| Summary: | 碩士 === 義守大學 === 資訊工程學系碩士班 === 99 === We propose two look-up table methods to improve the encoding redundant data speed on software RAID. One is the Lagrange-Like Method which calculating redundant data using base concept and the other is parallel Linear Feedback Shift Register (LFSR) which calculating parity using the generator polynomial of Reed-Solomon Code.
We use the polynomial base to calculate the redunant data and use the Horner’s rule with looup table to improvement the Lagrange-Like Method. The parallel LFSR can be calculated by module generator polynomail, and each result could be pre-calculated as constant date and store them in the memory.
The benefit of adopting lookup table does not need finite field operation. Therefore, computing P and Q method of the redundancy, namely, Lagrange-Like Method is faster than LFSR algorithm 40% and the memory size is required 128k bytes. The parallel LFSR algorithm is faster than LFSR method 80% and the memory size is required 126.5k bytes, on the improvement in speed has a considerable breakthrough.
|
|---|
