It is highly recommended to make use of a bytecode cache like OPcache. A bytecode cache removes the need for parsing PHP code on every request and can greatly improve performance.
"If you care about performance and don't use a bytecode cache then you don't really care about performance. Please get one and start using it."
Stas Malyshev, Core Contributor to PHP and Zend Employee
As already mentioned earlier in the chapter about configuring Doctrine, it is strongly discouraged to use Doctrine without a Metadata and Query cache (preferably with APC or Memcache as the cache driver). Operating Doctrine without these caches means Doctrine will need to load your mapping information on every single request and has to parse each DQL query on every single request. This is a waste of resources.
Make effective use of the available alternative query result formats like nested array graphs or pure scalar results, especially in scenarios where data is loaded for read-only purposes.
Starting with Doctrine 2.1 you can mark entities as read only (See metadata mapping references for details). This means that the entity marked as read only is never considered for updates, which means when you call flush on the EntityManager these entities are skipped even if properties changed. Read-Only allows to persist new entities of a kind and remove existing ones, they are just not considered for updates.
If entities hold references to large collections you will get performance and memory problems initializing them. To solve this issue you can use the EXTRA_LAZY fetch-mode feature for collections. See the tutorial for more information on how this fetch mode works.
A lot of the points mentioned in the Best Practices chapter will also positively affect the performance of Doctrine.
See Best Practices