I built my slackware system from source code. Bootstrapped, toolchained etc.. The only thing I didn't change was the chost thinking that life would be easier if left alone @ i486. Would I have gained anything if I had used i786 or even pentium4 (pentium4-slackware-linux-gnu)?

I already got all the benefits of optimization when I used (still using) my own CFLAGS CXXFLAGS right? So changing the chost won't do anything speed wise will it?

If I used march=prescott when compiling everything am I ok to just forget about the chost value? It's not gonna change anything will it?

EDIT: Crap I think I posted in the wrong forum. Could a Mod maybe put it where it belongs (I'm thinking slackware)

If you want your post moved you should click the Report button ...

1 members found this post helpful.

Why not simply use '-march=native -mtune=native' ? Works for all x86 processors (Intel, AMD, VIA, etc.).

All I want to know is if it's worth the hassle to change my chost... I can use march=whatever cflags but will there be any difference performance wise if I change the chost from i486 to i786 after I've already compiled everything?

Hi Holering,

Don't have much time to dialog -- this CHOST guide from Gentoo should get you
enough information and links to further study on your own and, in the end my
opinion is on Slackware just leave it like it is.

Slackware is designed and built for the lowest common denominator, and as you
are finding out in your quest, it's hard to change it too much without breaking the
system and going through Hades with SlackBuild scripts.

My journey has been similar to yours ... Gentoo runs on the same hardware like a
scalded dawg compared to Slackware. The Gentoo docs and wiki are far superior to
those of Slackware. The downside is Gentoo is going to require more maintenance
to keep it at that state because it's a "rolling distro", whereas Slackware can be
left pretty much alone with just security updates.

If you need to use your computers to produce work everyday as we do here, it's a
bother to keep Gentoo updated ... unless you never run into dep issues at all.

Slackware is also much more flexible for building your own software than Gentoo, at
the cost, once again, of the base system (toolchain, libs, etc) all being i486 with
some tuning for i686 (does not compare to proper CFLAGS for modern CPUs at all).

Just my opinion ... disregard the obvious difference with that of others.

I think I'll leave the chost alone... It's already enough hassle getting updated libraries and what not and finally being able to build everything else off of that...

I'm pretty sure changing the chost won't make any difference and is only more hassle to deal with in the long wrong... I mean if I compile something (or everything) with march=prescott and omit any mtune (or just mtune=prescott) then everything should be pretty much fully optimized for a prescott based chip (or lowest common denominator being prescott) even if my chost was unchanged and is still "labeled" i486. At least that's what I'm thinking...

Yeah gentoo can be unreliable when you wanna build up your system and then start building stuff from it or just maintaining it...

It'd be pretty cool if someone made a utility (ncurses gui or maybe gtk bleh) for source compiling-optimizing a slackware system easily (and easily change the chost haha), making slackbuild scripts for anonymous software and what not.

This document may help a bit. Really, what you're looking to do is to cross-compile the whole system for your specific sub-architecture, followed by building up your system with Slackware's packages (by using existing SlackBuilds). It's a lot of work, and the easiest way to accomplish full conversion to your sub-architecture is to start from scratch with a CLFS build followed by rebuilding Slackware on top of it. I don't necessarily see the benefits outweighing the costs here (the costs being a large amount of work), but if it's a learning experience then have at it. Once you get the system compiled and working once from scratch, keeping it up to date with further Slackware releases is much easier since you just have to recompile specific applications as they are released. However, if you want to create a real system designed for your specific sub-architecture, it'll be a lot of work, at least initially.

IMO what Holering really wants, and needs to build, is LFS.

You're not going to be happy with sluggish Slackware now that
you're used distros that run so much better. You're somewhat
used to Slackware convention and $PATHs, which you can easily
duplicate or improve with LFS.