Why aren't you using the best GNU/Linux distribution in existence Jow Forums?

Also Jow Forumsentoo general
See: wiki.gentoo.org/wiki/Benefits_of_Gentoo

Attached: gentoo-logo.png (666x800, 119K)

Other urls found in this thread:

wiki.gentoo.org/wiki/GCC_optimization
twitter.com/SFWRedditGifs

do cflags actually matter?

Bumping for interest. I heard that the speed gains from compiling are very small, is that true? How about memory usage? Will I be able to save lots of mb ram by removing certain features at compilation?

Do I have to use my own machines for distcc or can I find some spare online?

They do have a influence, big or small, really depends on the program itself. But it does matter. You can't install a program compiled for one machine for another, like the most basic comparison for a brainlet would be Amd and Intel
Speed gains are minimal. It exists though. My gentoo install on a core 2 duo was faster than debian 9 stretch. I don't have any benchmarks to support that claim, it responded hella faster.
>Will I be able to save lots of mb ram by removing certain features at compilation?
Ram, maybe not so much. CPU cycles at runtime?, ofcourse.
>Do I have to use my own machines for distcc or can I find some spare online?
Distcc is performed on a local network. Like PC's connected on your main router.
You can't distcc online, I think you can setup generic gentoo in a container or something and build the binaries, throw away after it completes and download the bin pkg to your pc to install in later.

Also flags like -O3, -Ofast and -Os have different consequences.
Read this : wiki.gentoo.org/wiki/GCC_optimization

Tldr the time you save from optimizations you lose from compiling everything. Unless you're running gentoo on hardware which is non standard to the point you'd have to compile everything anyway (like you're using some old ass architecture like PowerPC or SPARC or something) the optimizations are more of a cool add on that don't help a whole lot for the typical use case.

Another exception might be if you were doing something really computationally intensive like scientific computing. But then you'd have some machine in a cabinet with like two Xeons and 128gb of ram or some shit, and you'd me using a version of linux tailored to that kind of thing.

Tldr, learning about these optimizations is cool and instructive, but if you're like most people, they won't make a difference in your day today usage.

CFLAGS do matter, especially on older systems. The standard, recommended -O2 and -march=native will produce well-optimized, non-bloated code that often runs better than what you can find in binary packages. Libreoffice is the first thing that comes to mind: compiling it takes a long time, but the end result is a noticeably snappier executable.
But the real way you can squeeze out more performance and greatly reduce memory usage is not CFLAGS, and it's not even USE flags.... it's a little thing called LTO, something -O3 -funroll-loops -fomg-optimize memers don't even know exists. USE flags however still matter for excluding unused features and producing smaller binaries.
-funroll-loops and the like often just bloat your code and actually make it run slower. You shouldn't use them except for some very specific cases.
You can also consider -fipa-pta if you have a lot of RAM and graphite optmizations.

>the time you save from optimizations you lose from compiling everything
t. emerge --jobs=1, MAKEOPTS="j1", PORTAGE_NICENESS=0

Because its a meme

Because it's a waste of time.