# US 2021/0081027 A1 · Method of performing dynamic voltage and frequency scaling based on power step

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0081027 A1
- Document ID: US20210081027A1
- Status: Published application
- Kind: A1
- Application number: 16/861,383
- Filed: 2020-04-29
- Published: 2021-03-18
- Earliest claimed: 2019-09-16
- Assignees: Samsung Electronics Co., Ltd.
- Applicants: Samsung Electronics Co., Ltd.
- Inventors: Jonglae Park; Youngtae Lee; Choonghoon Park; Hyunchul Seok; Kwanjin Jung
- Web page: https://patentgrep.com/patent/US20210081027A1

## CPC

- G06F 1/3296: by lowering the supply or operating voltage
- G06F 1/3228: Monitoring task completion, e.g. by use of idle timers, stop commands or wait commands
- G06F 1/324: by lowering clock frequency
- G06F 1/329: by task scheduling
- Y02D 10/00: Energy efficient computing, e.g. low power processors, power management or thermal management
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- Y02D: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE

## Abstract

Dynamic voltage and frequency scaling (DVFS) is performed based on a power step by setting a plurality of power levels corresponding to a plurality of available frequencies of a clock signal for an operation of a processor, setting a plurality of power steps corresponding to the plurality of available frequencies, and controlling a conversion between the plurality of power levels based on a utilization of the processor and the plurality of power steps. Performance and power consumption of a processor are controlled efficiently by performing power level conversion based on the power step.

## Claims

### Claim 1

A method of performing dynamic voltage and frequency scaling (DVFS), the method comprising:
setting a plurality of power levels corresponding to a plurality of available frequencies of a clock signal for an operation of a processor;
setting a plurality of power steps corresponding to the plurality of available frequencies; and
controlling a conversion between the plurality of power levels based on a utilization of the processor and the plurality of power steps.

### Claim 2 (depends on 1)

The method of claim 1, wherein controlling the conversion comprises:
determining a power target frequency based on a current utilization of the processor and a current power step corresponding to a current power level determined from among the plurality of power levels; and
determining a next power level from among the plurality of power levels based on the power target frequency.

### Claim 3 (depends on 2)

The method of claim 2, wherein determining the power target frequency comprises:
determining a maximum power at the current power level;
determining a margin power corresponding to a sum of the maximum power and the current power step;
determining a target power corresponding to a multiplication value of the margin power and the current utilization; and
providing the power target frequency corresponding to the target power.

### Claim 4 (depends on 2)

The method of claim 2, wherein determining the next power level comprises:
determining a plurality of candidate frequencies from among the plurality of available frequencies, wherein each of the candidate frequencies is greater than the power target frequency; and
determining the next power level corresponding to a minimum frequency from among the candidate frequencies.

### Claim 5 (depends on 1)

The method of claim 1, wherein setting the plurality of power steps comprises:
setting a common power step corresponding to the plurality of available frequencies.

### Claim 6 (depends on 5)

The method of claim 5, wherein setting the common power step comprises:
setting a reference power; and
setting the common power step to be equal to the reference power divided by M, where M is a natural number.

### Claim 7 (depends on 6)

The method of claim 6, further comprising:
controlling a speed of the conversion by adjusting a value of M.

### Claim 8 (depends on 7)

The method of claim 7, wherein the speed of the conversion is decreased as the value of M is increased.

### Claim 9 (depends on 6)

The method of claim 6, wherein setting the reference power comprises:
changing the reference power based on at least one of an operation temperature of the processor and an available power of the processor.

### Claim 10 (depends on 1)

The method of claim 1, wherein setting the plurality of power steps comprises:
with respect to an i-th available frequency and an (i+1)-th available frequency from among the plurality of available frequencies, setting an i-th power step corresponding to the i-th available frequency to be smaller than an (i+1)-th power step corresponding to the (i+1)-th available frequency, where i is a natural number.

### Claim 11 (depends on 1)

The method of claim 1, wherein setting the plurality of power steps comprises:
decreasing at least one of the plurality of power steps as an available power of the processor decreases.

### Claim 12 (depends on 1)

The method of claim 1, wherein setting the plurality of power steps comprises:
decreasing at least one of the plurality of power steps as an operation temperature of the processor increases.

### Claim 13 (depends on 1)

The method of claim 1, wherein the processor is a multi-core processor comprising a first processor core operating based on a first clock signal and a second processor core operating based on a second clock signal.

### Claim 14 (depends on 13)

The method of claim 13, wherein setting the plurality of power steps comprises:
setting a plurality of first core power steps with respect to the first processor core; and
setting a plurality of second core power steps different from the first core power steps with respect to the second processor core.

### Claim 15 (depends on 1)

The method of claim 1, wherein the processor is a multi-core processor comprising a first cluster of first processor cores operating based on a first clock signal and a second cluster of second processor cores operating based on a second clock signal.

### Claim 16 (depends on 15)

The method of claim 15, wherein setting the plurality of power steps comprises:
setting a plurality of first core power steps with respect to the first cluster; and
setting a plurality of second core power steps different from the first core power steps with respect to the second cluster.

### Claim 17 (depends on 1)

The method of claim 1, wherein the processor is a multi-core processor comprising a plurality of processor cores, and
wherein controlling the conversion comprises:
determining a performance target frequency based on a current utilization of the processor and a current available frequency corresponding to a current power level determined from among the plurality of power levels,
wherein the current utilization corresponds to a task scheduling operation that assigns a plurality of tasks to the plurality of processor cores;
determining a power target frequency based on a current power step corresponding to the current power level and the current utilization; and
determining a next power level from among the plurality of power levels based on a selected frequency corresponding to a greater one of the performance target frequency and the power target frequency.

### Claim 18 (depends on 17)

The method of claim 17, wherein determining the performance target frequency comprises:
providing the performance target frequency corresponding to a multiplication value of the current available frequency, the current utilization and a margin rate, and
wherein determining the power target frequency comprises:
determining a maximum power at the current power level;
determining a margin power corresponding to a sum of the maximum power and the current power step;
determining a target power corresponding to a multiplication value of the margin power and the current utilization; and
providing the power target frequency corresponding to the target power.

### Claim 19

A computing system, comprising:
a processor comprising at least one processor core;
a power management circuit configured to provide a clock signal and a voltage for an operation of the processor; and
a memory storing a dynamic voltage and frequency scaling (DVFS) computer program,
wherein the processor is configured to execute the DVFS computer program to:
set a plurality of power levels corresponding to a plurality of available frequencies of the clock signal for the operation of the processor;
set a plurality of power steps corresponding to the plurality of available frequencies; and
control a conversion between the plurality of power levels based on a utilization of the processor and the plurality of power steps.

### Claim 20

A method of performing dynamic voltage and frequency scaling (DVFS), the method comprising:
setting a plurality of power levels corresponding to a plurality of available frequencies of a clock signal for an operation of a processor;
setting a plurality of power steps corresponding to the plurality of available frequencies based on at least one of an operation temperature of the processor and an available power of the processor;
determining a power target frequency based on a current utilization of the processor and a current power step corresponding to a current power level determined from among the plurality of power levels; and
determining a next power level from among the plurality of power levels based on the power target frequency.
