The current canonicalization code processes the code one byte at a time. The processing speed could be improved by using SIMD instructions. The methods described in this paper could be adapted for canonicalization of SQL statements.