Accepted Posters
The following posters will be presented at the Poster Session and Reception on Tuesday, February 24, 5:30 pm–7:00 pm. Please note that accepted papers will also present a poster during the Poster Session and Reception.
Courier: High-Performance Persistent Transactional Memory via Cooperative Concurrency Control
Hao Hu, Xinrui Zheng, Yizou Chen, and Xiangyu Zou, Harbin Institute of Technology, Shenzhen; Erci Xu, Shanghai Jiao Tong University and Alibaba Cloud; Hongpeng Wang and Wen Xia, Harbin Institute of Technology, Shenzhen
High-performance storage tier management for container-native AI workloads
Lei Pan, Frank Schmuck, Vasily Tarasov, and Veera Deenadhayalan, IBM Research
ConZone: Practical Zoned Flash Storage Emulation for Consumer Devices
Dingcui Yu, Jialin Liu, and Yumiao Zhao, East China Normal University; Xinghui Duan, Longsys Electronics Co., Ltd; Yina Lv, Xiamen University; Liang Shi, East China Normal University
The Pitfalls of Underspecified Workloads in Benchmarking
John Lewars, John Divirgilio, Frank Schmuck, and Vasily Tarasov, IBM Research
Beyond ARC - Introducing LLM-driven Adaptive Policy Replacement
Zhenjie Sun, Viraj Thakkar, and Qi Lin, Arizona State University; Avani Wildani, Emory University and Cloudflare; Zhichao Cao, Arizona State University
littlefs: rbyd B-trees: Flexible B-trees when RAM << block size
Christopher Haster, littlefs
Less Index, More Speed: Accelerating Large-Scale KVCache Index for Long-Context LLM Serving
Jun Kong, Chang Guo, Zhenyu Zhang, and Zhichao Cao, Arizona State University
Energy-per-Tile: Characterizing the Energy Footprint of AI Workloads on Heterogeneous, Recycled Infrastructure
Wenkai Guan, University of Minnesota, Morris; Yang Zhao, University of Minnesota, Twin Cities; Zhichao Cao, Arizona State University; Tianlong Chen, The University of North Carolina at Chapel Hill; Zishen Wan, Georgia Tech
Reducing Reconstruction I/O via Refactor-Aware Ordering and Tiered Storage
Qi Chen, Temple University; Qirui Tian and Qing Liu, New Jersey Institute of Technology; Xin Liang, University of Kentucky; Xubin He, Temple University