There are some room to optimize the EUC_JP code converters (Byte to Char).

  - Handle ASCII characters in convert() to avoid the convSingleByte() call overhead.

  - Look up byte1 in the order of the general priority
    (i.e., JIS X 0208, JIS X 0201:GR, JIS X 0212)

  - Eliminate unnecessary masking with 0xff

The revised ByteToCharEUC_JP.java is in Attachment. It gives 12% better performance (CPU time) with ASCII only data and 18% better with ASCII+Kanji data with the String(byte[]) constructor.

Probably similar optimizations would be applicable to other multi-byte code converters.