A new technical paper titled “Mind the Memory Gap: Unveiling GPU Bottlenecks in Large-Batch LLM Inference” was published by researchers at Barcelona Supercomputing Center, Universitat Politecnica de Catalunya, and IBM Research.
Abstract
“Large language models have been widely adopted across different tasks, but their auto-regressive generation nature often leads to inefficient resource utilization during inference. While batching is commonly used to increase throughput, performance gains plateau beyond a certain batch size, especially with smaller models, a phenomenon that existing literature typically explains as a shift to the compute-bound regime. In this paper, through an in-depth GPU-level analysis, we reveal that large-batch inference remains memory-bound, with most GPU compute capabilities underutilized due to DRAM bandwidth saturation as the primary bottleneck. To address this, we propose a Batching Configuration Advisor (BCA) that optimizes memory allocation, reducing GPU memory requirements with minimal impact on throughput. The freed memory and underutilized GPU compute capabilities can then be leveraged by concurrent workloads. Specifically, we use model replication to improve serving throughput and GPU utilization. Our findings challenge conventional assumptions about LLM inference, offering new insights and practical strategies for improving resource utilization, particularly for smaller language models.”
Find the technical paper here. March 2025.
Recasens, Pol G., Ferran Agullo, Yue Zhu, Chen Wang, Eun Kyung Lee, Olivier Tardieu, Jordi Torres, and Josep Ll Berral. “Mind the Memory Gap: Unveiling GPU Bottlenecks in Large-Batch LLM Inference.” arXiv preprint arXiv:2503.08311 (2025).
Leave a Reply