Class meeting for 10-405 Computing with GPUs

This is one of the class meetings on the schedule for the course Machine Learning with Large Datasets 10-405 in Spring 2018.

Slides

• ML and GPUs - Powerpoint
• ML and GPUs - PDF

Quiz

• No quiz for today!

Things to Remember

• The types of problems that can be parallized with map-reduce clusters versus GPUs
• SIMD versus MIMD parallization
• The actual process involved in using GPUs for ML problems:
  • Minibatches are generated by the CPU
  • Minibatch data is copied from CPU to GPU
  • Vectorized version of classifier update step is executed on the GPU
  • So the ML weights (the model) and one minibatch are stored on the GPU