Memory Sub-System CPU Cycle Tracking for Asynchronous Processes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to accurately monitor the execution time of asynchronous processes in memory sub-systems due to their parallel and out-of-order execution, leading to misleading performance measurements.
Innovation Solution
Implementing a process instance tagging solution in the memory sub-system controller to assign unique tags to each instance of a process and record start and stop values, allowing for precise calculation of execution time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional monitoring methods are used for asynchronous processes, then the monitoring system remains simple, but the measurement precision of execution time is poor
Solution Approach 1:
The patent segments the monitoring of asynchronous processes by introducing unique tags for each process instance. Each tag acts as an independent identifier that allows the system to track individual process executions separately, enabling precise measurement of execution time for each instance without conflating multiple parallel processes.
Solution Approach 2:
The patent introduces tags as intermediary elements between the asynchronous processes and the monitoring system. These tags serve as mediators that carry process instance information through the system, allowing the monitor to accurately associate completion events with their corresponding start events even when processes execute in parallel and out of order.
2Reliability
If asynchronous processes are monitored without instance tagging, then the monitoring overhead is low, but the reliability of performance data is poor
Solution Approach 1:
The patent applies preliminary action by assigning unique tags to each asynchronous process instance at the moment of initiation, before the process executes. This pre-tagging ensures that when the process completes, the monitoring system can immediately identify which instance has finished by reading the tag, eliminating ambiguity even when multiple instances are executing concurrently.
Solution Approach 2:
The patent implements feedback by using tags to create a closed-loop tracking system. The tag is assigned at process start, carried through execution, and read at completion, providing continuous feedback to the monitoring system about the state of each process instance. This feedback mechanism ensures accurate attribution of completion events to their corresponding start events.
Data Source
AI summary
A processing device in a memory sub-system initiates execution of an instance of a process on a memory device, associates a unique tag with the instance of the process, and stores an indication of a start time value for the instance of the process in an entry of a tag repository corresponding to the unique tag. Upon detecting a completion of the instance of the process, the processing device stores an indication of a stop time value for the instance of the process in the entry of the tag repository corresponding to the unique tag.