You are not logged in.
Pages: 1
just wondering if there is potential for a project for P4-optimized repository / or at least kernel26 packages for different architectures
vote what cpu-family you have and not how much the total is ... it's not about frequency, it's about comand-compatibility
The impossible missions are the only ones which succeed.
Offline
It would be also interesting having the possibility to cats more than once since I use the laptop which is Athlon XP and a pc whic is a celeron 1200 (so it should be more or less equivalent to an p3).
about the celeron I am not 100% sure about sse support. I think is missing...
May be we need another option in the pool for celeron/duron processor
Offline
Err, why is early K7 the same as PentiumPro ? IMHO every K7 is at least PIII or more.
Anyways, I need to do multpile answers :-)
Offline
...
about the celeron I am not 100% sure about sse support. I think is missing...
cat /proc/cpuinfo
says what?
bye neri
Offline
Now I am in the office with the lapto, later in the night.
Offline
Err, why is early K7 the same as PentiumPro ? IMHO every K7 is at least PIII or more.
in power, it is for sure, i mean the architecture integration in multimedia ... there the first really great thing from AMD was the 3dNOW!
Anyways, I need to do multpile answers :-)
i know what you mean ... but i didnt made this forum and dont know how to vote multiple (maybe as guest?)
The impossible missions are the only ones which succeed.
Offline
andy wrote:Err, why is early K7 the same as PentiumPro ? IMHO every K7 is at least PIII or more.
in power, it is for sure, i mean the architecture integration in multimedia ... there the first really great thing from AMD was the 3dNOW!
But 3dNOW! was introduced with the K6 or K6-II. I have a K7-600Mhz (yes, with that gigantic slot and even larger cooling-arrangement .... it doesn't get much earlier in K7 than that), and it has 3dNOW!. You must have mixed up something ;-)
Offline
um so are you gonna maintain an unofficial p4 repo because i doubt the current developers have time to create a repository for each processor type out there. the only thing i could possibly see them doing is changing the -march to -mcpu but even then i don't think -mcpu is as broad serving as -march instruction wise.
then there is the i586 port that some people want which has been on the want list for a long time. nope i think arch should just steer the course as is. it is too much work for the few devs there are to bother adding p4, i586, amd64, etc repos ... you have abs and makeworld for that.
AKA uknowme
I am not your friend
Offline
t it is too much work for the few devs there are to bother adding p4, i586, amd64, etc repos ... you have abs and makeworld for that.
Just that for amd64 I think it is worth to create an own port. However, as I am currently trying to do something like that, it turns out to be pretty ugly and throwing it simply into ABS/makeworld is not that simple. I have most of abs/base in 64bit, but some essential parts still are missing.
Offline
sarah31 wrote:t it is too much work for the few devs there are to bother adding p4, i586, amd64, etc repos ... you have abs and makeworld for that.
Just that for amd64 I think it is worth to create an own port. However, as I am currently trying to do something like that, it turns out to be pretty ugly and throwing it simply into ABS/makeworld is not that simple. I have most of abs/base in 64bit, but some essential parts still are missing.
yes it is. in your case though you may want to get a crosscompiler?
for me it was mostly nigly little things in the PKGBUILDs, etc, etc. porting is a grand nuisance if you have to rely on others
AKA uknowme
I am not your friend
Offline
about the k6 stuff i was wrong, sorry ... messed up earlier series of AMD with k6 -- it indeed has 3dnow! -> any idea how to change the entry for voting?
my shorttime-idea was to make a kernel26-P4 pkg that runs optimized on P4 and is in a repo for easy update for the whole comunity ... but as i have an upload of 2kb/s, it can take horribly long
the longtime-idea is to make a GRID that compiles every pkg optimized for several architectures and have them in new repositories
why this longterm idea? well once we figured out how to create the whole thing from tabula rasa, you can with the powerfull tools of archlinux: pacman/makepkg/abs have a great linux optimized for every architecture in binary form
i fully understand that this do not work the classical way with so little much people --- there should be some way of automatizing things (compiling new versions for all architectures -> compiled over GRID offering very fast results and possibility to have it really optimized for at least p4 and AMD64bit)
The impossible missions are the only ones which succeed.
Offline
then you have to have people in the GRID that you trust. allowing anyone the the ability to alter arch repos is just plain stupid then the develoers lose all control because anyone who wants to mod the packages could. only one or two people should have access to the base repo not every developers.
you cannot convince me that a GRID is not a security/control risk. and like i said before regardless of grids or no grids you will still be relying on the devs for properly made builds, etc.
having three million different architectures also makes managing pacman and such a little more complex.
who is going to fund all the bandwidth this will use repo wise? i know judd probably does not make enough in donations to pay for it.
logistically this would be a nightmare imho.
AKA uknowme
I am not your friend
Offline
yes it is. in your case though you may want to get a crosscompiler?
.... if it was just the cross compiler .... but crosscompiling glibc, gcc etc. is quite ...errr ... well ... most importantly : you need to get the right combinations of versions of gcc and glibc, otherwise it just won't work. Luckily, that thing is so damn fast that it doesn't matter how many times you recompile gcc and glibc. Still, it's a nightmare .... anyways, enough rambling ...
But I also need to make (non-trivial) modifications to the PKGBUILDS, and I still haven't really tested the results ....
Offline
then you have to have people in the GRID that you trust. allowing anyone the the ability to alter arch repos is just plain stupid then the develoers lose all control because anyone who wants to mod the packages could. only one or two people should have access to the base repo not every developers.
no intention to allow anyonw to do anything: GRID is centralized: only on one computer you have control over the project ... you only spread the "dirt work of computing" to the clients, that have no possibility to modify
you cannot convince me that a GRID is not a security/control risk. and like i said before regardless of grids or no grids you will still be relying on the devs for properly made builds, etc.
of course you will rely on the devs for properly made builds ... but when the dev starts makepkg, it generates one pkg ... my idea is to, instead of one, generate different pkgs
having three million different architectures also makes managing pacman and such a little more complex.
3<10^6
-P4
-AMD64bit
-PowerMac64bit
who is going to fund all the bandwidth this will use repo wise? i know judd probably does not make enough in donations to pay for it.
that's a good point!
logistically this would be a nightmare imho.
imagine FedEx about 100years ago .. everyone would say something like this would be a nightmare ... but now it's reality
it's not a project for next week ... it's the future 8)
The impossible missions are the only ones which succeed.
Offline
http://www.wbglinks.net/pages/reads/hac … mpson.html
It's these sorts of things that makes me not trust any sort of distributed compiling project.
I have discovered that all of mans unhappiness derives from only one source, not being able to sit quietly in a room
- Blaise Pascal
Offline
well in the future dp.
arch needs a much much larger development community and a different organizational structure (ie separate abs repos, different heads of each optimized tree head developers that can adjust the code of things like pacman for the various archtectures, developers competent enough to build various cvsup binaries (not an easy task) ) and so forth.
arch needs something like clc ( if you wanna know what clc iis just visit the crux homepage) this would make project like powerpc optimized arch linux a little more easier. it also absolves judd from being responsible for how other people want to manage his distro. at crux per provides tha core and clc provides the rest. what clc does is not per's responsiblity.
AKA uknowme
I am not your friend
Offline
Hi all,
For the information.
I'm a fresh french user of Arch Linux.
I'm installing it to have a test on my laptop Compaq Evo N600c powered by an Intel 1.2Ghz PIII-m.
Regards,
Joe.
Pages: 1