Hybrid Data Compression for Guaranteed Memory Bandwidth Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data compression methods in GPUs and CPUs face challenges in balancing memory bandwidth and power consumption while maintaining quality, especially when random access is required, as lossless compression techniques may not achieve sufficient compression ratios and lossy techniques degrade data quality.
Innovation Solution
A hybrid approach using both lossless and lossy compression techniques to ensure a guaranteed compression threshold is met, combining primary and reserve compression units to output either lossless or lossy compressed data blocks as needed, ensuring efficient memory usage and reduced bandwidth.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If lossless compression techniques are used, then data quality is maintained, but compression ratio is insufficient and memory bandwidth requirements remain high
Solution Approach 1:
The system dynamically switches between lossless and lossy compression modes based on whether a compression threshold is met. The primary compression unit uses lossless techniques, and only when the threshold is not satisfied does the system activate the reserve compression unit using lossy techniques, making the compression approach adaptive rather than static
Solution Approach 2:
The system changes the compression parameter (lossless vs lossy) based on the compression threshold satisfaction. When the lossless compression does not achieve the target compression ratio, the system transitions to lossy compression mode, effectively changing the compression parameter to meet the threshold requirement
2Productivity
If lossy compression techniques are used, then compression ratio is improved, but data quality deteriorates
Solution Approach 1:
The system dynamically switches between lossless and lossy compression modes based on whether a compression threshold is met. The primary compression unit uses lossless techniques, and only when the threshold is not satisfied does the system activate the reserve compression unit using lossy techniques, making the compression approach adaptive rather than static
Solution Approach 2:
The lossy compression is used as a fallback option only when necessary, similar to using a cheaper alternative only when the primary option is insufficient. The system prefers lossless compression but accepts lossy compression as a disposable solution when the compression threshold cannot be met otherwise
3Speed
If memory bandwidth is increased, then data transfer speed is improved, but power consumption increases significantly
Solution Approach 1:
The invention extracts and removes redundant data through compression before transfer to memory, reducing the volume of data that needs to be transmitted. This decreases both the time required for data transfer and the power consumption associated with memory bandwidth operations
Solution Approach 2:
The compression is performed as a preliminary action before data is transferred to memory. By compressing the data first, the system reduces the subsequent memory transfer workload, thereby reducing both transfer time and power consumption during the actual memory operations
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.