Writing code at assembler level for any processor is a tedious business, because you have to take care of every little detail yourself - and that is precisely what gives assembler its power!

Because you are entirely in control with assembler, the only limit is your own skill & experience - the only thing to stop you writing the tightest, fastest, smallest code is your own expertise.

Or, to put it another way, any bloat or inefficiency is entirely your own fault!


Processor resources (memory, MIPS, etc) are relatively cheap these days, but programmers' time is expensive - hence it often makes sense to use tools that maximise programmer productivity, even at the cost of some processor inefficiency.

In many cases, it's not just the cost of the development time, but the fact that an extended development time can lose you the market opportunity.

The exact break-even point will, of course, depend upon the scale & complexity of the problem, the skill and cost of your programmer resources, the production volume, and many other factors.