What is `git diff --patience` for? explains how patience diff may be better than myers (the default one).
Is there any reason against using it as a default?
name hints that it may be noticeably slower, speed issues are also mentioned in Examples of different results produced by the standard (Myers), minimal, patience and histogram diff algorithms But running time never was an issue for me - is it a real problem with massive files or other pathological situations?
patience algorithm also has poor result for some cases (like https://gist.github.com/roryokane/79b8ebcb3813ebd934c4) but at least in my experience it is better on typical source code diff (cases like https://gist.github.com/roryokane/6f9061d3a60c1ba41237) - though it is entirely my own observation, so most of noticed result may be a placebo.