Dual-Mode Data Compression for Guaranteed Memory Bandwidth Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data compression methods for GPUs and other processing units face challenges in balancing memory bandwidth and quality, as lossless compression may not achieve sufficient compression ratios, leading to inefficiencies in memory usage and power consumption, especially when dealing with image data.
Innovation Solution
A method involving a primary lossless compression unit and a reserve lossy compression unit, where the primary unit generates a compressed data block, and if it exceeds a target threshold, the reserve unit uses lossy techniques to ensure the data block meets the threshold, ensuring guaranteed compression ratios and reduced memory bandwidth.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If lossless compression technique is used, then data quality is maintained, but compression ratio is insufficient and memory bandwidth is not reduced enough
Solution Approach 1:
The system dynamically switches between lossless and lossy compression modes based on whether the compressed data block size meets the target threshold. The compression approach is made adaptive rather than static, allowing the system to optimize between quality and compression ratio in real-time
Solution Approach 2:
The system changes the compression parameter (lossless vs lossy) to achieve the desired compression ratio. By adjusting the compression mode parameter, the system can transition from maintaining perfect quality to achieving higher compression when needed
2Speed
If memory bandwidth is increased to handle uncompressed or poorly compressed data, then data transfer speed improves, but power consumption increases significantly
Solution Approach 1:
The system performs compression before data transfer to reduce the amount of data that needs to be transferred over the memory bus. This preliminary compression action reduces subsequent bandwidth requirements and power consumption during data transfer operations
3Productivity
If lossy compression technique is used to achieve target compression threshold, then memory bandwidth is reduced, but data quality is compromised
Solution Approach 1:
The system applies lossy compression locally only when necessary (when lossless compression fails to meet the target threshold), rather than applying it universally. This allows the system to maintain high quality where possible and only sacrifice quality when compression requirements demand it
4Device complexity
If a single compression unit is used, then device complexity is reduced, but ability to guarantee compression threshold is insufficient
Solution Approach 1:
The compression system is segmented into two specialized units: a lossless compression unit for maintaining quality and a lossy compression unit for achieving compression thresholds. This segmentation allows each unit to be optimized for its specific function while together providing guaranteed threshold compliance
Data Source
AI summary
A method of compressing data is described in which the compressed data is generated by either or both of a primary compression unit or a reserve compression unit in order that a target compression threshold is satisfied. If a compressed data block generated by the primary compression unit satisfies the compression threshold, that block is output. However, if the compressed data block generated by the primary compression unit is too large, such that the compression threshold is not satisfied, a compressed data block generated by the reserve compression unit using a lossy compression technique, is output.