一句话：提出 MSCCL++ 方法论，用底层原语、高层 DSL 与算法库实现高性能可移植 GPU 通信内核。
AI 应用需在快速演进的异构硬件上运行，通用库支持滞后，程序员自建通信栈虽快但易错且不可移植。MSCCL++ 提出高性能可移植通信内核的设计方法论：底层原语接口暴露最少硬件抽象并隐藏同步与一致性复杂性；高层 DSL 供开发者实现负载特定通信算法；算法库实现标准集合通信 API。评估显示，集合通信几何平均加速 1.7×（最高 5.4×），AI 推理负载加速 1.2×（最高 1.38×）。它已在 Microsoft Azure 生产使用并被 AMD RCCL 采用；经验表明其抽象稳健，multimem 等新硬件特性可在数周内支持。
AI applications increasingly run on fast-evolving, heterogeneous hardware to maximize performance, but general-purpose libraries lag in supporting these features. Performance-minded programmers often build custom communication stacks that are fast but error-prone and non-portable. This paper introduces MSCCL++, a design methodology for developing high-performance, portable communication kernels. It provides (1) a low-level, performance-preserving primitive interface that exposes minimal hardware abstractions while hiding the complexities of synchronization and consistency, (2) a higher-level DSL for application developers to implement workload-specific communication algorithms, and (3) a library of efficient algorithms implementing the standard collective API, enabling adoption by users with minimal expertise. Compared to state-of-the-art baselines, MSCCL++ achieves geomean speedups of 1.7× (up to 5.4×) for collective communication and 1.2× (up to 1.38×) for AI inference workloads. MSCCL++ is in production of multiple AI services provided by Microsoft Azure, and has also been adopted by RCCL, the GPU collective communication library maintained by AMD. MSCCL++ is open source and available at https://github.com/microsoft/mscclpp. Our two years of experience with MSCCL++ suggests that its abstractions are robust, enabling support for new hardware features, such as multimem, within weeks of development.
DOI 原文 · arXiv · PDF（开放获取） · DBLP
@inproceedings{Hwang0DJMMSSYLR26,
title = {MSCCL++: Rethinking GPU Communication Abstractions for AI Inference},
author = {Changho Hwang and Peng Cheng and Roshan Dathathri and Abhinav Jangda and Saeed Maleki and Madan Musuvathi and Olli Saarikivi and Aashaka Shah and Ziyue Yang and Binyang Li and Caio Rocha and Qinghua Zhou and Mahdieh Ghazimirsaeed and Sreevatsa Anantharamu and Jithin Jose},
booktitle = {ASPLOS},
year = {2026},
doi = {10.1145/3779212.3790188}
}