Rasterization Triage via Dual Precision Edge Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3-D rendering architectures struggle to scale with increasingly complex scenes and large screen resolutions, leading to bottlenecks in rasterization processes, which cannot be effectively addressed by traditional methods without increasing power consumption or reducing fabrication yield.
Innovation Solution
Implementing a rasterization process that uses low precision evaluation for conservative triage of polygon status, where graphics primitives are rasterized into tiles and sub-tiles, with low precision edge evaluation used to identify completely uncovered or covered areas, and high precision evaluation for partially covered areas, allowing for efficient resource allocation and performance scaling without increased power consumption or reduced yield.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional per pixel evaluation techniques are used to determine pixel coverage, then measurement precision is improved, but device complexity and power consumption increase
Solution Approach 1:
The patent segments the pixel evaluation process into two distinct stages: coarse evaluation that quickly identifies definitely covered or definitely uncovered pixels, and fine evaluation that accurately determines partially covered pixels. This segmentation allows the system to avoid performing complex fine evaluation on all pixels, thereby reducing overall device complexity while maintaining measurement precision for pixels that require it.
Solution Approach 2:
The patent applies different evaluation qualities to different spatial locations: pixels that are definitely covered or definitely uncovered receive a coarse evaluation using simplified tests, while only pixels near polygon edges receive the more computationally intensive fine evaluation. This local quality approach ensures measurement precision is applied only where necessary, reducing overall device complexity and power consumption.
2Measurement precision
If traditional per pixel evaluation techniques are used to determine pixel coverage, then measurement precision is improved, but use of energy increases
Solution Approach 1:
The patent segments the pixel evaluation process into two distinct stages: coarse evaluation that quickly identifies definitely covered or definitely uncovered pixels, and fine evaluation that accurately determines partially covered pixels. This segmentation allows the system to avoid performing complex fine evaluation on all pixels, thereby reducing overall device complexity while maintaining measurement precision for pixels that require it.
Solution Approach 2:
The patent applies different evaluation qualities to different spatial locations: pixels that are definitely covered or definitely uncovered receive a coarse evaluation using simplified tests, while only pixels near polygon edges receive the more computationally intensive fine evaluation. This local quality approach ensures measurement precision is applied only where necessary, reducing overall device complexity and power consumption.
3Productivity
If more rasterization resources are allocated to handle large screen sizes, then productivity is improved, but use of energy increases and device complexity increases
Solution Approach 1:
The patent segments the pixel evaluation process into two distinct stages: coarse evaluation that quickly identifies definitely covered or definitely uncovered pixels, and fine evaluation that accurately determines partially covered pixels. This segmentation allows the system to avoid performing complex fine evaluation on all pixels, thereby reducing overall device complexity while maintaining measurement precision for pixels that require it.
Solution Approach 2:
The patent performs coarse evaluation on all pixels (excessive action) to quickly eliminate definitely covered or definitely uncovered pixels, then performs fine evaluation only on the subset of pixels that require it (partial action). This approach achieves high productivity for large screen sizes without requiring proportionally large increases in device complexity, as the expensive fine evaluation is applied only where necessary.
4Productivity
If more rasterization resources are allocated to handle large screen sizes, then productivity is improved, but power consumption increases
Solution Approach 1:
The patent segments the pixel evaluation process into two distinct stages: coarse evaluation that quickly identifies definitely covered or definitely uncovered pixels, and fine evaluation that accurately determines partially covered pixels. This segmentation allows the system to avoid performing complex fine evaluation on all pixels, thereby reducing overall device complexity while maintaining measurement precision for pixels that require it.
Solution Approach 2:
The patent performs coarse evaluation on all pixels (excessive action) to quickly eliminate definitely covered or definitely uncovered pixels, then performs fine evaluation only on the subset of pixels that require it (partial action). This approach achieves high productivity for large screen sizes without requiring proportionally large increases in device complexity, as the expensive fine evaluation is applied only where necessary.
Data Source
AI summary
In a raster stage of a graphics processor, a method for using low precision evaluation and high precision evaluation for conservative triage of polygon status. The method includes receiving a graphics primitive for rasterization in a raster stage of a graphics processor and rasterizing the graphics primitive to generate a plurality of tiles of pixels related to the graphics primitive. The tiles are rasterized at a first level precision to generate a plurality of sub-tiles related to the graphics primitive, wherein the sub-tiles are evaluated against the graphics primitive at each of their respective corners. Each of the sub-tiles not related to the graphics primitive are discarded. The sub-tiles related to the graphics primitive are rasterized at a second level precision.