Method for regulating in real time the clock frequencies of at least one cluster of electronic machines, characterized in that it provides for carrying out the following steps: a) defining a finite number of discrete virtual capacity values f[l], f[2],.... 00 f[K], as global performance indices, of said cluster of machines; b) calculating by means of a randomized optimization procedure, for each value of said virtual capacity, a set of 1 vectors containing clock frequency values for each machine in said cluster; c) defin 00 ing a reference queue value, related to the number of processing requests received by said cluster; and provides for iteratively carry ing out the following steps: d) measuring the deviation between a current queue value, related to the number of processing requests in said current queue, and said reference queue value; e) selecting one of said discrete virtual capacity values on the basis of said » measured deviation, said selection occurring on the basis of a procedure that, at each iteration, analyzes said measured deviation, o compares it with the measured deviation value obtained at the previous iteration, and chooses whether to keep the current virtual ca - pacity value or to adopt one of the two adjacent or non-adjacent admissible virtual capacity values of said finite number of discrete values; f) selecting, based on said selected virtual capacity value, a vector of clock frequency values for each machine from said set o of 1 vectors, so as to optimize a multi-target performance index (J), and then setting the clock frequency of each machine in the cluster

Method and system for regulating in real time the clock frequencies of at least one cluster of electronic machines

Alessandro Pisano;
2020-01-01

Abstract

Method for regulating in real time the clock frequencies of at least one cluster of electronic machines, characterized in that it provides for carrying out the following steps: a) defining a finite number of discrete virtual capacity values f[l], f[2],.... 00 f[K], as global performance indices, of said cluster of machines; b) calculating by means of a randomized optimization procedure, for each value of said virtual capacity, a set of 1 vectors containing clock frequency values for each machine in said cluster; c) defin 00 ing a reference queue value, related to the number of processing requests received by said cluster; and provides for iteratively carry ing out the following steps: d) measuring the deviation between a current queue value, related to the number of processing requests in said current queue, and said reference queue value; e) selecting one of said discrete virtual capacity values on the basis of said » measured deviation, said selection occurring on the basis of a procedure that, at each iteration, analyzes said measured deviation, o compares it with the measured deviation value obtained at the previous iteration, and chooses whether to keep the current virtual ca - pacity value or to adopt one of the two adjacent or non-adjacent admissible virtual capacity values of said finite number of discrete values; f) selecting, based on said selected virtual capacity value, a vector of clock frequency values for each machine from said set o of 1 vectors, so as to optimize a multi-target performance index (J), and then setting the clock frequency of each machine in the cluster
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11584/299458
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact