TVM For Fun and Profit Tutorial, at FCRC 2019

June 22nd 2019: Phoenix Convention Center Room 104B, Phoenix, AZ


About

TVM is an open-source deep learning compiler stack for CPUs, GPUs, and specialized accelerators. It aims to close the gap between the productivity-focused deep learning frameworks, and the performance or efficiency-oriented hardware backends.

We are excited to hold a tutorial on TVM and showcase the different ways TVM can be used to facilitate research on deep learning systems, compilers, and hardware architectures. We welcome TVM contributors, potential users, collaborators, researchers, and practitioners from the broader community.

The presentations will be structured to provide a high-level overview on the research. We will complement the presentations with hands-on tutorials that can be run on your own laptop during, or after the tutorial.

Key Takeaways

  • Overview of the full TVM deep learning stack from frameworks, compilers, down to hardware.
  • Bringing compiler support to new deep learning accelerators.
  • Leveraging Machine Learning for automated program optimization.
  • Future trends in software hardware co-design.


Program

Browse the tentative program to learn more about exciting use cases for TVM. In addition, we will host “TVM Office Hours” during the breaks for those of you who have technical questions about TVM.

Time Talk Colab Notebook Presenter
8:15 Continental Breakfast    
9:00 Opening Remarks   Luis Ceze (UW)
9:05 TVM Stack Overview (clip)   Tianqi Chen (UW)
9:45 Relay: A Programming Language for Deep Learning (clip) nb Jared Roesch (UW)
10:30 AutoTVM: Automatically Optimize any Workload on any Hardware (clip) nbA nbB nbC Eddie Yan (UW)
11:00 Morning Break / TVM Office Hours    
11:20 VTA: Open Source Deep Learning Acceleration (clip) nbA nbB Thierry Moreau (UW)
12:00 TSIM: Cycle Accurate Simulation for Custom HW in TVM (clip) nb Luis Vega (UW)
12:30 Lunch    
2:00 Optimizing CNN Model Inference using Amazon SageMaker Neo (clip)   Yida Wang and Zhi Chen (Amazon AWS)
2:50 uTVM: Deep Learning on the Low-Power Edge (clip) nb Logan Weber and Pratyush Patel (UW)
3:20 Afternoon Break / TVM Office Hours    
4:00 Ultra Low Precision Operators (clip) nb Meghan Cowan (UW)
4:30 Bring Your Own Datatypes: Enabling Number Format Exploration with TVM (clip) nb Gus Smith (UW)
5:00 Closing Remarks (clip)   Luis Ceze (UW)
5:30 TVM Office Hours    

Follow us on Twitter


Register

You can register for the TVM tutorial at FCRC via the FCRC registration link below. Register by May 24 to receive the early pricing rate!


Click to Register


Hotels


Reserve your Hotel Room