FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Gentoo > Gentoo User

 
 
LinkBack Thread Tools
 
Old 02-05-2008, 12:33 PM
Jerry McBride
 
Default CHOST question.

Morning...

A small question to satisfy my curiosity about the CHOST setting
in /etc/make.conf...


Currently I have CHOST="i686-pc-linux-gnu" on a computer with a pentium4
processor. Would it make any differences, at all, to change this to
CHOST="pentium4-pc-linux-gnu" ?

Would the compiler then be optimized for the pentium4 and thus run a tad bit
faster?

Thank you, in advance....

P.S. Please don't warn me about changing chost, I've been through it
before.:')




--


From the Desk of: Jerome D. McBride
--
gentoo-user@lists.gentoo.org mailing list
 
Old 02-05-2008, 01:18 PM
"Benedikt Morbach"
 
Default CHOST question.

Hi,

no, it would not.
gcc would simply refuse to work, because CHOST="pentium4-pc-linux-gnu"
is not a valid CHOST.
CHOST describes the platform you build on. For optimizations take a
look at CFLAGS.

And by the way: Changing CHOST is not worth the trouble. Even if it
would be possible in your case, it would be better to do a new
install. It's faster (because when you change CHOST, you should at
least run emerge -e system && emerge -e world) and less likely to
break.
--
gentoo-user@lists.gentoo.org mailing list
 
Old 02-05-2008, 01:40 PM
Alan McKinnon
 
Default CHOST question.

On Tuesday 05 February 2008, Jerry McBride wrote:
> Would the compiler then be optimized for the pentium4 and thus run a
> tad bit faster?

See Benedikt's answer for why you should not go down this road.

If you did get it all to work right, and suffered through the emerge -e
world required, your computer would in fact run a tiny tad faster,
where tad is defined is "a teensy weensy little bit, so small you can
hardly see it with a magnifying glass"

Not worth the effort IMHO. Of course, there are ricers out there that
will swear by it and declare that their machine runs much faster, but
very few if any of them ever produce some actual numbers...

--
Alan McKinnon
alan dot mckinnon at gmail dot com
--
gentoo-user@lists.gentoo.org mailing list
 
Old 02-05-2008, 01:43 PM
Jerry McBride
 
Default CHOST question.

On Tuesday 05 February 2008 09:18:17 am Benedikt Morbach wrote:
> Hi,
>
> no, it would not.
> gcc would simply refuse to work, because CHOST="pentium4-pc-linux-gnu"
> is not a valid CHOST.
> CHOST describes the platform you build on. For optimizations take a
> look at CFLAGS.
>

Where do I find a list of valid chosts? I've been digging since my first post
and the Gentoo Handbook on this page says:

http://www.gentoo.org/doc/en/handbook/draft/complete/handbook.xml?part=2&chap=5

<QUOTE>
# info gcc
Select GCC Command Options,
Submodel Options,
and pick your architecture.
<UNQUOTE>

On the submodel page for i386, it clearly lists the pentium4...

My question is, what difference in performance would this change make?

Thank you, for the post.



--


From the Desk of: Jerome D. McBride
--
gentoo-user@lists.gentoo.org mailing list
 
Old 02-05-2008, 02:23 PM
Jerry McBride
 
Default CHOST question.

On Tuesday 05 February 2008 09:40:30 am Alan McKinnon wrote:
> On Tuesday 05 February 2008, Jerry McBride wrote:
> > Would the compiler then be optimized for the pentium4 and thus run a
> > tad bit faster?
>
> See Benedikt's answer for why you should not go down this road.
>
> If you did get it all to work right, and suffered through the emerge -e
> world required, your computer would in fact run a tiny tad faster,
> where tad is defined is "a teensy weensy little bit, so small you can
> hardly see it with a magnifying glass"
>
> Not worth the effort IMHO. Of course, there are ricers out there that
> will swear by it and declare that their machine runs much faster, but
> very few if any of them ever produce some actual numbers...
>

Thanks for the post.

I actually started working on this project late last night... My target test
machine is an "getting old" Compaq R3000 with a 3ghz P4. What I'm going to do
is just what you suggested.

First I'm going to finish freshening the laptop. This is my
daily "hack-n-slash" computer, so no worries clobbering it. I'm near the end
of finishing an "emerge -e world" that was preceded with two rounds
of "emerge -e system"

Next step is some exhaustive bench marking. All suggestions welcomed.

Then once completed, I' make the change to chost from i686 to pentium4,
following the docs on the net. Once done and smoothed out... another
freshening as mentioned above, followed up with identical runs of what ever
benchmarks I ran before...

Should be interesting... It'll lay to rest what everyone speculates or
postulates. :')

Cheers.


--


From the Desk of: Jerome D. McBride
--
gentoo-user@lists.gentoo.org mailing list
 
Old 02-05-2008, 02:28 PM
Dale
 
Default CHOST question.

Jerry McBride wrote:
> On Tuesday 05 February 2008 09:40:30 am Alan McKinnon wrote:
>
>> On Tuesday 05 February 2008, Jerry McBride wrote:
>>
>>> Would the compiler then be optimized for the pentium4 and thus run a
>>> tad bit faster?
>>>
>> See Benedikt's answer for why you should not go down this road.
>>
>> If you did get it all to work right, and suffered through the emerge -e
>> world required, your computer would in fact run a tiny tad faster,
>> where tad is defined is "a teensy weensy little bit, so small you can
>> hardly see it with a magnifying glass"
>>
>> Not worth the effort IMHO. Of course, there are ricers out there that
>> will swear by it and declare that their machine runs much faster, but
>> very few if any of them ever produce some actual numbers...
>>
>>
>
> Thanks for the post.
>
> I actually started working on this project late last night... My target test
> machine is an "getting old" Compaq R3000 with a 3ghz P4. What I'm going to do
> is just what you suggested.
>
> First I'm going to finish freshening the laptop. This is my
> daily "hack-n-slash" computer, so no worries clobbering it. I'm near the end
> of finishing an "emerge -e world" that was preceded with two rounds
> of "emerge -e system"
>
> Next step is some exhaustive bench marking. All suggestions welcomed.
>
> Then once completed, I' make the change to chost from i686 to pentium4,
> following the docs on the net. Once done and smoothed out... another
> freshening as mentioned above, followed up with identical runs of what ever
> benchmarks I ran before...
>
> Should be interesting... It'll lay to rest what everyone speculates or
> postulates. :')
>
> Cheers.
>
>
>

There is a script that will take care of the emerge and you only have to
do it once. It's on the forums but I still have a copy if you want me
to email it to you.

Dale

:-) :-)
--
gentoo-user@lists.gentoo.org mailing list
 
Old 02-05-2008, 02:35 PM
Graham Murray
 
Default CHOST question.

"Benedikt Morbach" <benedikt.morbach@googlemail.com> writes:

> Hi,
>
> no, it would not.
> gcc would simply refuse to work, because CHOST="pentium4-pc-linux-gnu"
> is not a valid CHOST.
> CHOST describes the platform you build on. For optimizations take a
> look at CFLAGS.

Though looking at /usr/share/gnuconfig/config.sub it looks as though it
might be valid, and be canonicalized to 'i786-pc-linux-gnu' (rather than
the more common i686-pc-linux-gnu)
--
gentoo-user@lists.gentoo.org mailing list
 
Old 02-05-2008, 02:35 PM
Alan McKinnon
 
Default CHOST question.

On Tuesday 05 February 2008, Jerry McBride wrote:
> Should be interesting... It'll lay to rest what everyone speculates
> or postulates. :')

No need. Been done. Question answered long ago. You are beating a dead
horse. We already know *exactly* what difference it makes - precious
little.

You want a machine that performs better? Stick in a disk drive with more
cache memory. Instant improvement that will dwarf any change you could
ever make with the compiler. Ever wondered why Ubuntu distributes 386
generic code? Because it makes no discernible difference whatsoever.

But if you wanna go ahead and prove to yourself something that the
toolchain world has know for like forever, then go ahead, don't let me
stop you <shrug>

--
Alan McKinnon
alan dot mckinnon at gmail dot com
--
gentoo-user@lists.gentoo.org mailing list
 
Old 02-05-2008, 02:36 PM
Jerry McBride
 
Default CHOST question.

On Tuesday 05 February 2008 10:28:01 am Dale wrote:
> Jerry McBride wrote:
> > On Tuesday 05 February 2008 09:40:30 am Alan McKinnon wrote:
> >> On Tuesday 05 February 2008, Jerry McBride wrote:
> >>> Would the compiler then be optimized for the pentium4 and thus run a
> >>> tad bit faster?
> >>
> >> See Benedikt's answer for why you should not go down this road.
> >>
> >> If you did get it all to work right, and suffered through the emerge -e
> >> world required, your computer would in fact run a tiny tad faster,
> >> where tad is defined is "a teensy weensy little bit, so small you can
> >> hardly see it with a magnifying glass"
> >>
> >> Not worth the effort IMHO. Of course, there are ricers out there that
> >> will swear by it and declare that their machine runs much faster, but
> >> very few if any of them ever produce some actual numbers...
> >
> > Thanks for the post.
> >
> > I actually started working on this project late last night... My target
> > test machine is an "getting old" Compaq R3000 with a 3ghz P4. What I'm
> > going to do is just what you suggested.
> >
> > First I'm going to finish freshening the laptop. This is my
> > daily "hack-n-slash" computer, so no worries clobbering it. I'm near the
> > end of finishing an "emerge -e world" that was preceded with two rounds
> > of "emerge -e system"
> >
> > Next step is some exhaustive bench marking. All suggestions welcomed.
> >
> > Then once completed, I' make the change to chost from i686 to pentium4,
> > following the docs on the net. Once done and smoothed out... another
> > freshening as mentioned above, followed up with identical runs of what
> > ever benchmarks I ran before...
> >
> > Should be interesting... It'll lay to rest what everyone speculates or
> > postulates. :')
> >
> > Cheers.
>
> There is a script that will take care of the emerge and you only have to
> do it once. It's on the forums but I still have a copy if you want me
> to email it to you.
>
> Dale
>
> :-) :-)

Thanks for the offer. I'm almost finished the re-compiling stuff however. Why
not post the script anyways? Someone else may be doing the same thing.

Cheers.

--


From the Desk of: Jerome D. McBride
--
gentoo-user@lists.gentoo.org mailing list
 
Old 02-05-2008, 02:52 PM
Jerry McBride
 
Default CHOST question.

On Tuesday 05 February 2008 10:35:34 am Alan McKinnon wrote:
> On Tuesday 05 February 2008, Jerry McBride wrote:
> > Should be interesting... It'll lay to rest what everyone speculates
> > or postulates. :')
>
> No need. Been done. Question answered long ago. You are beating a dead
> horse. We already know *exactly* what difference it makes - precious
> little.
>
> You want a machine that performs better? Stick in a disk drive with more
> cache memory. Instant improvement that will dwarf any change you could
> ever make with the compiler. Ever wondered why Ubuntu distributes 386
> generic code? Because it makes no discernible difference whatsoever.
>
> But if you wanna go ahead and prove to yourself something that the
> toolchain world has know for like forever, then go ahead, don't let me
> stop you <shrug>
>
> --
> Alan McKinnon
> alan dot mckinnon at gmail dot com

Are the numbers posted somewhere I can get to? It'd be good reading.

--


From the Desk of: Jerome D. McBride
--
gentoo-user@lists.gentoo.org mailing list
 

Thread Tools




All times are GMT. The time now is 07:56 PM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright 2007 - 2008, www.linux-archive.org