The second form may not work reliably. Consider:
Session A reads count. It's 320.
Session B reads count. It's 320.
Session A adds 1 to count and saves new count, value 321.
Session B adds 1 to the count it read and saves new count, value 321.
Count is now wrong by one. You can dodge this with SELECT... FOR UPDATE but that's not the best way to do it.
If you use count=count+1 both connections can get it right.
The count=321 form has to do at least two exchanges with the server, once to read the value, then to set the value. The count=count+1 form can be done with a single exchange so it'll be faster in elapsed time.
If using InnoDB you should also read
http://dev.mysql.com/doc/mysql/en/innodb-transaction-isolation.html