10-Bit Pixel Compression With Error Values for Lower GPU Bandwidth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data compression methods for GPUs and CPUs face challenges in balancing memory bandwidth and power consumption while maintaining high-quality rendering, especially in mobile devices, where memory resources are limited.
Innovation Solution
A method is described for converting 10-bit pixel data to 8-bit pixel data with error values, using truncation and error generation, and a hardware unit is implemented to perform this conversion, ensuring a guaranteed compression threshold is met by combining lossless and lossy compression techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If data compression is applied to reduce memory bandwidth and storage requirements, then memory bandwidth and storage requirements are reduced, but data quality and rendering accuracy deteriorate
Solution Approach 1:
The patent changes the bit depth parameter from 10-bit to 8-bit representation, which reduces the data size and memory bandwidth requirements. This parameter change achieves compression while maintaining acceptable visual quality through controlled truncation and error management techniques
Solution Approach 2:
The patent introduces error values as an intermediary element that captures the information lost during compression. By storing error values separately and using them during decompression, the system can reconstruct the original data with high accuracy, thus resolving the contradiction between compression and quality
2Manufacturing precision
If higher quality rendering algorithms are used to improve rendering quality, then rendering quality is improved, but memory bandwidth requirements increase
Solution Approach 1:
The patent applies compression to image data before it is stored in memory or transferred between memory and processing units. This preliminary compression action reduces the amount of data that needs to be moved, thereby reducing memory bandwidth requirements while still allowing high-quality rendering algorithms to process the data efficiently
3Manufacturing precision
If memory bandwidth is increased to support higher quality rendering, then rendering quality is improved, but power consumption increases
Solution Approach 1:
By changing the data representation parameter from 10-bit to 8-bit, the patent reduces the volume of data that must be transferred through the memory subsystem. This parameter change directly reduces power consumption associated with data movement while maintaining rendering quality through the error value compensation mechanism
Data Source
AI summary
A method of converting 10-bit pixel data (e.g. 10:10:10:2 data) into 8-bit pixel data involves converting the 10-bit values to 7-bits or 8-bits and generating error values for each of the converted values. Two of the 8-bit output channels comprise a combination of a converted 7-bit value and one of the bits from the fourth input channel. A third 8-bit output channel comprises the converted 8-bit value and the fourth 8-bit output channel comprises the error values. In various examples, the bits of the error values may be interleaved when they are packed into the fourth output channel.