Memory Bus Encoding for Burst Mode Energy Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing memory address encoding techniques are not compatible with burst mode accesses, leading to incorrect data retrieval when sequential logical addresses are encoded to reduce Hamming distance, as standard memory devices respond with non-encoded increments that do not follow the encoded sequence.
Innovation Solution
A memory controller that maps a portion of the logical memory address using Gray coding to create an encoded memory address, allowing standard memory devices to maintain burst mode capabilities while reducing energy consumption by minimizing Hamming distance between adjacent addresses, with burst control logic to handle situations where the burst length spans encoded memory address changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If address encoding is applied to reduce Hamming distance between sequential addresses, then energy consumption is reduced, but burst mode data retrieval becomes incorrect
Solution Approach 1:
The address is divided into two parts: an upper portion that is encoded using Gray code to reduce Hamming distance and energy consumption, and a lower portion that remains unencoded to maintain compatibility with burst mode sequential access. This segmentation allows each part to serve its specific function without interfering with the other.
Solution Approach 2:
Different encoding strategies are applied to different parts of the address. The upper portion uses Gray code encoding optimized for energy efficiency, while the lower portion uses standard binary encoding optimized for sequential access correctness. This local differentiation resolves the contradiction by allowing both encoding benefits to coexist in their respective domains.
2Reliability
If standard binary addressing is used for burst mode accesses, then sequential access correctness is maintained, but energy consumption increases due to higher Hamming distance
Solution Approach 1:
The address space is segmented into upper and lower portions, where the upper portion benefits from Gray code encoding to reduce transitions and energy consumption, while the lower portion maintains standard binary behavior for correct sequential bursting. This segmentation enables both correctness and energy efficiency.
Solution Approach 2:
The encoding strategy is localized to specific address bits. The upper portion uses Gray code where energy reduction is most beneficial, while the lower portion uses standard binary where sequential correctness is critical. This localized approach optimizes the overall system by applying the right encoding to the right location.
3Use of energy by moving object
If Gray code encoding is applied to the entire address, then energy consumption is minimized, but compatibility with standard burst mode memory operations is lost
Solution Approach 1:
The address is segmented such that only the upper portion is Gray coded while the lower portion remains in standard binary format. This segmentation preserves compatibility with standard burst mode operations that rely on sequential incrementing of the lower address bits, while still achieving energy savings in the upper address portion.
Solution Approach 2:
Gray code encoding is applied locally only where it provides benefit (upper address bits for bank/row selection) without affecting the lower address bits that are critical for burst mode sequential access. This localized encoding maintains adaptability while reducing energy consumption.
Data Source
AI summary
Encoding of logical addresses LA upon an off-chip memory bus 22 is performed to produce encoded addresses EA. The portion of the logical address encoded LA [9:3] does not include the least significant bits LA [2:0]. The number of bits LA [2:0] which are unencoded is chosen to correspond to a burst length BL supported by the memory 6 being accessed. Thus, burst mode accesses can be serviced by the memory 6 incrementing its memory address in the normal way. The encoding performed, such as Gray Encoding, reduces the Hamming distance between adjacent memory addresses in a sequence of memory addresses so as to reduce energy consumption.