Re: Perfomance difference between utf8_unicode_ci and utf8_general_ci

[Rick James]

UTF8_unicode_ci and UTF8_general_ci are collations; both apply to utf8 CHARACTER SET. Do not expect any data differences.

I have seen one datapoint -- utf8_unicode_ci was about 30% slower than utf8_general_ci. BUT... It will depend a lot on the data you have, and the operation you are performing.

This make take an hour on a 10 million row table. Is that what you are doing?
SELECT ... WHERE foo LIKE '%...'; (where ... is some strange char(s))
I would worry more about the hour than about the 20 minutes extra.

SELECT ... WHERE foo = '...';
will be sub-second, regardless of ... and collation (assuming an appropriate index on foo). So, use the 'right' collation and don't worry about speed.