Short answer: you can't measure performance for engineering. No matter what metric you use, it won't fit all cases and this is more so the bigger your team becomes.
Long answer: it is a very gray area. First of all, money is not a good motivator; if you want to motivate people, let them master a skill. Secondly, what works in a small software company will most likely not work with a bigger team. People will find tricks to game the system while PM will rely on "metrics" for a false sense of security rather than personal judgment and familiarity with individuals.
You can, however, add a system to make sure performance reviews are fair: transparency. Let the engineers chose their own goals for the next period (usually six months); have the manager agree on them. Make sure to throw in the mix a bit of technical and a bit of human side, e.g. "I will learn a new technology/language/library and give at least one tech talk on the topic to my fellow engineers". At the end of the period, review the goals, let the engineer self-assess his performance and the manager double check how they did. Be aware that engineers tend to under-estimate themselves. Be fair. Record everything in written form.
Then, if the majority of the goals are met, you can talk about raises. But do not base raises solely on this, otherwise you risk lots of anger if somebody's performance fails because of other team members.
See also this SO question, or this Dilbert comic for the funny side of things. Rands has, as always, some good insight on the topic (see also this other post).
One word of advice: do not assume that an appraisal system is a way to save time. Properly done, a good appraisal system takes more time than not doing it.
UPDATE: see also this answer on the Project Management stackexchange site.