Entropy-Agnostic Data Encoding for Low-Energy Memory Transfer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data encoding techniques are inefficient in reducing energy consumption during data movement in computer systems, particularly when data patterns are random or biased, as they are optimized for either random or biased patterns but not both, leading to high energy consumption in systems like DDR4, LPDDR4, and HBM.
Innovation Solution
The implementation of entropy agnostic data encoding, which uses a combination of deterministic biased and random binary vectors to encode data, selecting codewords that minimize bit transitions, logic high or low bit transmission, and crosstalk, regardless of data entropy, through circular shifts and XOR operations, and includes an encoder and decoder system to optimize energy consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If existing encoding techniques optimized for asymmetric polarity distribution are used, then energy consumption is reduced for biased data patterns, but energy consumption increases for high entropy random data patterns
Solution Approach 1:
The encoder dynamically adapts its encoding strategy based on the entropy characteristics of the input data. For low-entropy biased data, it uses deterministic biased vectors optimized for asymmetric polarity distribution. For high-entropy random data, it switches to random binary vectors that do not assume any polarity bias, thereby maintaining energy efficiency across varying data patterns.
Solution Approach 2:
The system changes the parameters of the binary vectors used for encoding based on data characteristics. When data exhibits asymmetric polarity distribution, the encoder uses vectors with corresponding bias. When data is random with high entropy, the encoder uses uniformly distributed random vectors, effectively adapting the encoding parameters to match the statistical properties of the input data.
2Loss of energy
If encoding techniques assume asymmetric polarity distribution are used, then energy consumption is reduced for biased patterns, but performance degrades when data is random
Solution Approach 1:
The encoder dynamically evaluates the entropy of incoming data and adjusts its encoding approach in real-time. This dynamic adaptation ensures that the encoding remains effective and energy-efficient regardless of whether the data follows an asymmetric distribution or is completely random, thereby maintaining reliability across different data scenarios.
Solution Approach 2:
The system modifies the statistical parameters of the encoding vectors based on data analysis. For asymmetric data, it uses vectors with matched polarity bias to minimize transitions. For random data, it uses vectors with uniform distribution parameters, ensuring that the encoding effectiveness is maintained by matching vector parameters to data characteristics.
3Use of energy by moving object
If deterministic biased binary vectors are used for encoding, then energy consumption is reduced for asymmetric data patterns, but the encoding becomes ineffective for random data
Solution Approach 1:
The encoding system transitions from static deterministic biased vectors to dynamic vector selection based on data entropy. The encoder continuously adapts between deterministic biased vectors for low-entropy data and random binary vectors for high-entropy data, ensuring energy efficiency and broad applicability across different entropy levels.
Solution Approach 2:
The system changes the distribution parameters of the binary vectors used for encoding. For asymmetric data patterns, it uses vectors with biased parameters that exploit the data's statistical properties. For random data with high entropy, it uses vectors with uniform distribution parameters, thereby maintaining encoding effectiveness across varying entropy conditions.
4Use of energy by moving object
If random binary vectors are used for encoding, then energy consumption is reduced for high entropy data, but the encoding is less efficient for biased patterns
Solution Approach 1:
The encoder dynamically selects between deterministic biased vectors and random binary vectors based on the entropy characteristics of the input data. This dynamic selection ensures that random vectors are used for high-entropy data to minimize energy consumption, while deterministic biased vectors are used for low-entropy data to maximize encoding efficiency.
Solution Approach 2:
The system adjusts the randomness parameter of the encoding vectors based on data entropy. For high-entropy random data, it uses vectors with high randomness parameters to achieve energy efficiency. For low-entropy biased data, it uses vectors with lower randomness parameters that are optimized for encoding efficiency, thereby balancing energy consumption and productivity across different data types.
Data Source
AI summary
Entropy agnostic data encoding includes: receiving, by an encoder, input data including a bit string; generating a plurality of candidate codewords, including encoding the input data bit string with a plurality of binary vectors, wherein the plurality of binary vectors includes a set of deterministic biased binary vectors and a set of random binary vectors; selecting, in dependence upon a predefined criteria, one of the plurality of candidate codewords; and transmitting the selected candidate codeword to a decoder.