find('a,b') is slower than find('a')+find('b'), why?

jsperf's link

I'm not a jQuery expert(not even a good user), i haven't studied the whole source co开发者_开发知识库de of it (only a little part which can't help me solve this problem).

Can somebody explain this for me?

This:

$p.find("input,select");

...uses the native querySelectorAll.

This:

$p.find("input");
$p.find("select");

...uses the native getElementsByTagName.

The getElementsByTagName is simply faster, perhaps because it is a very simple selection, and has been around longer, giving it more time for optimization.

Also, when jQuery uses querySelectorAll, it does so from the document even if you're searching from a specified context, so this has an impact. To narrow the results to the context you provided, it changes your selector to use the ID of the element, or gives it a temporary one if non exists.

I updated your test to give a few more comparisons. Shows that if you're really concerned about performance, you should use the native API.

What Ӫ_._Ӫ said above is true, but do note that the results aren't necessarily equivalent, since the first will (AFAIK) return inputs and selects in interspersed document order and the second will return them one after each other. This may not matter for your application, but it's something to note.