US 12,493,493 B2
Method and system for allocating graphics processing unit partitions for a computer vision environment
Ian Roche, Glanmire (IE); Philip Hummel, San Jose, CA (US); and Dharmesh M. Patel, Round Rock, TX (US)
Assigned to Dell Products L.P., Round Rock, TX (US)
Filed by Dell Products L.P., Round Rock, TX (US)
Filed on Jan. 21, 2022, as Appl. No. 17/581,399.
Prior Publication US 2023/0236887 A1, Jul. 27, 2023
Int. Cl. G06F 9/50 (2006.01); G06N 20/00 (2019.01)
CPC G06F 9/5016 (2013.01) [G06F 9/5022 (2013.01); G06F 9/5033 (2013.01); G06F 9/5044 (2013.01); G06F 9/505 (2013.01); G06N 20/00 (2019.01)] 8 Claims
OG exemplary drawing
 
1. A method for allocating graphics processing unit partitions for a computer vision environment, the method comprising:
obtaining, by a computer vision (CV) manager, an initial graphics processing unit (GPU) partition allocation request associated with a CV workload;
in response to obtaining the initial GPU partition allocation request:
obtaining CV workload information associated with the CV workload including video data complexity, CV workload type, and CV workload vendor, wherein the CV workload information specifies a plurality of CV nodes that comprise the CV environment and wherein the CV workload information specifies CV nodes of the plurality of CV nodes which may perform the CV workload;
obtaining first CV environment configuration information associated with the GPU partition allocation request, wherein the CV environment configuration information specifies:
the plurality of CV nodes, and
CV node characteristics including central processing unit (CPU) types, CPU utilization, and memory capacity and utilization associated with the plurality of CV nodes;
generating an optimal GPU partition allocation by executing a machine learning model, wherein the machine learning model is one of a plurality of models in a GPU partition model, wherein:
the first CV environment configuration information and the CV workload information are inputs for the machine learning model, and
the optimal GPU partition allocation specifies:
a CV node of the plurality of CV nodes to perform the CV workload,
a GPU partition to allocate to a plurality of CV workloads executing on the CV node of the plurality of CV nodes to perform the CV workload,
the plurality of CV workloads comprises the CV workload, and
GPU partition information including GPU operations per second, GPU partition utilization, and GPU partition temperature; and
initiating performance of the CV workload in a CV environment based on the optimal GPU partition allocation.