Nvidia Corporation

Arpit R Jain

Shounak Santosh Deshpande

Andrew Currid

Chenghuan Jia

Sneh Shikhar

Santosh Kumar Ravindranath Shukla

Apparatuses, systems, and techniques to use a graphics processing unit are disclosed. In at least one embodiment, a length of a timeslice for executing work on a virtual machine is determined based, at least in part, on the amount of time prior work performed by a GPU exceeded a prior length of a timeslice. In some embodiments, a work portion and a preemption portion of a timeslice are changed or updated based on an amount of time spent performing prior work and an amount of time spent performing preemption of the prior work.

Adaptive task scheduling for virtualized environments

Systems and methods are disclosed for throttling memory bandwidth accessed by virtual machines (VMs). A technique for dynamically throttling the virtual computer processing units (vCPUs) assigned to a VM (tenant) controls the memory access rate of the VM. When the memory is shared by multiple VMs in a cloud-computing environment, one VM increasing its memory access rate may cause another VM to suffer memory access starvation. This behavior violates the principle of VM isolation in cloud computing. In contrast to conventional systems, a software solution for dynamically throttling the vCPUs may be implemented within a hypervisor and is therefore portable across CPU families and doesn't require specialized server-class CPU capabilities or limit the system configuration.

Growing community of inventors

Maharashtra, India

Arpit R Jain