Central office based virtual personal computer

摘要

A virtual personal computer is implemented in a communication system comprising a plurality of central offices each of which communicates with a plurality of client devices over a corresponding access network. A given one of the central offices comprises at least one compute server and at least one storage server. The virtual personal computer is configured by allocating physical processing resources of the compute server and physical storage resources of the storage server to that virtual personal computer. User access is provided to the virtual personal computer via one of the client devices. The virtual personal computer can be dynamically reconfigured by altering the allocation of at least one of the physical processing resources and the physical storage resources to the given virtual personal computer responsive to particular applications selected by the user to run on the given virtual personal computer.

主权项

1. A method of implementing virtual personal computers in a communication system, the communication system comprising a plurality of central offices each of which communicates with a plurality of client devices over a corresponding access network, the method comprising the steps of:
providing in a given one of the central offices at least one compute server and at least one storage server; configuring a given one of the virtual personal computers by allocating physical processing resources of the compute server and physical storage resources of the storage server to that virtual personal computer; and providing access to the given virtual personal computer to a user via a corresponding one of the client devices; wherein the compute server comprises a plurality of interconnectable graphics processing unit (GPU) hardware subcomponents, the compute server configuring a particular GPU processing pipeline for the given virtual personal computer from selected ones of said subcomponents; and wherein a frame buffer associated with the GPU processing pipeline is allocated such that it utilizes an address space of a processor core that runs a corresponding virtual machine, but is physically located internal to an integrated circuit comprising at least a portion of the GPU hardware subcomponents.