IC Chip Telemetry Reporting for Bandwidth-Aware Fleet Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Managing and controlling large numbers of computing devices with multiple integrated circuit (IC) chips for distributed tasks like cryptocurrency mining or AI computations poses challenges in data collection, power consumption management, and adherence to electrical grid regulations, requiring improved control processes and data handling.
Innovation Solution
A system and method for controlling and managing multiple IC chips in computing devices, involving a controller that adjusts and compresses status data, sends it to a remote server, and receives computational parameters to optimize power usage and compliance with grid regulations, using an asset management server for centralized control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If all IC chips send detailed status data to the server, then data completeness is improved, but network bandwidth consumption increases
Solution Approach 1:
The patent extracts only the necessary status data from each IC chip based on operational needs and device grouping, rather than transmitting all available data. The controller selectively determines which status parameters to collect from each IC chip, reducing unnecessary data transmission while maintaining data completeness for monitoring and control purposes.
Solution Approach 2:
The patent segments the fleet of IC chips into different device groups with varying data collection requirements. Different reporting configurations are applied to different groups based on their operational characteristics and monitoring needs, allowing optimized data transmission for each segment rather than a uniform approach to all chips.
2Measurement precision
If the server queries each IC chip for status data, then data accuracy is improved, but communication overhead increases
Solution Approach 1:
The patent inverts the traditional client-server query model by having controllers push status data to the server proactively. Instead of the server querying each IC chip individually, the controllers collect status data from their respective IC chips and push the aggregated data to the server, reducing communication overhead while maintaining data accuracy.
Solution Approach 2:
The controller acts as an intermediary between the IC chips and the server. It collects status data from multiple IC chips, processes and aggregates the information, then pushes the consolidated data to the server. This intermediary role reduces the number of direct communications required between the server and individual IC chips.
3Measurement precision
If detailed status data is collected from all devices, then monitoring accuracy is improved, but data processing time increases
Solution Approach 1:
The patent applies different data collection and processing qualities to different device groups based on their specific monitoring requirements. Critical devices receive more detailed monitoring with higher data granularity, while less critical devices use simplified monitoring approaches, optimizing the balance between monitoring accuracy and processing time for each local context.
Solution Approach 2:
The patent implements partial data collection by focusing on essential status parameters for each device group rather than collecting all possible data points. This selective approach to data collection reduces processing time while maintaining sufficient monitoring accuracy for operational decision-making.
Data Source
AI summary
A computing device includes: a plurality of integrated circuit (IC) chips arranged on one or more substrates; and a controller. The controller is configured to: control the plurality of integrated circuit (IC) chips to perform a common type of computation in parallel with one another; receive, from one or more of the plurality of IC chips, IC chip data indicative of performance of the IC chips; generate status data based on the IC chip data; and push the status data to a remote asset management server.