Automatic localization of acceleration in edge computing environments

1. An edge computing device in an edge computing system, comprising:

acceleration circuitry;

processing circuitry; and

a memory device comprising instructions stored thereon, wherein the instructions, when executed by the processing circuitry, configure the processing circuitry to perform operations to:

obtain first telemetry information that indicates availability of the acceleration circuitry to execute a function;

obtain second telemetry information that indicates of a remote acceleration resource to execute the function, the remote acceleration resource located at a remote location in the edge computing system that is remote from the edge computing device;

identify an estimated time to execute the function at the acceleration circuitry or the remote acceleration resource, based on evaluation of the first and second telemetry information;

identify an urgency level for execution of the function; and

select use of the acceleration circuitry or the remote acceleration resource, to execute the function on a workload, based on identification of the estimated time to execute the function at the remote acceleration resource or the acceleration circuitry in relation to a service level agreement and the urgency level, wherein use of the acceleration circuitry or the remote acceleration resource includes:

in response to a determination that the remote acceleration resource can execute the entire function based on the service level agreement and the urgency level faster than the acceleration circuitry, causing the remote acceleration resource to execute the function;

in response to a determination that the acceleration circuitry can execute the entire function based on the service level agreement and the urgency level faster than the remote acceleration resource, causing the acceleration circuitry to execute the function; and

in response to a determination that the remote acceleration resource can execute a portion of the function based on the service level agreement and the urgency level faster than the acceleration circuitry, causing the remote acceleration resource to execute the portion of the function and causing the acceleration circuitry to execute a remaining portion of the function.

2. The edge computing device of claim 1 , wherein the second telemetry information includes an indication of the remote acceleration resource, a type of measurement for the remote acceleration resource, and a measurement value for the type of measurement.

3. The edge computing device of claim 2 , wherein the type of measurement relates to: resource utilization, resource availability, resource capabilities, performance measurements, power measurements, throughput, computational benchmarks, response time, monetary costs, resource usage costs, or latency.

4. The edge computing device of claim 2 , wherein the second telemetry information includes indications of a plurality of remote acceleration resources, wherein the type of measurement and the measurement value are provided for respective resources of the plurality of remote acceleration resources.

5. The edge computing device of claim 4 , wherein the plurality of remote acceleration resources includes a first acceleration resource at the remote location and a second acceleration resource at a peer of the remote location.

6. The edge computing device of claim 1 , wherein the estimated time to execute the function at the remote acceleration resource is based on a time to send the workload to the remote location, a time to execute the function on the workload at the remote location, and a time to obtain results of the function from the remote location, and wherein the use of the acceleration circuitry or the remote acceleration resource is further selected based on costs associated with a transport to the remote location and use of the acceleration resource at the remote location.

7. The edge computing device of claim 1 , the instructions further to perform operations to obtain features and hints of the remote acceleration resource for use of the remote acceleration resource at the remote location, wherein selecting use of the acceleration circuitry or the remote acceleration resource is further based on the features and hints obtained for the remote acceleration resource.

8. The edge computing device of claim 1 , the instructions further to perform operations to request, from the remote location, the second telemetry information, wherein the request communicates an identification of the function, a resource type, and requirements associated with the service level agreement.

9. The edge computing device of claim 8 , wherein the availability of the remote acceleration resource to execute the function is determined by the remote location, based on an estimated time to execute the function at the remote location which meets the requirements associated with the service level agreement.

10. The edge computing device of claim 1 , wherein a proxy function is used to produce an estimated time to execute the function at the remote acceleration resource, and wherein the acceleration circuitry or the remote acceleration resource is invoked based on a lower estimated time.

11. The edge computing device of claim 1 , wherein the acceleration circuitry includes at least one of a:

field-programmable gate array (FPGA), a graphics processing unit (GPU) hardware, central processing unit (CPU) hardware, or artificial intelligence (AI) processor hardware.

12. The edge computing device of claim 1 , wherein the remote location is a gateway, base station, or access point of the edge computing system, and wherein the edge computing device is a mobile computing device connected to the remote location via a wireless network.

13. At least one non-transitory machine-readable storage device comprising instructions stored thereupon, which when executed by a processing circuitry of a computing device, cause the processing circuitry to:

access first telemetry information that indicates availability of a local acceleration resource of the computing device to execute a function;

access second telemetry information that indicates availability of a remote acceleration resource to execute the function, the remote acceleration resource located at a remote location in an edge computing system;

identify an estimated time to execute the function at the remote acceleration resource or the local acceleration resource, based on evaluation of the first and second telemetry information;

identify an urgency level for the function; and

select use of the local acceleration resource or the remote acceleration resource, to execute the function on a workload, based on identification of the estimated time to execute the function at the remote acceleration resource or the local acceleration resource in relation to a service level agreement and the urgency level, wherein use of the acceleration circuitry or the remote acceleration resource includes:

in response to a determination that the remote acceleration resource can execute the entire function based on the service level agreement and the urgency level faster than the acceleration circuitry, causing the remote acceleration resource to execute the function;

in response to a determination that the acceleration circuitry can execute the entire function based on the service level agreement and the urgency level faster than the remote acceleration resource, causing the acceleration circuitry to execute the function; and

in response to a determination that the remote acceleration resource can execute a portion of the function based on the service level agreement and the urgency level faster than the acceleration circuitry, causing the remote acceleration resource to execute the portion of the function and causing the acceleration circuitry to execute a remaining portion of the function.

14. The machine-readable storage device of claim 13 , wherein the second telemetry information includes an indication of the remote acceleration resource, a type of measurement for the remote acceleration resource, and a measurement value for the type of measurement.

15. The machine-readable storage device of claim 14 , wherein the type of measurement relates to:

resource utilization, resource availability, resource capabilities, performance measurements, power measurements, throughput, computational benchmarks, response time, monetary costs, resource usage costs, or latency.

16. The machine-readable storage device of claim 14 , wherein the second telemetry information includes indications of a plurality of remote acceleration resources, wherein the type of measurement and the measurement value are provided for respective resources of the plurality of remote acceleration resources.

17. The machine-readable storage device of claim 16 , wherein the plurality of remote acceleration resources includes a first acceleration resource at the remote location and a second acceleration resource at a peer of the remote location.

18. The machine-readable storage device of claim 13 , wherein the estimated time to execute the function at the remote acceleration resource is based on a time to send the workload to the remote location, and a time to execute the function on the workload at the remote location, and a time to obtain results of the function from the remote location.

19. The machine-readable storage device of claim 13 , wherein the instructions further cause the processing circuitry to:

access features and hints of the remote acceleration resource for use of the remote acceleration resource at the remote location, wherein selecting use of the local acceleration resource or the remote acceleration resource is further based on the features and hints obtained from the remote location.

20. The machine-readable storage device of claim 13 , wherein the instructions further cause the processing circuitry to:

invoke the local acceleration resource or the remote acceleration resource, based on the invoked resource having a lower estimated time.

21. The machine-readable storage device of claim 13 , wherein the instructions further cause the processing circuitry to:

request, from the remote location, the second telemetry information, wherein the request communicates an identification of the function, a resource type, and requirements associated with the service level agreement.

22. The machine-readable storage device of claim 21 , wherein the availability of the remote acceleration resource to execute the function is determined by the remote location, based on an estimated time to execute the function at the remote location which meets the requirements associated with the service level agreement, and wherein the use of the local acceleration resource or the remote acceleration resource is further selected based on costs associated with a transport to the remote location and use of the acceleration resource at the remote location.

23. The machine-readable storage device of claim 13 , wherein a proxy function is used to produce an estimated time to execute the function at the remote acceleration resource.

24. A method performed by a computing system, comprising:

obtaining first telemetry information that indicates availability of a local acceleration resource of the computing system to execute a function;

obtaining second telemetry information that indicates availability of a remote acceleration resource to execute the function, the remote acceleration resource located at a remote location in an edge computing system;

identifying an estimated time to execute the function at the local acceleration resource or the remote acceleration resource, based on evaluation of the first and second telemetry information;

identifying an urgency level for execution of the function; and

selecting use of the local acceleration resource or the remote acceleration resource, to execute the function on a workload, based on identification of the estimated time to execute the function at the local acceleration resource or remote acceleration resource in relation to a service level agreement and the urgency level, wherein use of the local acceleration resource or the remote acceleration resource includes:

in response to determining that the remote acceleration resource can execute the entire function based on the service level agreement and the urgency level faster than the local acceleration resource, causing the remote acceleration resource to execute the function;

in response to determining that the local acceleration resource can execute the entire function based on the service level agreement and the urgency level faster than the remote acceleration resource, causing the local acceleration resource to execute the function; and

in response to a determination that the remote acceleration resource can execute a portion of the function based on the service level agreement and the urgency level faster than the local acceleration resource, causing the remote acceleration resource to execute the portion of the function and causing the local acceleration resource to execute a remaining portion of the function.

25. The method of claim 24 , wherein the second telemetry information includes an indication of the remote acceleration resource, a type of measurement for the remote acceleration resource, and a measurement value for the type of measurement, and wherein the type of measurement relates to: resource utilization, resource availability, resource capabilities, performance measurements, power measurements, throughput, computational benchmarks, response time, monetary costs, resource usage costs, or latency.