ScaleRAN boosts GPU use by sharing resources between network and AI tasks
ScaleRAN: A Runtime for Elastic GPU Resource Management in AI-RAN
Networking and Internet ArchitectureDistributed, Parallel, and Cluster Computing
Summary
Mobile networks use special computers called GPUs to handle both signal processing and AI tasks. The challenge is that these tasks need different amounts of GPU power at different times, and existing methods can't quickly adjust resources without risking delays. The authors introduce ScaleRAN, a software tool that divides GPU power more flexibly between network and AI jobs, adjusting every few milliseconds to keep everything running smoothly. Their tests show this approach can improve AI performance without slowing down network processing.
What this means in practice
- •For telecom infrastructure teams: Dynamically allocate GPUs in base stations to improve AI application use without disrupting network signal processing.
- •For cloud platform operators: Manage GPU resources in Open-RAN cloud deployments to meet strict network timing while running AI workloads efficiently.
Authors
Arthur Hamard, Jérémie Leguay, Gopalasingham Aravinthan, Alberto Conte, Julien Lallet, Abdelkader Outtagarts
Abstract
AI-RAN is driving the convergence of radio access network (RAN) processing and artificial intelligence (AI) workloads on GPU-enabled computing platforms, ranging from GPU-accelerated base stations to cloud-native infrastructures. While GPU acceleration improves the efficiency of virtualized RAN functions, the computational demand of latency-critical baseband processing varies with traffic load, creating transient spare GPU capacity that can be reclaimed by co-located AI applications. Existing GPU-sharing mechanisms rely on coarse-grained hardware partitioning or static resource allocation, limiting their ability to adapt to slot-level workload fluctuations while preserving strict timing requirements. This paper presents ScaleRAN, a node-local runtime for elastic GPU resource management in AI-RAN. ScaleRAN leverages NVIDIA Green Contexts to dynamically partition Streaming Multiprocessors (SMs) between RAN and AI workloads through a ladder of pre-created GPU partitions, enabling low-overhead resource reallocation at RAN slot granularity. The proposed runtime naturally supports Open-RAN (O-RAN) O-Cloud deployments, one representative AI-RAN deployment scenario. To drive these reallocations, we propose a reactive controller inspired by Kubernetes autoscaling and a proactive controller based on quantile regression and conformal prediction. We also develop LDPCSim, a GPU-accelerated workload generator built on NVIDIA Aerial for reproducible evaluation. Results show that predictive fine-grained GPU partitioning satisfies slot-level RAN deadlines while increasing the GPU resources available to co-located AI workloads, demonstrating the practicality of elastic GPU sharing for future AI-RAN platforms.