This is an excellent question. My general feeling about this is that you should not write parallel optimised code if you are not going to target a multi-core machine as there is a performance overhead. However, in terms of respecting the scalability of your application it is perhaps something you should consider if there is a possibility your code will eventually run on a multi-core machine or even a mixed envoironment.
Personally, I would tend not to write parallel optimised code until I feel there is a serious need for it. E.g. a slow running operation or hefty report etc. In these cases you can always look into running processes on a separate thread so as to not lock up the user interface while processing is being completed. I found the this article which also discusses this topic and maybe of some use.