Contention-Aware GPU Thread Block Scheduler for Efficient GPU-SSD
- 7 July 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Computer Architecture Letters
- Vol. 24 (2) , 257-260
Abstract
The growing dataset sizes in LLM have made low-cost SSDs a popular solution for extending GPU memory in mobile devices. In this paper, we introduce CA-Scheduler, a contention-aware scheduling scheme for GPU-initiated SSD access. The key insight behind CA-Scheduler is leveraging the BSP GPU programming model, which allows reordering work at the thread block level to optimize SSD throughput. By capitalizing on the predictable memory access patterns of GPU thread blocks, CA-Scheduler anticipates SSD locations to minimize contention and improve performance.Keywords
Funding Information
- NSF (PPoSS 2316176)
- research grant from Samsung Electronics
This publication has 5 references indexed in Scilit:
- LLM in a flash: Efficient Large Language Model Inference with Limited Memory2024
- Locality-Centric Data and Threadblock Management for Massive GPUs2020
- ZnG: Architecting GPU Multi-Processors with New Flash for Scalable Data Analysis2020
- CODAACM Transactions on Architecture and Code Optimization, 2018
- Performance Evaluation of Dynamic Page Allocation Strategies in SSDsACM Transactions on Modeling and Performance Evaluation of Computing Systems, 2016