noteworthy papers include Toward Heterogeneous, Distributed, and Energy-Efficient Computing with SYCL, which introduces a SYCL-based high-level programming interface with advanced techniques such as task distribution and energy optimization. Another notable paper is Accelerating Fast Ewald Summation with Prolates for Molecular Dynamics Simulations, which develops a novel method that reduces the size of the fast Fourier transform required for Ewald summation, resulting in better strong scaling with respect to the number of cores.

Sources

Toward Heterogeneous, Distributed, and Energy-Efficient Computing with SYCL

Taming Offload Overheads in a Massively Parallel Open-Source RISC-V MPSoC: Analysis and Optimization

RCOMPSs: A Scalable Runtime System for R Code Execution on Manycore Systems

Comparing Parallel Functional Array Languages: Programming and Performance

Accelerating Fast Ewald Summation with Prolates for Molecular Dynamics Simulations

FLASH: Fast All-to-All Communication in GPU Clusters

KAITIAN: A Unified Communication Framework for Enabling Efficient Collaboration Across Heterogeneous Accelerators in Embodied AI Systems

Built with on top of