Exploring Deep Neural Network Applications on Theta

The objective of this project is twofold: (1) to learn and explore DNN applications on Theta at ALCF by evaluating two neural networks (AlexNet and ResNet-50), and (2) to analyze the performance (communication and memory) and power consumption of the DNN applications.

The project is conducted by senior undergraduate student Sergio Servantez during the Spring'19 (Jan. 14 - April 26, 2019). This webpage lists their bi-weekly reports.


Bi-weekly reports:
  • report 1 on 2/3 [Link]
  • report 2 on 2/27 [Link]
  • report 3 on 3/6 [Link]
  • report 4 on 3/20 [Link]
  • report 5 on 4/7 [Link]
  • final report on 5/31 [Link]