RE: 'There are a lot of things like time frequency, types of schedulers, less no. of namespaces, size optimization etc...'
These are sooner 'cosmetic changes' rather than improvements and could impose additional risks to the system. (What about data safety, satisfaction of customers, reliability of computation, efficiency of processes, horizontal and vertical compatibility of the OS, etc.)
RE: ... 'time frequency' ...
Time frequency often only increases the energy consumption without improving the computation (unless some radical change in the method of computer processing is done), for the system usually 'skips the pulses' that it cannot process.
A lot of cosmetic changes which's bound to make the system more responsive or have better throughput (either one). Speaking of responsiveness, over a generic kernel any Linux distro does not feel as responsive as Windows, this includes the higher timer frequency which does have lower throughput as you suggest but makes the system more responsive.
Originally Posted by user-f11
RE: 'Speaking of responsiveness, over a generic kernel any Linux distro does not feel as responsive as Windows'
Right away I can give you an example to the contrary of this statement. My computer platform was slower under XP than it is under F11, and under F10 was even faster than under F11. Not to talk about the printer (HP). I even had somewhere measurements with a chronometer, but I cannot find them now.
RE: The linux kernel compilations
The idea was that if the newly compiled kernel is not better than the currently existing ones, it makes no sense. Of course a definition of what does 'better' mean should be given.
There's a difference between slow, less responsive and less throughput. If you remember, there were talks about a 200 lined patch which made Desktop Linux more responsive, the thing which Con Kolivas was talking about from day 1 he entered the kernel development and till the last day of his development days.
This's what RT kernels is all about.