# US 2024/0004443 A1 · Thermal and performance management

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

- Number: US 2024/0004443 A1
- Document ID: US20240004443A1
- Status: Published application
- Kind: A1
- Application number: 17/852,699
- Filed: 2022-06-29
- Published: 2024-01-04
- Earliest claimed: 2022-06-29
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Xin Zhang; Shun Zhang; Shaoze Fan; Xiaoxiao Guo; Chuang Gan
- Web page: https://patentgrep.com/patent/US20240004443A1

## CPC

- G06F 1/26: Power supply means, e.g. regulation thereof (for memories G11C)
- G06F 1/206: {comprising thermal management}
- G06F 1/28: Supervision thereof, e.g. detecting power-supply failure by out of limits supervision
- G06F 1/3203: Power management, i.e. event-based initiation of a power-saving mode
- G06N 3/006: based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
- G06N 3/045: Combinations of networks
- G06N 3/092: Reinforcement learning
- G06N 20/00: Machine learning
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Described aspects include a system for optimizing performance of a functional circuit unit, a method of optimizing performance of a functional circuit unit, and a computer program product. In one embodiment, the system may include a functional circuit unit having an associated cooling device and power converter, one or more sensors for the functional circuit unit, the one or more sensors including a power sensor and a temperature sensor, and a first machine learning model. The first machine learning model may be adapted to receive temperature data and power data from the one or more sensors, and to generate control signals for the cooling device and the power converter to optimize performance of the functional circuit unit.

## Claims

### Claim 1

A system for optimizing performance of a functional circuit unit, the system including:
a functional circuit unit having an associated cooling device and power converter;
one or more sensors for the functional circuit unit, the one or more sensors including a power sensor and a temperature sensor; and
a first machine learning model adapted to:
receive temperature data and power data from the one or more sensors; and
generate control signals for the cooling device and the power converter to optimize performance of the functional circuit unit.

### Claim 2 (depends on 1)

The system of claim 1, wherein the functional circuit unit comprises an electronic card of a server.

### Claim 3 (depends on 1)

The system of claim 1, wherein the functional circuit unit is chosen from the group consisting of a microprocessor, a graphical processing unit, and a machine learning accelerator.

### Claim 4 (depends on 1)

The system of claim 1, wherein the temperature data and power data further include performance signals associated with the functional circuit unit.

### Claim 5 (depends on 1)

The system of claim 1, wherein the first machine learning model is trained at least in part on simulated performance and temperature and power data using reinforcement learning techniques.

### Claim 6 (depends on 1)

The system of claim 1, further comprising a second machine learning model adapted to:
predict a state representation of the functional circuit unit based at least in part on the temperature data and power data; and
generate a vector that captures the predicted state representation.

### Claim 7 (depends on 6)

The system of claim 6, wherein the second machine learning model is adapted to recognize temporal dependencies in the temperature data and power data.

### Claim 8 (depends on 1)

The system of claim 1, further comprising a control parameter generator operably connected to the first machine learning model and adapted to generate control actions for the associated cooling device and power converter from the control signals.

### Claim 9 (depends on 1)

The system of claim 1, wherein the power sensor comprises a partial state sensor adapted to measures a power consumption of only functional circuit unit.

### Claim 10 (depends on 1)

The system of claim 1, further comprising a plurality of temperature sensors that measures temperatures at a plurality of various locations associated with the functional circuit unit.

### Claim 11

A method of optimizing performance of a functional circuit unit, comprising:
generating, by one or more sensors associated with a functional circuit unit having an associated cooling device and power converter, temperature data and power data; and
generating, by a first machine learning model, control signals for the cooling device and the power converter to optimize performance of the functional circuit unit.

### Claim 12 (depends on 11)

The method of claim 11, further comprising:
partially training the first machine learning model using simulated data in a simulated environment; and
transferring the first machine learning model to an online environment; and
additionally training the first machine in the online environment.

### Claim 13 (depends on 11)

The method of claim 11, further comprising:
predicting, by a second machine learning model, a state representation of the functional circuit unit based at least in part on the temperature data and power data; and
generating, by the second machine learning model, a vector that captures the predicted state representation.

### Claim 14 (depends on 13)

The method of claim 13, wherein the second machine learning model is adapted to recognize temporal dependencies in the temperature data and power data.

### Claim 15 (depends on 11)

The method of claim 11, further comprising generating, by a control parameter generator from the control signals, control actions for the associated cooling device and power converter.

### Claim 16

A computer program product for optimizing performance of a functional circuit unit, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to perform a method comprising:
generating, by one or more sensors associated with a functional circuit unit having an associated cooling device and power converter, temperature data and power data; and
generating, by a first machine learning model, control signals for the cooling device and the power converter to optimize performance of the functional circuit unit.

### Claim 17 (depends on 16)

The computer program product of claim 16, wherein the method further comprises:
partially training the first machine learning model using simulated data in a simulated environment; and
transferring the first machine learning model to an online environment; and
additionally training the first machine in the online environment.

### Claim 18 (depends on 16)

The computer program product of claim 16, wherein the method further comprises:
predicting, by a second machine learning model, a state representation of the functional circuit unit based at least in part on the temperature data and power data; and
generating, by the second machine learning model, a vector that captures the predicted state representation.

### Claim 19 (depends on 18)

The computer program product of claim 18, wherein the second machine learning model is adapted to recognize temporal dependencies in the temperature data and power data.

### Claim 20 (depends on 16)

The computer program product of claim 16, wherein the method further comprises generating, by a control parameter generator from the control signals, control actions for the associated cooling device and power converter.
