Importance of tweaking/stripping the kernel?

[aimeec1995]

So I have a question about the FreeBSD kernel.
I have ever been a user of the NODEBUG, but I have a question about tweaking it ...

Are there advantages like performance or boot time increases for removing support for unused devices or compiling drivers into the kernel?

[Oko]

Short answer is no. Long answer is if you were competent to increase performance of FreeBSD kernel by tweaking it you would not be asking the question on a hobbyist forum. Most definitely do not touch your kernel configuration files unless you are doing this for educational purposes and you can effort to break the system. Boot times are irrelevant for servers. The only time server is rebooted is when kernel security patches are applied (few times a month on a well maintained systems).

[hanzer]

Originally Posted by aimeec1995

[snip]
Are there advantages like performance or boot time increases for removing support for unused devices or compiling drivers into the kernel?

Sure. Although, an end user on modern hardware probably wouldn't notice the difference; and the maintenance overhead is probably much higher; but if you have a spare machine for experimentation, exploring the FreeBSD source-build infrastructure can be a wonderfully informative exercise - it is nicely engineered, well documented, and plenty of people understand it. Also, if you ever need to deploy a low-resource embedded or dedicated system, building and tuning know-how could be especially useful. So, if you have a spare machine, I say go for it! There is nothing exotic about building a custom system from source and it is perfectly within reach of a semi-serious hobbyist, IMO. In many ways, FreeBSD is an awesome choice as a roll-your-own OS.