Virtual graphics processing unit scheduling based preemption overhead

A system comprising:

• one or more processing units to:
• identify a plurality of virtual processing units associated with a physical processing unit, wherein each of the plurality of virtual processing units is associated with a virtual machine of a plurality of virtual machines that execute one or more processing tasks on respective virtual processing units in round-robin order during respective allocated predefined discrete time slices;
• obtain, for a first virtual machine of the plurality of virtual machines, a first overhead time value associated with the first virtual machine using a first virtual processing unit of the plurality of virtual processing units;
• obtain, for a second virtual machine of the plurality of virtual machines, a second overhead time value associated with the second virtual machine using a second virtual processing unit of the plurality of virtual processing units;
• determine, based on the first overhead time value associated with the first virtual machine, whether the second overhead time value associated with the second virtual machine satisfies a compensation threshold criterion, wherein the compensation threshold criterion is satisfied if a difference between the second overhead time value associated with the second virtual machine and the first overhead time value associated with the first virtual machine is greater than a respective assigned execution time period of the second virtual machine;
• and
• responsive to determining that the second overhead time value associated with the second virtual machine satisfies the compensation threshold criterion, cause allocation of another predefined discrete time slice to the second virtual machine after a previous predefined discrete time slice of the second virtual machine based on the difference between the second overhead time value associated with the second virtual machine and the first overhead time value associated with the first virtual machine prior to allocating a predefined discrete time slice to any other of the plurality of virtual machines.

A non-transitory computer-readable storage medium comprising instructions that, when executed by a processing device, cause the processing device to perform operations comprising:

• for each run of a first virtual machine of a plurality of virtual machines on a first virtual processing unit of a plurality of virtual processing units associated with a physical processing unit for an assigned predefined execution time, adding, by a hypervisor an overhead time value associated with a respective run to a first total overhead time value maintained by the hypervisor for the first virtual machine;
• for each run of a second virtual machine of the plurality of virtual machines on a second virtual processing unit of the plurality of virtual processing units associated with the physical processing unit for the assigned predefined execution time, adding, by the hypervisor, an overhead time value associated with a respective run to a second total overhead time value maintained by the hypervisor for the second virtual machine;
• prior to a run of a next virtual machine, determine a difference between the first total overhead time value for the first virtual machine and the second total overhead time value for the second virtual machine;
• responsive to determining that the difference exceeds the assigned predefined execution time, causing a run of the first virtual machine to be repeated prior to running the next virtual machine by allocating the assigned predefined execution time to the first virtual machine after the assigned predefined execution time of a previous run of the first virtual machine based on the difference between the first total overhead time value for the first virtual machine and the second total overhead time value for the second virtual machine.