5.3k citations
- Chinese Academy of SciencesCN241 papers
- Chinese University of Hong KongHK84 papers
- University of Chinese Academy of SciencesCN55 papers
- Southern University of Science and TechnologyCN27 papers
- University of Hong KongHK24 papers
- Shenzhen UniversityCN19 papers
- University of MacauMO19 papers
- Hong Kong University of Science and TechnologyHK18 papers
- Chinese University of Hong Kong, ShenzhenCN15 papers
- Nanyang Technological UniversitySG14 papers
- Peking UniversityCN14 papers
- Shanghai Jiao Tong UniversityCN14 papers
18 papers · 1 filter
BanaServe: Unified KV Cache and Dynamic Module Migration for Balancing Disaggregated LLM Serving in AI Infrastructure
Yiyuan He, Minxian Xu, Jingfeng Wu +7
Large language models (LLMs) are increasingly deployed in AI infrastructure, driving the need for high throughput, resource efficient serving systems. Disaggregated LLM serving, wh…
FlexPipe: Adapting Dynamic LLM Serving Through Inflight Pipeline Refactoring in Fragmented Serverless Clusters
Yanying Lin, Shijie Peng, Chengzhi Lu +2
Serving Large Language Models (LLMs) in production faces significant challenges from highly variable request patterns and severe resource fragmentation in serverless clusters. Curr…
BrownoutServe: SLO-Aware Inference Serving under Bursty Workloads for MoE-based LLMs
Jianmin Hu, Minxian Xu, Kejiang Ye +1
In recent years, the Mixture-of-Experts (MoE) architecture has been widely applied to large language models (LLMs), providing a promising solution that activates only a subset of t…
Edge AI: A Taxonomy, Systematic Review and Future Directions
Sukhpal Singh Gill, Muhammed Golec, Jianmin Hu +12
Edge Artificial Intelligence (AI) incorporates a network of interconnected systems and devices that receive, cache, process, and analyze data in close communication with the locati…
Exploiting Scratchpad Memory for Deep Temporal Blocking: A case study for 2D Jacobian 5-point iterative stencil kernel (j2d5pt)
Lingqi Zhang, Mohamed Wahib, Peng Chen +4
General Purpose Graphics Processing Units (GPGPU) are used in most of the top systems in HPC. The total capacity of scratchpad memory has increased by more than 40 times in the las…
Revisiting Temporal Blocking Stencil Optimizations
Lingqi Zhang, Mohamed Wahib, Peng Chen +4
Iterative stencils are used widely across the spectrum of High Performance Computing (HPC) applications. Many efforts have been put into optimizing stencil GPU kernels, given the p…