Multi-Core GPU Draw Splitting with Topology-Aware Triangle Integrity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multicore GPUs face challenges in efficiently splitting geometry processing tasks without creating incomplete triangles, especially when triangles share vertices, leading to rendering errors.
Innovation Solution
A multicore GPU architecture with a master unit that distributes geometry processing tasks by grouping vertices or indices into primitives using topology information, splitting tasks into triggers, and assigning them to cores while maintaining the correct rendering order.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If geometry processing tasks are split into smaller groups for parallel processing, then processing speed is improved, but rendering accuracy deteriorates due to incomplete triangles
Solution Approach 1:
The draw call is segmented into multiple triggers, each containing a subset of primitives. The master unit divides the geometry processing tasks into smaller manageable groups that can be distributed to different cores, ensuring each core receives complete, valid primitives for processing
Solution Approach 2:
The master unit acts as an intermediary between the draw call and the individual cores. It receives the complete draw call, processes topology information to identify valid primitives, creates triggers with complete primitive sets, and distributes them to cores. This intermediary ensures rendering accuracy is maintained while enabling parallel processing
2Ease of operation
If a central hub is used to distribute geometry processing tasks, then task distribution control is improved, but device complexity increases
Solution Approach 1:
The master unit combines multiple functions into a single integrated component: receiving draw calls, processing topology information, identifying valid primitives, creating triggers, and distributing to cores. This consolidation provides centralized control while managing complexity by merging related functions
Solution Approach 2:
The master unit is designed as a multi-functional component that handles various tasks including draw call reception, topology processing, primitive validation, trigger creation, and task distribution. This universal design simplifies the overall architecture by consolidating control logic in one unit
Data Source
AI summary
A multicore graphics processing unit (GPU) and a method of operating a GPU are provided. The GPU comprises at least a first core and a second core. At least one of the cores in the multicore GPU comprises a master unit configured to distribute geometry processing tasks between at least the first core and the second core.