well you can use either one. There are advantages and disadvantages to both methods. If you are going to texture the poly model and then convert to subdivs you have to be aware that you may end up with parts of the texture looking very clear cut (with lots of edges) but it's easy enough to simply edit your texture and reload it. Laying out uv's can be a tricky thing at first but here goes...

okay if you have a polymodel, (you must select the model you want to layout first) in the modelling menu set select edit polygons>texture and you will have a range of different options. if you select the option box next to automatic mapping you have two main options to choose from 1) less distortion and 2) fewer pieces. Choose which one you want and click project.

now if you open up the UV texture editor by selecting panels>panel>UV texture editor you should see several shapes of your model, in here you can cut and paste your UV's in a way that makes it easy to texture (if you need help with that then ask me). then if you go to polygons>uv snapshot and save the image you will have a snapshot of those UV's.

Then what you can do is go into something like photoshop open your snapshot (don't forget where you saved the snapshot) and create a new layer to paint on and then if you delete the snapshot layer and save that file as a texture then you can load that texture as your colour file, it will map it in the way that the snapshot is arranged. This just one particular method of mapping there are others but this offers the best arrangement without distorting too much for begginners.

There are ways of selecting particular groups of UV's and texturing each group seperately but that would take too long to explain. The same things apply for subdiv UV layouts as well. Hopefully that has helped if you are totally confusedd buy all of this then ur best bet is to have a look on maya's help it tends to be pretty good at explaining things.