GPU Source Location Mapping via Predefined Decoding Rules
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current graphics processing units (GPUs) face performance and power efficiency issues due to the significant memory consumption required for loading and storing source location data for parallel operations, which reduces overall system resource utilization.
Innovation Solution
Configuring a GPU with predefined rules to decode data source parameters, determining source locations and relationships between task invocations, and loading input data from specific memory addresses for efficient execution by logic units, thereby reducing the need for extensive local storage allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the shader unit allocates local storage to load and store all source location data for the workload, then the operation can be executed, but significant memory is consumed which reduces overall performance and power of the GPU
Solution Approach 1:
The patent extracts only the essential source location data needed for operation execution from the complete workload data, storing only this critical subset in local storage while keeping the rest in external memory. This selective extraction minimizes local storage consumption while maintaining execution capability.
Solution Approach 2:
The system performs preliminary decoding of source location data before execution to determine which specific data elements need to be loaded into local storage. By pre-identifying and loading only the necessary source location information, the system avoids consuming excessive memory while ensuring all required data is available for execution.
2Quantity of substance
If significant memory is allocated for source location data, then complete workload can be processed, but overall performance and power efficiency of the GPU is reduced
Solution Approach 1:
The patent changes the parameter of data representation by decoding source location data into compact indices or offsets that reference external memory locations. This parameter transformation allows the system to store minimal reference information locally while maintaining access to the complete workload data in external memory, thereby improving performance without sacrificing data completeness.
3Speed
If all source location data is loaded into local storage, then data access is fast, but power consumption increases
Solution Approach 1:
The system applies partial action by loading only the portion of source location data that is immediately needed for the current operation into local storage, rather than loading all possible data. This partial loading strategy maintains fast access speed for required data while avoiding the power consumption associated with loading and maintaining excessive data in high-speed memory.
Data Source
AI summary
The present disclosure relates to methods and apparatus for mapping a source location of input data for processing by a graphics processing unit. The apparatus can configure a processing element of the graphics processing unit with a predefined rule for decoding a data source parameter for executing a task by the graphics processing unit. Moreover, the apparatus can store the parameter in local storage of the processing element and configure the processing element to decode the parameter according to the at least one predefined rule to determine a source location of the input data and at least one relationship between invocations of the task. The apparatus can also load, to the local storage of the processing element, the input data from a plurality of memory addresses of the source location determined by the parameter. A one logic unit can then execute the task on the loaded input data.