Skip to content

yunoJ/pytorch

 
 

Repository files navigation

PyTorch-P2P (GPU-SSD team)

This repository is DNN training system with SSD as augmented memory of GPU based on PyTorch. Final submission plan of this research is NSDI 2020.

Step (Motivation)

  • TensorFlow-serving system with vDNN (Latency?)
  • Spark ML with vDNN (Throughput of vDNN?)
  • Computing utilization with throughput graph
  • Memory usage breakdown

Step (Implementation)

  • Implementing vDNN on PyTorch
  • Implementing ARCP2P on PyTorch
  • Porting CycleGAN / BERT
  • Performance optimization

License

PyTorch-P2P is BSD-style licensed, as found in the LICENSE file (same as original PyTorch).

About

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • C++ 67.2%
  • Python 18.0%
  • Cuda 9.3%
  • C 3.5%
  • CMake 1.0%
  • Objective-C++ 0.4%
  • Other 0.6%