GPGPU systems and services
AssigneeCreativeC LLC
InventorsGreg Scantlen, Gary Scantlen
Life of this patent
Estimating the term…
Abstract
Graphics processing units (GPUs) deployed in general purpose GPU (GPGPU) units are combined into a GPGPU cluster. Access to the GPGPU cluster is then offered as a service to users who can use their own computers to communicate with the GPGPU cluster. The users develop applications to be run on the cluster and a profiling module tracks the applications' resource utilization and can report it to the user and to a subscription server. The user can examine the report to thereby optimize the application or the cluster's configuration. The subscription server can interpret the report to thereby invoice the user or otherwise govern the users' access to the cluster.
Claims16
A GPGPU cluster computing service comprising:
- a subscription server module communicating with a user's computer and controlling a user's access to the GPGPU cluster computing service;
- a plurality of GPGPU units each comprising an enclosure, a cooling fan, a CPU, system memory, GPU memory, a GPU, and a communications interface, wherein each GPU on the plurality of GPGPU units is accessible by the user computer subject to permission determined by the subscription server;
- a GPGPU application created by a user on the user's computer to include a specification for the GPGPU cluster configuration, the GPGPU application received by the plurality of GPGPU units from the user's computer, the GPGPU application configured to be run by the plurality of GPGPU units to produce application results based on the user created specification wherein the GPGPU units are interconnected in accordance with a first configuration, wherein the GPGPU application comprises GPU instructions that are performed by the GPUs, and wherein the GPGPU application comprises application data that is operated on by the GPU instructions;
- a second GPGPU application created by a user on the user's computer to include a second specification for the GPGPU cluster configuration, the second GPGPU application designed to also produce the application results using the second specification but wherein the GPGPU units are interconnected in accordance with a second configuration such that the user receives secondary resource utilization information to thereby guide the user to selecting a preferred configuration, wherein the user can test and profile the first GPGPU application and the second GPGPU application;
- and
- a profiling module that tracks the GPGPU application's resource utilization during run time such that the user receives resource utilization information.
A GPGPU cluster computing service comprising:
- a subscription server module communicating with a user's computer and controlling a user's access to the GPGPU cluster computing service;
- a plurality of GPGPU units each comprising and enclosure, a cooling fan, a CPU, system memory, GPU memory, a GPU, and a communications interface, wherein each GPU on the plurality of GPGPU units is accessible by the user computer subject to permission determined by the subscription server;
- a first GPGPU application created by a user on the user's computer to include a specification for the GPGPU cluster configuration, the first GPGPU application received by the plurality of GPGPU units from the user's computer, the first GPGPU application configured to be run by the GPGPU units to produce application results based on the user created specification wherein the GPGPU units are interconnected in accordance with a first configuration, wherein the first GPGPU application comprises GPU instructions that are performed by the GPUs, and wherein the first GPGPU application comprises application data that is operated on by the GPU instructions;
- a profiling module that tracks the first GPGPU application's resource utilization during run time such that the user receives primary resource utilization information;
- a second GPGPU application created by a user on the user's computer to include a second specification for the GPGPU cluster configuration, the second GPGPU application received by the plurality of GPGPU units from the user's computer, the second GPGPU application designed to also produce the application results using the second specification but wherein the GPGPU units are interconnected in accordance with a second configuration such that the user receives secondary resource utilization information to thereby guide the user to test and profile the first GPGPU application and the second GPGPU application to select a preferred configuration.
Dependencies are inferred from the claim text.