Placement optimization for virtualized graphics processing
Methods, systems, and computer-readable media for placement optimization for virtualized graphics processing are disclosed. A provider network comprises a plurality of instance locations for physical compute instances and a plurality of graphics processing unit (GPU) locations for physical GPUs. A GPU location for a physical GPU or an instance location for a physical compute instance is selected in the provider network. The GPU location or instance location is selected based at least in part on one or more placement criteria. A virtual compute instance with attached virtual GPU is provisioned. The virtual compute instance is implemented using the physical compute instance in the instance location, and the virtual GPU is implemented using the physical GPU in the GPU location. The physical GPU is accessible to the physical compute instance over a network. An application is executed using the virtual GPU on the virtual compute instance.
1. A system, comprising:
one or more processors and memory storing program instructions executable by the one or more processors to:
select in a network, based at least in part on one or more placement criteria, a graphics processing unit (GPU) location of a physical GPU and an instance location for a physical compute instance, wherein the network comprises a plurality of instance locations for physical compute instances and a plurality of GPU locations for physical GPUs; and
provision a virtual compute instance and a virtual GPU attached to the virtual compute instance, wherein the virtual compute instance is implemented using the physical compute instance in the instance location, wherein the virtual GPU is implemented using the physical GPU in the GPU location, and wherein the physical GPU is accessible to the physical compute instance over the network.
2. The system as recited in claim 1 , wherein the one or more placement criteria are based at least in part on a performance metric associated with the use of the physical GPU by the physical compute instance.
3. The system as recited in claim 1 , wherein the one or more placement criteria are based at least in part on a cost metric associated with the use of the physical GPU by the physical compute instance.
4. The system as recited in claim 1 , wherein the GPU location or the instance location is selected based at least in part on network locality.
5. A computer-implemented method, comprising:
selecting in a network, based at least in part on one or more placement criteria, a graphics processing unit (GPU) location for a physical GPU or an instance location for a physical compute instance, wherein the network comprises a plurality of instance locations for physical compute instances and a plurality of GPU locations for physical GPUs;
provisioning a virtual compute instance and a virtual GPU attached to the virtual compute instance, wherein the virtual compute instance is implemented using the physical compute instance in the instance location, wherein the virtual GPU is implemented using the physical GPU in the GPU location, and wherein the physical GPU is accessible to the physical compute instance over the network.
6. The method as recited in claim 5 , wherein the one or more placement criteria comprise improvement of one or more metrics.
7. The method as recited in claim 5 , wherein the one or more placement criteria are based at least in part on a performance metric associated with use of the physical GPU by the physical compute instance.
8. The method as recited in claim 5 , wherein the one or more placement criteria are based at least in part on a cost metric associated with use of the physical GPU by the physical compute instance.
9. The method as recited in claim 5 , wherein the one or more placement criteria are based at least in part on an energy metric associated with use of the physical GPU by the physical compute instance.
10. The method as recited in claim 5 , wherein the GPU location or the instance location is selected based at least in part on network locality.
11. The method as recited in claim 5 , wherein the GPU location is selected based at least in part on network latency between the physical GPU and a client device.
12. The method as recited in claim 5 , wherein provisioning the virtual compute instance and the virtual GPU comprises:
migrating the virtual compute instance from an existing physical compute instance in an existing instance location to the physical compute instance in the instance location; or
migrating the virtual GPU from an existing physical GPU instance in an existing GPU location to the physical GPU in the GPU location.
13. A non-transitory computer-readable storage medium storing program instructions computer-executable to perform:
selecting in a network, based at least in part on one or more placement criteria, a graphics processing unit (GPU) location for a physical GPU and an instance location for a physical compute instance, wherein the network comprises a plurality of instance locations for physical compute instances and a plurality of GPU locations for physical GPUs; and
provisioning a virtual compute instance and a virtual GPU attached to the virtual compute instance, wherein the virtual compute instance is implemented using the physical compute instance in the instance location, wherein the virtual GPU is implemented using the physical GPU in the GPU location, and wherein the physical GPU is accessible to the physical compute instance over the network.
14. The non-transitory computer-readable storage medium as recited in claim 13 , wherein the one or more placement criteria are based at least in part on one or more metrics.
15. The non-transitory computer-readable storage medium as recited in claim 13 , wherein the one or more placement criteria are based at least in part on a performance metric associated with use of the physical GPU by the physical compute instance.
16. The non-transitory computer-readable storage medium as recited in claim 13 , wherein the one or more placement criteria are based at least in part on a cost metric associated with use of the physical GPU by the physical compute instance.
17. The non-transitory computer-readable storage medium as recited in claim 13 , wherein the one or more placement criteria are based at least in part on an energy metric associated with use of the physical GPU by the physical compute instance.
18. The non-transitory computer-readable storage medium as recited in claim 13 , wherein the GPU location and the instance location are selected based at least in part on network locality.
19. The non-transitory computer-readable storage medium as recited in claim 13 , wherein the GPU location is selected based at least in part on network latency between the physical GPU and a client device.
20. The non-transitory computer-readable storage medium as recited in claim 13 , wherein provisioning the virtual compute instance and the virtual GPU comprises:
migrating the virtual compute instance from an existing physical compute instance in an existing instance location to the physical compute instance in the instance location; or
migrating the virtual GPU from an existing physical GPU instance in an existing GPU location to the physical GPU in the GPU location.