Dynamic allocation of physical graphics processing units to virtual machines

A non-transitory computer-readable medium stored in a server computing system containing content for dynamically allocating physical graphics processing unit (GPU) resources of at least one physical GPU among a plurality of virtual machines by performing a method comprising:

• determining a processing capacity of each physical GPU resource;
• determining a GPU benefit factor for each of the plurality of virtual machines, the GPU benefit factor determined based upon an assessment of benefit to the virtual machine of using a GPU;
• and
• dynamically allocating the physical GPU resources among the plurality of virtual machines by assigning the physical GPU resources to one or more of the plurality of virtual machines based upon the determined processing capacity of each physical GPU and the determined priority and GPU benefit factor of each of the plurality of virtual machines, wherein the GPU benefit factors of one or more of the virtual machines are adjusted or set while running the one or more virtual machines, based upon one or more of a dynamic assessment of workload indicating potential benefit from use of a GPU resource, scaling of workloads, and/or determination of runtime profiles.

A server computing system, comprising:

• a memory;
• a computer processor;
• and
• virtualization logic, stored in the memory, and configured, when executed on the computer processor, to:
• assess a processing capacity of one or more physical GPUs associated with the server computing system;
• determine a GPU benefit factor for each of the plurality of virtual machines, wherein the GPU benefit factor for at least one of the plurality of virtual machines is adjusted or set after running the at least one virtual machine;
• and
• dynamically allocate the physical GPUs among the plurality of virtual machines by:
• rank ordering the plurality of virtual machines according to the GPU benefit factor of each virtual machine in the plurality of virtual machines;
• determining a subset of the rank ordered virtual machines that can run on the physical GPUs based upon the assessed capacities of the one or more physical GPUs as compared to requirements of some of the plurality of virtual machines in rank order;
• and
• allocating a virtual GPU to each virtual machine in the determined subset.