1. 性能

对于使用Unicode字符集的情况，一般使用_general_ci会比xxx_unicode_ci快。

For any Unicode character set, operations performed using the xxx_general_ci collation are faster than those for the xxx_unicode_ci collation.

via:https://dev.mysql.com/doc/refman/5.0/en/charset-unicode-sets.html#idm140247913962048

2. 精确度

utf8_general_ci对排序支持不完善。

utf8_general_ci fails to implement all of the Unicode sorting rules, which will result in undesirable sorting in some situations, such as when using particular languages or characters.

via: http://stackoverflow.com/questions/766809/whats-the-difference-between-utf8-general-ci-and-utf8-unicode-ci

3. 结论

综上，在性能差异上可能微乎其微，重要的是精确度的情况下，还是选择utf8_unicode_ci。