AI Command Queue Recovery for Fault-Tolerant Job Execution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hardware partitioning methods for fault tolerance in artificial intelligence calculation are costly and resource-intensive, necessitating a more economical solution.
Innovation Solution
A method involving resetting the command queue and recovering descriptors associated with target jobs based on tracking data to ensure fault tolerance without partitioning hardware resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If hardware resources are partitioned to ensure fault tolerance in artificial intelligence calculation, then reliability is improved, but device complexity and hardware resource requirements increase
Solution Approach 1:
The patent segments the command queue into multiple independent job queues, where each job queue handles specific jobs independently. This segmentation allows fault isolation at the software level without requiring physical hardware partitioning, thereby maintaining reliability while reducing device complexity
Solution Approach 2:
The patent introduces a job queue as an intermediary layer between the host system and the command queue. This intermediary manages job descriptors and coordinates command execution, enabling fault tolerance through software-based job isolation rather than direct hardware partitioning
2Reliability
If hardware resources are partitioned to ensure fault tolerance in artificial intelligence calculation, then reliability is improved, but hardware resource consumption increases
Solution Approach 1:
The patent makes the command queue universal by allowing it to handle multiple jobs through job queues and descriptors. A single command queue can serve multiple computational contexts without requiring separate hardware resources for each job, achieving fault tolerance through software multiplexing rather than hardware duplication
Solution Approach 2:
The patent merges multiple job management functions into a unified command queue structure with job descriptors. Instead of requiring separate hardware resources for each job, the system combines job scheduling, command management, and fault isolation into a single integrated software framework
Data Source
AI summary
Provided is a method for recovering data for artificial intelligence calculation, which is performed by one or more processors and which includes extracting, from a command queue, a descriptor associated with a target job, that is a job to be executed, of a plurality of jobs, executing at least one command associated with the extracted descriptor to execute artificial intelligence calculation associated with the target job, resetting the command queue if an error occurs while executing the artificial intelligence calculation, determining at least one descriptor to be recovered, based on tracking data associated with at least one job, and recovering the determined at least one descriptor to the reset command queue.