OSDI 2022

Meta Info

16th USENIX Symposium on Operating Systems Design and Implementation

Homepage: https://www.usenix.org/conference/osdi22

Paper list: https://www.usenix.org/conference/osdi22/technical-sessions

Papers

Automated Parallelism for DNN Training

• Alpa: Automating Inter- and Intra-Operator Parallelism for Distributed Deep Learning [Code]

• Unity: Accelerating DNN Training Through Joint Optimization of Algebraic Transformations and Parallelization [Code]

Scheduling for DNN Training

• Looking Beyond GPUs for DNN Scheduling on Multi-Tenant Clusters [Code]

Model Serving

• Orca: A Distributed Serving System for Transformer-Based Generative Models [Personal Notes]

• Microsecond-scale Preemption for Concurrent GPU-accelerated DNN Inferences [Code, Benchmark, Artifact]

Sparse Models

• SparTA: Deep-Learning Model Sparsity via Tensor-with-Sparsity-Attribute [Code]

DL Compiler

• ROLLER: Fast and Efficient Tensor Compilation for Deep Learning [Code]

Collaborative ML

• Walle: An End-to-End, General-Purpose, and Large-Scale Production System for Device-Cloud Collaborative Machine Learning [Code, 中文官网]