Workload allocation based on downstream thermal impacts

Patent metadata

Field	Value
Publication number	US-10209749-B2
Application number	US-201514740157-A
Country	US
Kind code	B2
Filing date	Jun 15, 2015
Priority date	Jun 15, 2015
Publication date	Feb 19, 2019
Grant date	Feb 19, 2019

Abstract

An apparatus for workload management includes a thermal relationship module, a thermal effect module, and a resource allocation module. The thermal relationship module determines a plurality of thermal relationships among components of an electronic device. Each thermal relationship quantifies a thermal impact on one of the components of utilizing another of the components. The thermal effect module determines one or more potential thermal effects of a workload on the components based on the thermal relationships. The one or more potential thermal effects correspond to one or more possible allocations of resources of at least a portion of the components to run the workload. The resource allocation module selects an allocation of resources from the one or more possible allocations of resources to run the workload, based on the determined potential thermal effects.

First claim

What is claimed is:

1. An apparatus comprising:
a thermal relationship module that determines a plurality of thermal relationships among components of an electronic device by quantifying a thermal impact on one of the components of utilizing another of the components;
a thermal effect module that, for a plurality of possible allocations of resources of at least a portion of the components to run a workload, determines, using predictive modeling, a plurality of potential thermal effects of the workload on the components over time based on the thermal relationships by, determining a primary thermal effect on a first set of components based on the workload utilizing the first set of components and a secondary thermal effect on a second set of components based on heating of the second set of components by the first set of components; and
a resource allocation module that selects an allocation of resources from the plurality of possible allocations of resources to run the workload, based on the determined primary and secondary thermal effects, wherein selecting an allocation of resources comprises allocating the resources,
wherein said modules comprise one or more of hardware circuits, a programmable hardware device, and a processor executing code.

2. The apparatus of claim 1 , further comprising a thermal state module that determines a thermal state of the electronic device using one or more temperature sensors, the thermal state comprising at least one temperature for at least one of the components, wherein the resource allocation module selects the allocation of resources based on the thermal state and the determined potential thermal effects, wherein the thermal state module comprises one or more of hardware circuits, a programmable hardware device, and a processor executing code.

3. The apparatus of claim 2 , further comprising the electronic device, the electronic device comprising the components and the one or more temperature sensors.

4. The apparatus of claim 1 , wherein the components comprise one or more of one or more processors, one or more memory modules, and one or more storage devices.

5. The apparatus of claim 1 , wherein the plurality of thermal relationships comprises a network of thermal relationships between multiple components of the electronic device, wherein at least one of the components links two or more of the thermal relationships.

6. The apparatus of claim 1 , wherein selecting an allocation of resources comprises reallocating resources while the workload is running, to avoid exceeding a thermal limit.

7. The apparatus of claim 1 , wherein selecting an allocation of resources comprises allocating resources for the workload before the workload runs, to distribute a thermal load among the components.

8. The apparatus of claim 1 , wherein determining the thermal relationships comprises computing a thermal relationship based on a topology for the electronic device.

9. The apparatus of claim 1 , wherein determining the thermal relationships comprises utilizing one component and measuring the thermal impact on another component, using one or more temperature sensors.

10. The apparatus of claim 1 , wherein selecting an allocation of resources comprises allocating resources, based on the potential thermal effect over time, to avoid exceeding a thermal limit.

11. The apparatus of claim 1 , wherein selecting an allocation of resources comprises using a workload manager to allocate resources on a component by component basis, wherein the workload manager comprises one or more of hardware circuits, a programmable hardware device, and a processor executing code.

12. A method comprising:
determining a plurality of thermal relationships among components of an electronic device by quantifying a thermal impact on one of the components of utilizing another of the components;
determining, using predictive modeling for a plurality of possible allocations of resources of at least a portion of the components to run a workload, a plurality of potential thermal effects of the workload on the components over time based on the thermal relationships by determining a primary thermal effect on a first set of components based on the workload utilizing the first set of components, and determining a secondary thermal effect on a second set of components based on heating of the second set of components by the first set of components; and
selecting an allocation of resources from the plurality of possible allocations of resources to run the workload, based on the determined primary and secondary thermal effects, wherein selecting an allocation of resources comprises allocating the resources.

13. The method of claim 12 , wherein selecting an allocation of resources comprises reallocating resources while the workload is running, to avoid exceeding a thermal limit.

14. The method of claim 12 , wherein selecting an allocation of resources comprises allocating resources for the workload before the workload runs, to distribute a thermal load among the components.

15. The method of claim 12 , wherein determining the thermal relationships comprises utilizing one component and measuring the thermal impact on another component.

16. The method of claim 12 , wherein determining the potential thermal effects comprises predictively modeling a potential thermal effect over time, and wherein selecting an allocation of resources comprises allocating resources, based on the potential thermal effect over time, to avoid exceeding a thermal limit.

17. A program product comprising a non-transitory computer readable storage medium that stores code executable by a processor, the executable code comprising code to perform:
determining a plurality of thermal relationships among components of an electronic device by quantifying a thermal impact on one of the components of utilizing another of the components;
determining, using predictive modeling for a plurality of possible allocations of resources of at least a portion of the components to run a workload, a plurality of potential thermal effects of the workload on the components over time based on the thermal relationships by determining a primary thermal effect on a first set of components based on the workload utilizing the first set of components, and determining a secondary thermal effect on a second set of components based on heating of the second set of components by the first set of components; and
selecting an allocation of resources from the plurality of possible allocations of resources to run the workload, based on the determined primary and secondary thermal effects, wherein selecting an allocation of resources comprises allocating the resources.

Assignees

• Lenovo Entpr Solutions Singapore Pte Ltd

Inventors

• Cudak Gary D
• Alshinnawi Shareef F
• Holland Jeffrey S
• Ramineni Pradeep

Classifications

• H10W90/00

  Package configurations · CPC title

• H01L25/00

  Electricity · mapped topic

• Y02D10/16

  Cross-Sectional Technologies · mapped topic

• G06F1/206 Primary

  comprising thermal management · CPC title

• Y02D10/00

  Energy efficient computing, e.g. low power processors, power management or thermal management · CPC title