cd /news/artificial-intelligence/every-microsecond-matters-achieving-… · home topics artificial-intelligence article
[ARTICLE · art-66037] src=arxiv.org ↗ pub= topic=artificial-intelligence verified=true sentiment=↑ positive

Every Microsecond Matters:Achieving Near SpeedOfLight Latency in GPU Collectives

A new study from researchers at an undisclosed institution demonstrates GPU collective communication kernels that reduce latency to within 7% of the hardware Speed-of-Light lower bound, achieving substantial improvements for small and medium messages. Building on NCCL's device-side API, the work introduces barrier-free synchronization and symmetric collectives that accelerate long-context LLM inference and cuSOLVERMp, addressing latency bottlenecks in both AI and HPC workloads.

read2 min views5 publishedJul 20, 2026
Every Microsecond Matters:Achieving Near SpeedOfLight Latency in GPU Collectives
Image: source
[Submitted on 17 Jul 2026]


[View PDF](/pdf/2607.16100)

[HTML (experimental)](https://arxiv.org/html/2607.16100v1)

Abstract:GPU collective communication is typically optimized for bandwidth, yet many emerging workloads are increasingly limited by latency. Long-context decode-heavy large language model (LLM) inference is a prime example, where serving large models requires multiple GPUs, and many small collectives lie directly on the critical path of token generation. Therefore, even microsecond of overhead can impact performance and cost. In this work, we study how to approach the hardware Speed-of-Light (SoL) lower bound for GPU collectives within a scale-up network. We identify key principles for near-optimal designs, including barrier-free synchronization and efficient use of symmetric memory and multicast. Building on NCCL's device-side API, we develop low-latency interfaces for constructing custom collective kernels and use them to implement new symmetric collectives in NCCL. Microbenchmarks show substantial latency reductions for small and medium messages, reducing overhead to within 7% of the absolute SoL lower bound. When integrated into real applications, these kernels improve inter-token latency and throughput in LLM inference and accelerate cuSOLVERMp, demonstrating benefits for both AI inference and traditional HPC workloads.

References & Citations

...

Bibliographic Explorer

(What is the Explorer?) Connected Papers

(What is Connected Papers?) Litmaps

(What is Litmaps?) scite Smart Citations

(What are Smart Citations?)# Code, Data and Media Associated with this Article alphaXiv

(What is alphaXiv?) CatalyzeX Code Finder for Papers

(What is CatalyzeX?) DagsHub

(What is DagsHub?) Gotit.pub

(What is GotitPub?) Hugging Face

(What is Huggingface?) ScienceCast

(What is ScienceCast?)# Demos Influence Flower

(What are Influence Flowers?) CORE Recommender

(What is CORE?)# arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

── more in #artificial-intelligence 4 stories · sorted by recency
── more on @nccl 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/every-microsecond-ma…] indexed:0 read:2min 2026-07-20 ·