Redwan Khan created a novel approach, SHADE, to significantly speedup deep learning training (DLT). SHADE exploits DLT behaviors for enabling fundamental I/O cacheability. More details in his USENIX FAST’23 paper available at https://t.co/PNx4kip5k5
@StackCSVT @VT_CS
Congratulations to DSSL alum @yuecheng87 on receiving IEEE-CS TCHPC Early Career Researchers Award for Excellence in High Performance Computing https://t.co/vtromAcy66
@StackCSVT @VT_CS@VT_alumni
Jingoo Han introduces a novel tokenized incentive approach to improve participant engagement and efficiency for Federated Learning in his IEEE Cloud 2022 paper. https://t.co/PNx4ki7tVv
Hadeel Albahar is designing techniques to efficiently support deep learning applications on heterogeneous GPUs. Her IEEE/ACM CCGrid 2022 paper describes a new workload scheduler for this. Read more at
https://t.co/PNx4kip5k5
Arnab K. Paul's upcoming IEEE HiPC paper shows how application-agnostic file system statistics can provide a deep understanding of HPC application I/O behavior. #hipc2020 Read more at
https://t.co/qwWMB3ChQc
Container storage costs got you down? Unleash Nannan Zhao’s DupHunter on your Docker registries to rein them in. Find out more in our upcoming USENIX ATC’20 paper at https://t.co/PNx4kip5k5
Subil Abraham's upcoming IEEE Cloud (#ieeecloud2020) paper quantitatively analyzes the I/O efficiency of Docker/Podman/Singularity/Charliecloud containers in HPC storage. Find out more at https://t.co/qwWMB3ChQc
(2/2) HPC storage manages more data than ever. How do you even find the information? Arnab K. Paul (with Cray Inc.) has a solution, BRINDEXER, which provides efficient metadata indexing and search in HPC storage.
Find out more in our CCGrid’20 paper: https://t.co/qwWMB3ChQc
(1/2) Scheduling deep leaning workloads on HPC systems is hard, especially if we want to utilize the multiple available GPUs. Jingoo Han has designed MARBLE, a practical solution to this problem.
Read more about it in our upcoming CCGrid’20 paper at https://t.co/qwWMB3ChQc
DSSL receives $1.3M NSF SPX award in collaboration with @yuecheng87 to develop cross-stack memory optimizations for boosting I/O performance of Deep Learning HPC applications. Congrats team! https://t.co/Al63ESbQ1V
DSSL has three papers at #CLUSTER2019. Jingoo Han's first-of-its-kind quantitative study of deep learning training on supercomputers; Nannan Zhao's analysis of Docker Hub registry; and Arnab K. Paul's tool for HPC file systems monitoring. Congratulations!! https://t.co/qwWMB3ChQc
Bharti Wadhwa has invented a new load balancing system to give a much-needed performance boost to HPC storage systems. Read about it in our @IPDPS 2019 paper. https://t.co/qwWMB3ChQc
Luna Xu’s upcoming IEEE Cluster 2018 paper describes innovative heterogeneity-aware task scheduling for Spark, especially for HPC workloads. Congratulations!!
https://t.co/5HgbDz55xi
Ali Anwar’s upcoming FAST’18 paper presents comprehensive analysis and insights into production-level container registry workloads and data center design implications. Paper and tools @ https://t.co/sc2R5hMhaL
@VT_CS
Congratulations to Luna Xu for her paper on scaling up data parallel analytics platforms, and to Arnab Paul on his paper on I/O load balancing for HPC centers, to appear in IEEE Big Data.
https://t.co/qwWMB3kGrC