Near Memory Compaction for Reduced Footprint
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current memory hierarchies in graphics and central processors have limited near memory (NM) capacity due to inefficient use of space, as compression techniques do not reduce the memory footprint in near memory, leading to increased reliance on slower far memory accesses, which are costly in terms of performance and power consumption.
Innovation Solution
The implementation of a compaction technique that reduces the footprint of data in near memory by compacting compressed data into fewer, contiguous blocks, utilizing a compaction buffer and daemon to manage and optimize memory usage, allowing for more efficient storage and access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If data is compressed and stored sparsely in near memory, then random access speed is improved, but memory footprint is not reduced leading to increased far memory accesses
Solution Approach 1:
The patent divides compressed data into fixed-size blocks and reorganizes them from sparse noncontiguous storage to compact contiguous storage. Each block is processed and relocated to eliminate gaps, transforming the storage layout from segmented-sparse to segmented-contiguous, thereby reducing overall memory footprint while preserving block-level access efficiency.
Solution Approach 2:
The patent implements a dynamic compaction process that continuously reorganizes data blocks in near memory. A compaction mechanism actively monitors and repositions data blocks to eliminate unused space, making the storage layout adaptive and dynamic rather than static, thus optimizing memory utilization without sacrificing access speed.
2Reliability
If near memory capacity is increased to store more data, then hit rate is improved, but cost and power consumption increase
Solution Approach 1:
The patent changes the storage density parameter in near memory through compaction. By transforming sparse storage to compact storage, the effective capacity of near memory increases without physical expansion, allowing more data to be stored in the same footprint, thereby improving hit rate without increasing power consumption or cost.
3Quantity of substance
If lossy compression is applied to reduce footprint, then memory usage is reduced, but data quality deteriorates
Solution Approach 1:
The patent extracts and removes the sparseness characteristic from compressed data storage. By taking out the gaps and unused spaces from the sparse storage layout and eliminating them through compaction, the system achieves footprint reduction without altering the actual compressed data content, thus preserving data quality while reducing memory usage.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
In accordance with some embodiments, compaction, as contrasted with compression, is used to reduce the footprint of a near memory. In compaction, the density of data storage within a storage device is increased. In compression, the number of bits used to represent information is reduced. Thus you can have compression while still having sparse or non-contiguously arranged storage. As a result, compression may not always reduce the memory footprint. By compacting compressed data, the footprint of the information stored within the memory may be reduced. Compaction may reduce the need for far memory accesses in some cases.