Adaptive Memory Bandwidth Compression in CPU Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current memory bandwidth compression schemes in CPU-based systems face limitations in achieving high compression ratios while minimizing memory bandwidth consumption, leading to increased latency due to queued memory access requests when external memory bandwidth is reached.
Innovation Solution
An adaptive memory bandwidth compression system using a compressed memory controller (CMC) that employs two compression mechanisms: one for small data amounts and another for larger data groups, determining the best compression method based on the size of the results to store either separately or together in system memory, with a master table for tracking compression data and a prefetch accuracy indicator for optimizing data retrieval.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If memory bandwidth compression schemes are employed to reduce bandwidth consumption, then memory bandwidth efficiency is improved, but memory access latency increases due to queued requests
Solution Approach 1:
The patent implements dynamic compression ratio adjustment based on data characteristics and memory controller status. The compression ratio is not fixed but adapts in real-time to optimize between bandwidth savings and access latency, allowing the system to respond to changing workload conditions and memory queue states
Solution Approach 2:
The system changes multiple parameters including compression algorithm selection, compression ratio, and memory line size based on workload characteristics. By dynamically adjusting these parameters, the system optimizes the trade-off between compression efficiency and access performance for different data patterns and memory conditions
2Quantity of substance
If memory line size is increased to achieve higher compression ratios, then compression efficiency is improved, but wasted memory bandwidth increases
Solution Approach 1:
The patent segments memory lines into smaller units that can be independently compressed and transferred. This allows the system to apply compression to only the necessary data portions rather than entire memory lines, reducing wasted bandwidth while maintaining compression efficiency through selective processing
Solution Approach 2:
The system applies partial compression actions by compressing only the portions of memory data that benefit from compression rather than processing entire memory lines uniformly. This partial action approach avoids the bandwidth waste associated with compressing or transferring unnecessary data
3Device complexity
If conventional compression schemes are used, then implementation simplicity is maintained, but compression ratio and bandwidth efficiency are limited
Solution Approach 1:
The patent implements a universal compression architecture that can handle multiple data types, compression ratios, and memory access patterns through a single integrated system. The memory controller is designed to perform multiple functions including compression, decompression, and adaptive parameter adjustment, improving bandwidth efficiency without requiring separate specialized systems
Data Source
AI summary
Providing memory bandwidth compression using adaptive compression in central processing unit (CPU)-based systems is disclosed. In one aspect, a compressed memory controller (CMC) is configured to implement two compression mechanisms: a first compression mechanism for compressing small amounts of data (e.g., a single memory line), and a second compression mechanism for compressing large amounts of data (e.g., multiple associated memory lines). When performing a memory write operation using write data that includes multiple associated memory lines, the CMC compresses each of the memory lines separately using the first compression mechanism, and also compresses the memory lines together using the second compression mechanism. If the result of the second compression is smaller than the result of the first compression, the CMC stores the second compression result in the system memory. Otherwise, the first compression result is stored.