Graphics Processor Task Allocation Across Specialized Execution Units
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional graphics processors inefficiently allocate resources due to all programmable execution units being configurable for either all or none of specialized tasks like machine learning or graphics processing, leading to increased silicon area and power consumption, especially in mobile devices.
Innovation Solution
A graphics processor with a subset of programmable execution units dedicated to specialized tasks and a processing resource that dynamically allocates tasks between these units, restricting capacity to prioritize specialized tasks when both task types are present.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If all programmable execution units are made configurable for specialized tasks like machine learning or graphics processing, then the processor can handle diverse task types, but silicon area and power consumption increase
Solution Approach 1:
The patent applies local quality by making only a subset of programmable execution units (specifically 50-75% of shader cores) configurable for specialized tasks like machine learning and ray tracing, while other units remain dedicated to traditional graphics processing. This selective configuration reduces overall power consumption while maintaining the capability to handle diverse task types through the configurable subset.
2Adaptability or versatility
If all programmable execution units are made configurable for specialized tasks, then the processor can perform multiple specialized operations, but the silicon area increases
Solution Approach 1:
The patent implements local quality by enabling specialized task configurability in only a portion (50-75%) of the programmable execution units rather than all units. This selective approach allows the processor to perform multiple specialized operations including machine learning and ray tracing while significantly reducing the silicon area required compared to making all units configurable.
3Device complexity
If programmable execution units are restricted to process only one task type, then resource allocation becomes simpler, but throughput decreases when multiple task types are present
Solution Approach 1:
The patent applies dynamics by implementing a dynamic task allocation mechanism that automatically distributes tasks between configurable and non-configurable execution units based on real-time workload composition. When specialized tasks are present, the system dynamically assigns them to configurable units while routing traditional graphics tasks to non-configurable units, thereby maintaining high throughput without requiring complex manual resource management.
4Use of energy by moving object
If a subset of execution units is dedicated to specialized tasks, then power consumption and silicon area are reduced, but task allocation complexity increases
Solution Approach 1:
The patent implements self-service by enabling the graphics processor to automatically manage its own task allocation between configurable and non-configurable execution units without requiring external intervention. The built-in task allocation mechanism autonomously identifies task types and assigns them to appropriate units, reducing the need for complex external resource management while maintaining efficient power utilization.
Data Source
AI summary
The present disclosure relates to a graphics processor having a plurality of programmable execution units operable to process tasks of a first task type, a subset of the plurality of programmable execution units further operable to process tasks of a second task type, wherein the second task type is different to the first task type, and restricting a capacity of the subset of programmable execution units to process one or more tasks of a first task type when tasks of both the first task type and the second task type are to be allocated to the programmable execution units.