A specification of the machines and devices that will be allocated to run the ComputeGraph.
An allocator of shared latches that are guaranteed to produce correct results even with an out-of-order command queue.
An allocator of shared latches that are guaranteed to produce correct results even with an out-of-order command queue.
Produces a set of shared latches guaranteed to work on any kernel launch order that preserves the semantics of the calculation. For an OpenCL out-of-order queue, the kernels are allowed to run in an order unrelated to the order in which the kernels were enqueued, although ordering established by event linking is preserved.
An allocator of shared latches that are guaranteed to produce correct results even with an out-of-order command queue.
An allocator of shared latches that are guaranteed to produce correct results even with an out-of-order command queue.
Produces a set of shared latches guaranteed to work on any kernel launch order that preserves the semantics of the calculation. For an OpenCL out-of-order queue, the kernels are allowed to run in an order unrelated to the order in which the kernels were enqueued, although ordering established by event linking is preserved.
Holds the single shared FieldRegister for a set of kernels
Created by carteric on 10/15/14.
This package handles all allocation of compute resources for a computation.