Well first of all i wouldnt ge tinto the habit of making api calls with the url loading functions of simpleXML and DOMDocument... Youre better off using cURL, file_get_contents, fopen, etc.. That way you can catch any issues with the request/response before you get the parsing part, ie. youve got a better ability to handle errors and direct control flow and log/present more meaningful messages about whats going wrong.
With that said i would never create and destroy multiple cURL resource handlers (if thats what you mean) jsut open a single one and then keep resetting the attributes and dispatching a new request.
As far as which one is actually faster or less resource intensive, i think thats the least of the issues when compared to the other things i mentioned above... but then thats jsut my opinion i suppose.