READ ME

Please refer to our report for a detailed explanation of our project, including the following sections:

• Introduction
• Background
  • Neural Network Inference
  • Traditional OS Scheduling
• Design
  • Client
  • Scheduler
• Evaluation
  • Meeting SLOs Better
  • Tradeoffs and Limitations
• Related Work
  • Benefits of Overlapping CPUs and GPUs
  • Overview of the Triton Server and Client
• Work Distribution

Acknowledgements

Thanks to course staff of CS 744 Big Data Systems for their guidance and blueprint for this project.