PCIe Topology GPU Temperature Monitoring for Server Fan Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern computing systems face overheating issues due to the lack of I²C connections in some GPUs, which prevents effective temperature monitoring and cooling, potentially leading to physical damage and system failures.
Innovation Solution
The system retrieves temperature information of GPUs via PCIe topology and uses a management controller to control fan speeds, leveraging PCIe BusID information to access temperature data from a GPU library and manage cooling fans, even in the absence of I²C connections, ensuring optimal GPU operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If I²C connections are used for temperature monitoring, then temperature monitoring is reliable, but device complexity increases and not all GPUs have I²C connections
Solution Approach 1:
The patent uses PCIe BusID as an intermediary to indirectly identify and retrieve GPU temperature information. Instead of directly accessing GPU sensors via I²C connections, the management controller obtains BusID through PCIe topology and uses it as a key to retrieve temperature data from the GPU library, eliminating the need for direct I²C connections while maintaining reliable temperature monitoring.
Solution Approach 2:
The patent replaces the physical I²C connection mechanism with a software-based PCIe topology and library retrieval system. Temperature monitoring transitions from hardware-level direct sensor access via I²C to software-level data retrieval through PCIe BusID identification and library queries, simplifying the physical connection requirements.
2Adaptability or versatility
If PCIe topology is used to retrieve temperature information, then compatibility with GPUs without I²C connections is improved, but measurement precision may be affected
Solution Approach 1:
The patent creates a virtual representation of GPU temperature data through the GPU library, which stores temperature information accessible via PCIe BusID. This copying mechanism allows the management controller to access accurate temperature readings without requiring physical I²C connections, maintaining measurement precision while improving compatibility.
3Temperature
If fan speed is controlled based on GPU temperature, then cooling effectiveness is improved, but device complexity increases due to additional control mechanisms
Solution Approach 1:
The patent implements a feedback control loop where the management controller continuously monitors GPU temperature through PCIe BusID and dynamically adjusts fan speed accordingly. Temperature readings are retrieved via the GPU library, and fan control decisions are made based on these readings, creating a closed-loop system that maintains optimal GPU temperature while using existing management controller capabilities.
Data Source
Figure 1A
Figure 1B
Figure 1C
AI summary
The present disclosure provides a system and method for retrieving temperature information of GPUs of a server system via PCIe topology, and using a baseboard management controller (BMC) to control fan speed(s) of cooling fans based at least upon the temperatures of the GPUs. In some implementations, the management controller can determine a PCIe topology of the server system via an operating system (OS), and get the BusID of each GPU or GPU card of the server system. Based upon the BusID, the management controller can retrieve temperature information of a corresponding GPU from a GPU library (e.g., NVIDIA Management Library [NVML]), and further control fan speed of cooling fan(s) based at least upon the temperature information.