Once you submit a manuscript (MS) and have had it reviewed, you are not supposed to make any substantial changes such as adding data, running additional model runs or anything that would negate earlier reviews and make new reviews necessary unless specifically asked to by the editor, for example, in response to reviews.
If you receive reviews, no matter of what sort, but so that the editor allows you revisions, you should make the revisions necessary to sort out the problems. This could mean rewriting to clear out fuzzy formulations or improving figures and potentially remaking calculations, runs whatever. But, remember that typically there is a specific time frame for such revisions and any experienced editor would also judge if the revisions necessary may fit in such a time frame. If they don't rejection would probably be the proper verdict. This means that revisions cannot be extremely large and thereby time consuming for a MS to be revised.
If you think your necessary revisions need a lot of work you should possibly withdraw the MS.
In essence the editor will (or should) determine what changes are sufficient and provide you with directions. If you think (significantly) more is necessary you MS may become so different that it essentially is a different MS. You do need to discuss this with the editor to make sure this is acceptable. After all, the editor should be able to tell you what is within reason and the time frame of the publication of papers for the conference or journal to which you submit.