Liang Geng
Liang Geng

Ph.D. Student

About Me

Hi, I’m Liang Geng (耿亮), a third-year Ph.D. student in the Computer Science Department, OSU. I’m interested in computer systems, especially GPU and RDMA programming. My current research is repurposing NVIDIA RT Cores to accelerate non-graphics workloads, such as geospatial data processing. My advisors are Prof. Xiaodong Zhang and Dr. Rubao Lee. Before joining the OSU, I was a senior engineer at Alibaba DAMO Academy supervised by Dr. Wenyuan Yu, where I developed GPU support for libgrape-lite and was the initial contributor to GrapeScope and Vineyard. I was fortunately advised by Prof. Yanfeng Zhang and Dr. Hao Wang while pursuing my Master’s Degree. They helped me to start my research journey.

Interests
  • Geospatial Data Processing
  • Parallel Computing
  • Distributed Systems
Education
  • Ph.D., Computer Science, 2022 - Current

    The Ohio State University, USA

  • M.Eng., Computer Science, 2016 - 2019

    Northeastern University, China

  • B.Eng., Software Engineering, 2012 - 2016

    Liaoning Technical University, China

Recent Publications
(2024). RayJoin: Fast and Accurate Spatial Join with Ray Tracing. In ICS.
(2024). RR-Compound: RDMA-fused gRPC for Low Latency and High Throughput with an Easy Interface. In TPDS.
(2024). Ingress: an automated incremental graph processing system. In VLDB Journal.
(2023). Efficient Multi-GPU Graph Processing with Remote Work Stealing. In ICDE.
(2022). Linking Entities across Relations and Graphs. In ICDE.
(2022). An RDMA-enabled In-memory Computing Platform for R-tree on Clusters. In TSAS.
(2021). Automating Incremental Graph Processing with Flexible Memoization. In VLDB.
(2020). Automating Incremental and Asynchronous Evaluation for Recursive Aggregate Data Processing. In SIGMOD.
(2019). Catfish: Adaptive RDMA-enabled R-Tree for Low Latency and High Throughput. In ICDCS.
(2019). HYPHA: a framework based on separation of parallelisms to accelerate persistent homology matrix reduction. In ICS.
(2019). SEP-graph: finding shortest execution paths for graph processing under a hybrid framework on GPU. In PPoPP.