← back to search

Xiaoyi Lu

University of California System · US
🔎 Find collaborators in Computer Networks and Communications · Information Systems →
Search 5.9M scientists by topic, h-index, country & funding — free.
Area of research
Computer Networks and Communications · Information Systems
Research interest
Research topics from publications: gZCCL: Compression-Accelerated Collective Communication Framework for GPU Clusters; An Optimized Error-controlled MPI Collective Framework Integrated with Lossy Compression; hZCCL: Accelerating Collective Communication with Co-Designed Homomorphic Compression. Representative work: GPU-aware collective communication has become a major bottleneck for modern computing platforms as GPU computing power rapidly rises. A traditional approach is to directly integrate lossy compression into GPU-aware collectives, which can lead to serious performance issues such as underutilized GPU devices and uncontrolled data distortion. In order to address these issues, in this paper, we propose gZCCL, a first-ever general framework that designs and optimizes GPU-aware, compression-enabled collectives with an accuracy-aware design to control error propagation. To validate our framework, we evaluate the performance on up to 512 NVIDIA A100 GPUs with real-world applications and datasets. Exp With the ever-increasing computing power of supercomputers and the growing scale of scientific applications, the efficiency of MPI collective communications turns out to be a critical bottleneck in large-scale distributed and parallel processing. The large message size in MPI collectives is particularly concerning because it can significantly degrade the overall parallel performance. To address this issue, prior research simply applies the off-the-shelf fix-rate lossy compressors in the MPI collectives, leading to suboptimal performance, limited generalizability, and unbounded errors. In this paper, we propose a novel solution, called C-Coll, which leverages error-bounded lossy compression t
h-index
citations
0
works
0
NIH funding
primary concept
email

Recent publications

gZCCL: Compression-Accelerated Collective Communication Framework for GPU Clusters
2024cited by 17position: middledoi
An Optimized Error-controlled MPI Collective Framework Integrated with Lossy Compression
2024cited by 16position: middledoi
hZCCL: Accelerating Collective Communication with Co-Designed Homomorphic Compression
2024cited by 6position: middledoi

Grants

No grants ingested yet.

Frequent collaborators

Kai Zhao · University of Alabama at Birmingham3 papers (2024–2024)Jinyang Liu · University of California, Riverside3 papers (2024–2024)Jiajun Huang · University of California, Riverside3 papers (2024–2024)Rajeev Thakur · Argonne National Laboratory3 papers (2024–2024)Yanfei Guo · Argonne National Laboratory3 papers (2024–2024)Yujia Zhai · University of California, Riverside3 papers (2024–2024)Franck Cappello · University of Iowa3 papers (2024–2024)Zizhong Chen · Shandong University3 papers (2024–2024)Xiaodong Yu · University of Illinois Urbana-Champaign2 papers (2024–2024)Ken Raffenetti · Argonne National Laboratory2 papers (2024–2024)Sheng Di · University of California, Riverside2 papers (2024–2024)Zizhe Jian · University of California, Riverside1 papers (2024–2024)Yafan Huang · University of Iowa1 papers (2024–2024)Hui Zhou · Soochow University1 papers (2024–2024)Zhaorui Zhang · Hong Kong Polytechnic University1 papers (2024–2024)Xiaodong Yu · City University of Hong Kong1 papers (2024–2024)Xin Liang · University of Kentucky1 papers (2024–2024)Sheng Di · University of California, Riverside1 papers (2024–2024)Hui Zhou · Yan'an University1 papers (2024–2024)
Looking for a research collaborator?
Search millions of scientists by field, institution, impact, and funding status — see their work, find their email, and reach out directly.
Find collaborators in Computer Networks and Communications · Information Systems →