If you're not going to use timers, the only solution is to use some piece of software with a known execution time (4s in this example). The commonest way to do this is with a loop, where you count the total number of instruction cycles "consumed" by the loop - and adjust that to your required delay.

But, if your processor is stuck in such a timing loop, it will be totally 100% dedicated to that task: it will not be ale to do any other thing - let alone "lots of things at the same time".
If you want timing without locking-up your processor, you must use a hardware timing function - whether on-chip or external hardware (unless you can precisely calculate the execution times of every other part of your system...)