US 11,874,715 B2
Dynamic power budget allocation in multi-processor system
Nikos Kaburlasos, Folsom, CA (US); Iqbal Rajwani, Roseville, CA (US); Bhushan Borole, Rancho Cordova, CA (US); Kamal Sinha, Folsom, CA (US); and Sanjeev Jahagirdar, Folsom, CA (US)
Assigned to INTEL CORPORATION, Santa Clara, CA (US)
Filed by Intel Corporation, Santa Clara, CA (US)
Filed on Oct. 14, 2022, as Appl. No. 17/966,151.
Application 17/966,151 is a continuation of application No. 16/994,073, filed on Aug. 14, 2020, granted, now 11,493,974.
Application 16/994,073 is a continuation of application No. 16/004,647, filed on Jun. 11, 2018, granted, now 10,747,286, issued on Aug. 18, 2020.
Prior Publication US 2023/0030396 A1, Feb. 2, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 1/00 (2006.01); G06F 1/28 (2006.01); G06F 1/3203 (2019.01); G06F 1/3206 (2019.01)
CPC G06F 1/28 (2013.01) [G06F 1/3203 (2013.01); G06F 1/3206 (2013.01)] 20 Claims
OG exemplary drawing
 
1. An apparatus comprising:
a plurality of processors including a plurality of graphics processing units (GPUs); and
a power control hardware component to control power allocated to each of the plurality of GPUs;
the apparatus to perform dynamic power budget allocation for the plurality of GPUs, performing dynamic power budget allocation including:
assigning workloads to one or more of the plurality of GPUs,
establishing an initial power budget from a global power budget for operation of each of the plurality of GPUs, the plurality of GPUs sharing the global power budget,
tracking utilization of each of the plurality of GPUs in operation and identifying any of the plurality of GPUs that are under-utilized or over-utilized based at least in part on the tracking of utilization of each of the plurality of GPUs, and
dynamically reallocating an amount of the global power budget between the plurality of GPUs based at least in part on identification of one or more GPUs of the plurality of GPUs that are under-utilized or over-utilized.