关于MYSQL中每个用户取1条记录的三种写法(group by xxx)

今天以前的同学问我关于这方面的sql语句，我特意记忆一下，毕竟这个也比较常见了

select * from (select * from member_payment
                order by id desc) t group by member_id limit 10

第一种是先排序，然后group，这样的话自然可以取到最适合的一条数据。
缺点很明显：using temporary; using filesort

select s.*
from (select max(id) as id from `member_payment` group by `member_id` limit 10) t
left join `member_payment` as s on t.id=s.id

第二种是联合查询

select * from `member_payment` where exists (
 select `id` from (
  select max(`id`) as id from `member_payment` group by `member_id` limit 10) t
 where t.`id`=`member_payment`.`id`
)

第三种是子查询
窃以为第二种效率最高