Workload Priority Cooling for Server BMCs

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing server cooling systems lack an efficient method to dynamically adjust cooling based on the priority levels of workloads, leading to suboptimal temperature management and increased power consumption.

Innovation Solution

A method that determines the priority level of a workload, sets a threshold cooling level for the computing unit based on that priority, and adjusts the usage level of the cooling system by receiving the current temperature and comparing it to the threshold, thereby optimizing fan speed and energy usage.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Temperature

If cooling system usage is increased to maintain lower temperatures for high-priority workloads, then temperature control for critical processes is improved, but power consumption increases

Engineering Contradiction:
Improvecomputing unit temperatureVSAvoidcooling system power consumption
Core Design Contradiction:
TemperatureVSUse of energy by moving object

Solution Approach 1:

The cooling system operates dynamically by continuously monitoring workload priority levels and adjusting fan speeds accordingly. The BMC receives priority level indications from computing units and modifies cooling usage in real-time, transitioning from static to dynamic operation to match actual thermal requirements

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes operational parameters by adjusting fan speed based on workload priority levels. When high-priority workloads are detected, the cooling system increases fan speed to maintain lower temperatures; when low-priority workloads are present, fan speed is reduced to conserve energy

Inventive Principle:
Principle #35Parameter changes

2Use of energy by moving object

If cooling resources are conserved for low-priority workloads, then power consumption is reduced, but temperature management for critical processes may be compromised

Engineering Contradiction:
Improvecooling system power consumptionVSAvoidtemperature management reliability
Core Design Contradiction:
Use of energy by moving objectVSReliability

Solution Approach 1:

The system implements feedback control where the BMC continuously monitors workload priority levels and temperature conditions, then adjusts cooling system operation accordingly. This closed-loop feedback ensures that temperature management reliability is maintained for high-priority processes while optimizing energy consumption

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The cooling system applies differentiated quality control by providing enhanced cooling specifically to computing units running high-priority workloads, while reducing cooling for low-priority processes. This localized approach ensures critical processes receive adequate temperature management without wasting energy on non-critical tasks

Inventive Principle:
Principle #3Local quality

3Temperature

If fan speed is increased to cool computing units, then temperature control is improved, but energy consumption increases

Engineering Contradiction:
Improvecomputing unit temperatureVSAvoidfan power consumption
Core Design Contradiction:
TemperatureVSUse of energy by stationary object

Solution Approach 1:

Fan speed is controlled dynamically based on real-time workload priority levels rather than operating at constant high speed. The BMC adjusts fan rotation speed in response to changing workload conditions, ensuring fans run at minimum necessary speed to reduce energy consumption while maintaining adequate cooling when needed

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS20250094222A1Cooling based on workload priority level
Publication Date: 2025.03.20 MICROSOFT TECHNOLOGY LICENSING LLC
  • US20250094222A1 patent drawing
  • US20250094222A1 patent drawing
  • US20250094222A1 patent drawing

AI summary

The described technology provides a method including determining a priority level associated with a workload, determining, based on the priority level associated with the workload, a threshold cooling level of a computing unit implementing the workload, receiving, at a base motherboard controller (BMC) associated with the computing unit implementing the workload, a current temperature of the computing unit implementing the workload, and adjusting, based on the threshold cooling level of the computing unit implementing the workload and the current temperature of the computing unit implementing the workload, a usage level of a cooling system of the computing unit implementing the workload.