Oral
Automatically batching control-intensive programs for modern accelerators
Alexey Radul · Brian Patton · Dougal Maclaurin · Matthew Hoffman · Rif A. Saurous

Tue Mar 3rd 08:25 -- 08:50 AM @ Ballroom A

We present a general approach to batching arbitrary computations for accelerators such as GPUs. We show orders-of-magnitude speedups using our method on the No U-Turn Sampler (NUTS), a workhorse algorithm in Bayesian statistics. The central challenge of batching NUTS and other Markov chain Monte Carlo algorithms is data-dependent control flow and recursion. We overcome this by mechanically transforming a single-example implementation into a form that explicitly tracks the current program point for each batch member, and only steps forward those in the same place. We present two different batching algorithms: a simpler, previously published one that inherits recursion from the host Python, and a more complex, novel one that implemenents recursion directly and can batch across it. We implement these batching methods as a general program transformation on Python source. Both the batching system and the NUTS implementation presented here are available as part of the popular TensorFlow Probability software package.

Author Information

Alexey Radul (Google)
Brian Patton (Google Inc.)
Dougal Maclaurin (Google Inc.)
Matthew Hoffman (Google)
Rif A. Saurous (Google)

Related Events (a corresponding poster, oral, or spotlight)