.png){kind=icon}
.png){kind=icon}
Return
Scalable OpenMP Remote Offloading via Asynchronous MPI and Coroutine-Driven Communication
.png){kind=icon}
PRE
AI
DOI：10.1007/978-3-031-99872-0_18.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Heterogeneous multi-node clusters with accelerators, such as GPUs, are increasingly the standard for HPC, where the MPI+OpenMP approach is commonly used for application development. However, this approach poses significant challenges for developers, especially in managing communication, synchronization, and load balancing across distributed nodes and accelerators. To address these challenges, this paper proposes the MPI Proxy Plugin (MPP), an extension of the LLVM OpenMP Offloading runtime that transparently offloads OpenMP target regions to remote accelerators via MPI. By abstracting communication and using the asynchronous mechanisms of MPI with C++20 coroutines, MPP provides a scalable alternative to MPI+OpenMP, enabling simpler development of heterogeneous HPC applications through the familiar OpenMP programming model. Experimental results show that MPP achieves excellent scalability. For compute-intensive proxy applications, it scales nearly linearly, reaching.63x speedup from 1 to 64 GPUs. While naive data transfers can degrade performance, this research reveals that extending OpenMP Target with collective operations (e.g., multi-device broadcast) simplifies development and improves performance, achieving up to.10x speedup in communication-bound benchmarks.
Keywords:
HPC
OpenMP
MPI
Remote Offloading
GPUs
Journal
E
IF:
0
Papers:
21
Citations:
0