Characteristic-Based SIMD Task Assembly for Graphics Cache Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tile-based graphics processing systems face inefficiencies in handling the increased geometry data complexity of computer graphics applications, particularly in the rasterization phase, due to duplicated processing stages and memory transfer delays when generating sub-primitives.
Innovation Solution
A graphics processing system with a cache system and task assembly unit that stores and allocates computation instances for generating graphics data items, utilizing SIMD processing logic to execute these instances efficiently, thereby reducing redundant processing and improving efficiency in both geometry and rasterization phases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If tile-based rendering is used to reduce memory transfer, then memory transfer amount is reduced, but processing complexity increases due to duplicated processing stages
Solution Approach 1:
The patent applies preliminary action by performing geometry processing in advance during the geometry phase, storing results in tile memory. This allows the rasterization phase to directly access pre-processed geometry data without duplicating processing stages, thereby reducing both memory transfers and processing complexity.
Solution Approach 2:
The rendering process is segmented into distinct phases (geometry phase and rasterization phase) with clear separation of responsibilities. Geometry processing is completed and stored in tile memory before rasterization begins, allowing independent optimization of each phase and reducing overall system complexity.
2Reliability
If geometry data is processed in both geometry phase and rasterization phase, then rendering completeness is ensured, but redundant computations increase processing time
Solution Approach 1:
Geometry processing is performed as a preliminary action in the geometry phase, with results stored in tile memory. The rasterization phase then uses these pre-processed results directly, eliminating redundant computations while ensuring rendering completeness through the structured two-phase approach.
Solution Approach 2:
The patent ensures continuity of useful action by maintaining an unbroken data flow from geometry processing through rasterization. Pre-processed geometry data is continuously available in tile memory during rasterization, eliminating idle time and redundant processing while maintaining complete rendering functionality.
3Productivity
If computation instances are allocated without optimization, then task execution is simple, but SIMD processing efficiency decreases
Solution Approach 1:
The patent optimizes SIMD processing efficiency by dynamically changing allocation parameters - specifically, by allocating a predetermined number of computation instances to tasks based on task characteristics. This parameter optimization enables better utilization of SIMD units while maintaining manageable task allocation complexity through systematic rules.
Data Source
AI summary
A cache system in a graphics processing system stores graphics data items for use in rendering primitives. It is determined whether graphics data items relating to primitives to be rendered are present in the cache, and if not then computation instances for generating the graphics data items are created. Computation instances are allocated to tasks using a task assembly unit which stores task entries for respective tasks. The task entries indicate which computation instances have been allocated to the respective tasks. The task entries are associated with characteristics of computation instances which can be allocated to the respective tasks. A computation instance to be executed is allocated to a task based on the characteristics of the computation instance. SIMD processing logic executes computation instances of a task outputted from the task assembly unit to thereby determine graphics data items, which can be used to render the primitives.