Technologies for coordinating disaggregated accelerator device resources

Abstract

A compute device to manage workflow to disaggregated computing resources is provided. The compute device comprises a compute engine receive a workload processing request, the workload processing request defined by at least one request parameter, determine at least one accelerator device capable of processing a workload in accordance with the at least one request parameter, transmit a workload to the at least one accelerator device, receive a work product produced by the at least one accelerator device from the workload, and provide the work product to an application.

1 . Cloud computing system for use in providing at least one cloud-based service associated with execution of at least one workload, the cloud computing system being for use in association with at least one network communication link, the cloud computing system comprising:
 compute resources comprising at least one central processing unit and memory circuitry;   accelerator resources comprising graphics processing unit (GPU) accelerator circuitry;   network fabric for use in communicatively coupling at least certain of the compute resources and/or accelerator resources; and
 management resources for use in allocating, based at least in part upon received request data, the compute resources and the accelerator resources for use in the execution of the at least one workload; 
 wherein:
 the at least one workload comprises at least one virtual machine workload and/or at least one container workload; 
 the accelerator resources are configurable to comprise local GPU accelerator circuitry and remote GPU accelerator circuitry; 
 the compute resources and the remote GPU accelerator circuitry are comprised in one or more cloud computing data centers; 
 the local GPU accelerator circuitry is comprised in at least one housing that is remote from the one or more cloud computing data centers; 
 the local GPU accelerator circuitry is to be communicatively coupled to the management resources via the at least one network communication link; 
 the management resources are configurable to obtain configuration-related data associated with the local GPU accelerator circuitry for use in management of the accelerator resources; and 
 the cloud computing system is configurable to dynamically reallocate, based upon past resource utilization data, resource utilization prediction data, and machine-learning, the compute resources and/or the accelerator resources for use in the execution of the at least one workload. 
 
   
     
     
         2 . The cloud computing system of  claim 1 , further comprising:
 storage resources for use in association with the compute resources and/or the accelerator resources; and   the compute resources, the accelerator resources, and storage resources are comprised, at least in part, in one or more pools of resources for being dynamically allocated based upon the resource utilization data.   
     
     
         3 . The cloud computing system of  claim 2 , wherein:
 the cloud computing system is configurable to implement hardware attestation associated, at least in part, with the compute resources and/or the accelerator resources.   
     
     
         4 . The cloud computing system of  claim 3 , wherein:
 the management resources are configurable to maintain directory data comprising:
 identification data to identify the remote GPU accelerator circuitry; and 
 configuration data comprising accelerator architecture-related data associated with the remote GPU accelerator circuitry. 
   
     
     
         5 . At least one machine-readable storage medium storing instructions for being executed by at least one machine associated with a cloud computing system, the cloud computing system being for use in providing at least one cloud-based service associated with execution of at least one workload, the cloud computing system being for use in association with at least one network communication link, the cloud computing system including compute resources, accelerator resources, network fabric, and management resources, the compute resources including at least one central processing unit and memory circuitry, the accelerator resources including graphics processing unit (GPU) accelerator circuitry, the network fabric being for use in communicatively coupling at least certain of the compute resources and/or accelerator resources, the instructions when executed by the at least one machine resulting in the cloud computing system being configured to perform operations comprising:
 allocating, by the management resources, based at least in part upon received request data, the compute resources and the accelerator resources for use in the execution of the at least one workload;   wherein:
 the at least one workload comprises at least one virtual machine workload and/or at least one container workload; 
 the accelerator resources are configurable to comprise local GPU accelerator circuitry and remote GPU accelerator circuitry; 
 the compute resources and the remote GPU accelerator circuitry are comprised in one or more cloud computing data centers; 
 the local GPU accelerator circuitry is comprised in at least one housing that is remote from the one or more cloud computing data centers; 
 the local GPU accelerator circuitry is to be communicatively coupled to the management resources via the at least one network communication link; 
 the management resources are configurable to obtain configuration-related data associated with the local GPU accelerator circuitry for use in management of the accelerator resources; and 
 the cloud computing system is configurable to dynamically reallocate, based upon past resource utilization data, resource utilization prediction data, and machine-learning, the compute resources and/or the accelerator resources for use in the execution of the at least one workload. 
   
     
     
         6 . The at least one machine-readable storage medium of  claim 5 , wherein the cloud computing system further comprises:
 storage resources for use in association with the compute resources and/or the accelerator resources; and   the compute resources, the accelerator resources, and storage resources are comprised, at least in part, in one or more pools of resources for being dynamically allocated based upon the resource utilization data.   
     
     
         7 . The at least one machine-readable storage medium of  claim 6 , wherein:
 the cloud computing system is configurable to implement hardware attestation associated, at least in part, with the compute resources and/or the accelerator resources.   
     
     
         8 . The at least one machine-readable storage medium of  claim 7 , wherein:
 the management resources are configurable to maintain directory data comprising:
 identification data to identify the remote GPU accelerator circuitry; and 
 configuration data comprising accelerator architecture-related data associated with the remote GPU accelerator circuitry. 
   
     
     
         9 . A method implemented using a cloud computing system, the cloud computing system being for use in providing at least one cloud-based service associated with execution of at least one workload, the cloud computing system being for use in association with at least one network communication link, the cloud computing system including compute resources, accelerator resources, network fabric, and management resources, the compute resources including at least one central processing unit and memory circuitry, the accelerator resources including graphics processing unit (GPU) accelerator circuitry, the network fabric being for use in communicatively coupling at least certain of the compute resources and/or accelerator resources, the method comprising:
 allocating, by the management resources, based at least in part upon received request data, the compute resources and the accelerator resources for use in the execution of the at least one workload;   wherein:
 the at least one workload comprises at least one virtual machine workload and/or at least one container workload; 
 the accelerator resources are configurable to comprise local GPU accelerator circuitry and remote GPU accelerator circuitry; 
 the compute resources and the remote GPU accelerator circuitry are comprised in one or more cloud computing data centers; 
 the local GPU accelerator circuitry is comprised in at least one housing that is remote from the one or more cloud computing data centers; 
 the local GPU accelerator circuitry is to be communicatively coupled to the management resources via the at least one network communication link; 
 the management resources are configurable to obtain configuration-related data associated with the local GPU accelerator circuitry for use in management of the accelerator resources; and 
 the cloud computing system is configurable to dynamically reallocate, based upon past resource utilization data, resource utilization prediction data, and machine-learning, the compute resources and/or the accelerator resources for use in the execution of the at least one workload. 
   
     
     
         10 . The method of  claim 9 , wherein the cloud computing system further comprises:
 storage resources for use in association with the compute resources and/or the accelerator resources; and   the compute resources, the accelerator resources, and storage resources are comprised, at least in part, in one or more pools of resources for being dynamically allocated based upon the resource utilization data.   
     
     
         11 . The method of  claim 10 , wherein:
 the cloud computing system is configurable to implement hardware attestation associated, at least in part, with the compute resources and/or the accelerator resources.   
     
     
         12 . The method of  claim 11 , wherein:
 the management resources are configurable to maintain directory data comprising:
 identification data to identify the remote GPU accelerator circuitry; and 
 configuration data comprising accelerator architecture-related data associated with the remote GPU accelerator circuitry.