Habitat - A Runtime-Based Computational Performance Predictor for Deep Neural Network Training

Habitat - A Runtime-Based Computational Performance Predictor for Deep Neural Network Training

USENIX via YouTube Direct link

Intro

1 of 11

1 of 11

Intro

Class Central Classrooms beta

YouTube videos curated by Class Central.

Classroom Contents

Habitat - A Runtime-Based Computational Performance Predictor for Deep Neural Network Training

Automatically move to the next video in the Classroom when playback concludes

  1. 1 Intro
  2. 2 What this talk is about The problem: • Many GPUs available for deep neural network (DNN) training . Each has a different cost and performance
  3. 3 A Cambrian explosion in hardware for training
  4. 4 Choosing a GPU: The paradox of choice
  5. 5 Key observations • Deep learning users may already have an existing GPU
  6. 6 Habitat: A runtime-based performance predictor
  7. 7 One last wrinkle: Kernel-varying operations Wave scaling assumes the same kernel is used across GPUS
  8. 8 Evaluation
  9. 9 How accurate is Habitat?
  10. 10 Rent a GPU in the cloud? Scenario: Want to train GNMT, have access to a P4000. Which cloud GPU to use, if any?
  11. 11 Key takeaways . DNN computation is special (repetitive), enabling new analysis opportunities

Never Stop Learning.

Get personalized course recommendations, track subjects and courses with reminders, and more.

Someone learning on their laptop while sitting on the floor.