Detailed Description

Google docs:

ParallelTaskAssigner computes target parallel task counts for all HLOs in the module, then assigns parallel task counts to HLOs in the entry computation, or to HLOs in embedded computations invoked by (potentially nested) kWhile or kCall instructions. Each HLO which is assigned parallel task counts is outlined into its own embedded computation, which is compiled as a parallel compute function, and which is invoked from a kCall instruction that is lowered in codegen to a runtime parallel fork/join call.

The documentation for this class was generated from the following files:

tensorflow/tensorflow/compiler/xla/service/cpu/parallel_task_assignment.h
tensorflow/tensorflow/compiler/xla/service/cpu/parallel_task_assignment.cc