Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
GCC 4.1.0 - is it safe/worth ?
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3, 4, 5, 6  
Reply to topic    Gentoo Forums Forum Index Portage & Programming
View previous topic :: View next topic  
Author Message
SnEptUne
l33t
l33t


Joined: 23 Aug 2004
Posts: 656

PostPosted: Sun Jun 04, 2006 1:31 pm    Post subject: Re: following the gcc hype Reply with quote

nivw wrote:
I just tried re-emerging world on two difrent pcs, and got ERROR: sys-devel/gcc-4.1.1 failed
one is a pentium-4 with CFLAGS="-march=pentium4 -O2 -pipe -fomit-frame-pointer" ; CXXFLAGS="${CFLAGS}"
other is a athlon-xp with CFLAGS="-O2 -march=athlon-xp -pipe -fomit-frame-pointer -ffast-math -ftracer -fprefetch-loop-arrays -fno-ident"
the log for the athlon pc is here www.cs.bgu.ac.il/~nivv/gcc-4.11.fail.tar.bz2

uname -r for the athlon: 2.6.16-ck11

I used to have nosrip in both machines in the FEATURES , and just took it out.
I couldnt find any open bug , or issues in the forums.

Thanks,
NIv


I couldn't compile gcc 4.1.1 either. I am currently using gcc-3.4.4-r1 with glibc-2.3.6-r3. My CFLAGS is stable on Gcc 3.x, but I am not sure about GCC 4.1.x. Here's my CFLAGS:

Code:
CFLAGS="-O2 -march=athlon -pipe"


Do you have any success on compiling GCC 4.1? Or should I wait before it stablized?
_________________
"There will be more joy in heaven over the tear-bathed face of a repentant sinner than over the white robes of a hundred just men." (LM, 114)
Back to top
View user's profile Send private message
boniek
Guru
Guru


Joined: 26 Mar 2005
Posts: 373

PostPosted: Sun Jun 04, 2006 1:56 pm    Post subject: Reply with quote

Just another success story. ~x86 with CFLAGS -march=athlon-xp -O2 -fomit-frame-pointer -pipe
Back to top
View user's profile Send private message
Halcy0n
Developer
Developer


Joined: 17 Sep 2003
Posts: 1682
Location: Freehold, NJ

PostPosted: Sun Jun 04, 2006 9:18 pm    Post subject: Re: following the gcc hype Reply with quote

SnEptUne wrote:

I couldn't compile gcc 4.1.1 either. I am currently using gcc-3.4.4-r1 with glibc-2.3.6-r3. My CFLAGS is stable on Gcc 3.x, but I am not sure about GCC 4.1.x. Here's my CFLAGS:

Code:
CFLAGS="-O2 -march=athlon -pipe"


Do you have any success on compiling GCC 4.1? Or should I wait before it stablized?


What is the error you are getting? We need something to work off of :) Give us the last 20 or so lines from the build log. Wherever it looks like the error begins, but atleast 20 to get context.
_________________
Mark Loeser
http://www.halcy0n.com
Back to top
View user's profile Send private message
Alejandro Nova
n00b
n00b


Joined: 08 Sep 2004
Posts: 50

PostPosted: Sun Jun 04, 2006 10:22 pm    Post subject: Reply with quote

I found that the -freorder-blocks-and-partition CFLAG speeds up things, but it also break some things, the same way that -fvisibility-inlines-hidden does. I can't compile the following binaries (they give me the beautiful message "relocation R_X86_64_32 against `a local symbol' can not be used when making a shared object; recompile with -fPIC"): gettext, gmp, libwmf, imagemagick, libsvg, db, vdrift (custom ebuild), enchant, libxslt, foomatic-db-engine. Also I'm having trouble with the media icon placement on desktop with KDE. So, i'm rebuilding world without that flag.
_________________
Becoming someone beautiful, through my music, my silent devotion...
Alejandro Nova™.
Back to top
View user's profile Send private message
katswaio
n00b
n00b


Joined: 28 Jan 2006
Posts: 21
Location: Iasi,ro

PostPosted: Mon Jun 05, 2006 9:38 pm    Post subject: Reply with quote

Upgraded from gcc-3.3.6 to 4.1.1. Reemerged system packages and also kde packages in order to use the new kdehiddenvisibility use flag in kde-3.5.3.
I was thinking about skipping the 'emerge -e world' step, since it would take a long time on my system. So I emerged libstdc++-v3 for packages compiled with gcc 3.x. But, for example, amarok wouldn't work until I recompiled taglib. I figured there are issues with programs compiled with 4.1.1 that are linking against libraries compiled with 3.3.6. Is it so?
Is there any way around this or I really have to 'emerge -e world' ?
Back to top
View user's profile Send private message
SnEptUne
l33t
l33t


Joined: 23 Aug 2004
Posts: 656

PostPosted: Tue Jun 06, 2006 2:10 am    Post subject: Re: following the gcc hype Reply with quote

Halcy0n wrote:
SnEptUne wrote:

I couldn't compile gcc 4.1.1 either. I am currently using gcc-3.4.4-r1 with glibc-2.3.6-r3. My CFLAGS is stable on Gcc 3.x, but I am not sure about GCC 4.1.x. Here's my CFLAGS:

Code:
CFLAGS="-O2 -march=athlon -pipe"


Do you have any success on compiling GCC 4.1? Or should I wait before it stablized?


What is the error you are getting? We need something to work off of :) Give us the last 20 or so lines from the build log. Wherever it looks like the error begins, but atleast 20 to get context.


That's strange. When I compile GCC 4.1 again, it works fine. I haven't update anything. Maybe it wouldn't work if I compile several software at once?
_________________
"There will be more joy in heaven over the tear-bathed face of a repentant sinner than over the white robes of a hundred just men." (LM, 114)
Back to top
View user's profile Send private message
red-wolf76
l33t
l33t


Joined: 13 Apr 2005
Posts: 714
Location: Rhein-Main Area

PostPosted: Sat Jun 24, 2006 7:43 am    Post subject: Re: following the gcc hype Reply with quote

SnEptUne wrote:
Halcy0n wrote:
SnEptUne wrote:

I couldn't compile gcc 4.1.1 either. I am currently using gcc-3.4.4-r1 with glibc-2.3.6-r3. My CFLAGS is stable on Gcc 3.x, but I am not sure about GCC 4.1.x. Here's my CFLAGS:

Code:
CFLAGS="-O2 -march=athlon -pipe"


Do you have any success on compiling GCC 4.1? Or should I wait before it stablized?


What is the error you are getting? We need something to work off of :) Give us the last 20 or so lines from the build log. Wherever it looks like the error begins, but atleast 20 to get context.


That's strange. When I compile GCC 4.1 again, it works fine. I haven't update anything. Maybe it wouldn't work if I compile several software at once?
That's generally a Bad Thing®. Compiling two or more packages at once may seem feasible (I've done it before, too!) but won`t help you all that much, as the MAKEOPTS variable in your make.conf will govern how many instances will hit your processor(s) at a time. As I read elsewhere, opening new emerges in other consoles will make the compiles fight for processor time and stress your hard drive. I guess for long compiles that include many large updates it may be clever to start an instance with "--fetchonly" some time before starting the actual emerge. That will keep your processor from idling while portage fetches source from the servers.

Cheers!
_________________
0mFg, G3nt00 r0X0r$ T3h B1g!1111 ;)

Use sane CFLAGS! If for no other reason, do it for the lulz!
Back to top
View user's profile Send private message
spielc
Guru
Guru


Joined: 20 Apr 2004
Posts: 452

PostPosted: Sat Jun 24, 2006 8:36 am    Post subject: Reply with quote

katswaio wrote:
Upgraded from gcc-3.3.6 to 4.1.1. Reemerged system packages and also kde packages in order to use the new kdehiddenvisibility use flag in kde-3.5.3.
I was thinking about skipping the 'emerge -e world' step, since it would take a long time on my system. So I emerged libstdc++-v3 for packages compiled with gcc 3.x. But, for example, amarok wouldn't work until I recompiled taglib. I figured there are issues with programs compiled with 4.1.1 that are linking against libraries compiled with 3.3.6. Is it so?
Is there any way around this or I really have to 'emerge -e world' ?


Code:
revdep-rebuild
is your friend in this case!

be aware that it might miss some things tho... So if you want to be on the safe side you have to do the emerge -e world
Back to top
View user's profile Send private message
yngwin
Retired Dev
Retired Dev


Joined: 19 Dec 2002
Posts: 4572
Location: Suzhou, China

PostPosted: Sat Jun 24, 2006 11:46 am    Post subject: Re: following the gcc hype Reply with quote

red-wolf76 wrote:
I guess for long compiles that include many large updates it may be clever to start an instance with "--fetchonly" some time before starting the actual emerge. That will keep your processor from idling while portage fetches source from the servers.

If you update portage to 2.1, then you can use FEATURES="parallel-fetch" (in make.conf) with the same result.

Quote:
I really have to 'emerge -e world' ?

It is highly recommended.
_________________
"Those who deny freedom to others deserve it not for themselves." - Abraham Lincoln
Free Culture | Defective by Design | EFF
Back to top
View user's profile Send private message
vovin
n00b
n00b


Joined: 07 May 2006
Posts: 16

PostPosted: Sat Jun 24, 2006 1:04 pm    Post subject: Reply with quote

I ran benchmark nbench compiled with gcc 3.4.6 and 4.1.1,
CFLAGS="-O2 -finline-functions -g0 -mmmx -msse -march=pentium3 -mfpmath=sse -fomit-frame-pointer -s -mtune=pentium3 -ffast-math"
with following results:
Code:

BYTEmark* Native Mode Benchmark ver. 2 (10/95)
Index-split by Andrew D. Balsa (11/97)
Linux/Unix* port by Uwe F. Mayer (12/96,11/97)

TEST                : Iterations/sec.  : Old Index   : New Index
                    :                  : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT        :          949.44  :      24.35  :       8.00
STRING SORT         :          39.162  :      17.50  :       2.71
BITFIELD            :      1.8586e+08  :      31.88  :       6.66
FP EMULATION        :          65.641  :      31.50  :       7.27
FOURIER             :           12877  :      14.64  :       8.23
ASSIGNMENT          :          9.5352  :      36.28  :       9.41
IDEA                :          1770.8  :      27.08  :       8.04
HUFFMAN             :          871.63  :      24.17  :       7.72
NEURAL NET          :          12.964  :      20.83  :       8.76
LU DECOMPOSITION    :           509.4  :      26.39  :      19.06
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX       : 26.889
FLOATING-POINT INDEX: 20.040
Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
CPU                 : GenuineIntel Intel(R) Celeron(TM) CPU                1200MHz 1193MHz
L2 Cache            : 256 KB
OS                  : Linux 2.6.17
C compiler          : 3.4.6
libc                :
MEMORY INDEX        : 5.537
INTEGER INDEX       : 7.750
FLOATING-POINT INDEX: 11.115
Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.

BYTEmark* Native Mode Benchmark ver. 2 (10/95)
Index-split by Andrew D. Balsa (11/97)
Linux/Unix* port by Uwe F. Mayer (12/96,11/97)

TEST                : Iterations/sec.  : Old Index   : New Index
                    :                  : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT        :          604.16  :      15.49  :       5.09
STRING SORT         :          39.521  :      17.66  :       2.73
BITFIELD            :      2.0827e+08  :      35.73  :       7.46
FP EMULATION        :          78.058  :      37.46  :       8.64
FOURIER             :           12771  :      14.52  :       8.16
ASSIGNMENT          :          11.089  :      42.20  :      10.94
IDEA                :          1356.2  :      20.74  :       6.16
HUFFMAN             :          862.41  :      23.91  :       7.64
NEURAL NET          :          13.701  :      22.01  :       9.26
LU DECOMPOSITION    :          599.28  :      31.05  :      22.42
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX       : 25.826
FLOATING-POINT INDEX: 21.489
Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
CPU                 : GenuineIntel Intel(R) Celeron(TM) CPU                1200MHz 1193MHz
L2 Cache            : 256 KB
OS                  : Linux 2.6.17
C compiler          : 4.1.1
libc                :
MEMORY INDEX        : 6.066
INTEGER INDEX       : 6.744
FLOATING-POINT INDEX: 11.919
Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.
Back to top
View user's profile Send private message
red-wolf76
l33t
l33t


Joined: 13 Apr 2005
Posts: 714
Location: Rhein-Main Area

PostPosted: Sat Jun 24, 2006 8:19 pm    Post subject: Re: following the gcc hype Reply with quote

yngwin wrote:
If you update portage to 2.1, then you can use FEATURES="parallel-fetch" (in make.conf) with the same result.

Yay! I didn't know that. It's going in my make.conf right away... Thanks...
_________________
0mFg, G3nt00 r0X0r$ T3h B1g!1111 ;)

Use sane CFLAGS! If for no other reason, do it for the lulz!
Back to top
View user's profile Send private message
bur
Apprentice
Apprentice


Joined: 20 Feb 2004
Posts: 229

PostPosted: Wed Jul 05, 2006 2:01 am    Post subject: Reply with quote

I recompiled all of system and all kde-base packages with 4.1.1 and
cflags:

-O2 -march=athlon-xp -fomit-frame-pointer -pipe -ftree-vectorize

and all works well. Only Firefox quit with a segmentation fault. I had just started recompiling it without tree-vectorize, when I read that not Fx, but zlib has trouble with the vectorizer. So instead of an hour-long emerge mozilla-firefox, I remerged zlib and now really all works well.
Back to top
View user's profile Send private message
radfoj
Guru
Guru


Joined: 31 Dec 2004
Posts: 490
Location: Tísek, Czech Republic

PostPosted: Mon Jul 10, 2006 3:40 pm    Post subject: Reply with quote

Cleaners for unneeded dependencies like "dep -s" or "emerge -pv --depclean" offers me to unmerge virtual/libstdc++ and sys-libs/libstdc++-v3. Arent this packages needen anymore after switch to gcc-4.1?

I am considering it as cleaner mistake. Am I right?
Code:
# equery d libstdc++
[ Searching for packages depending on libstdc++... ]
sys-devel/gcc-4.1.1
Back to top
View user's profile Send private message
Halcy0n
Developer
Developer


Joined: 17 Sep 2003
Posts: 1682
Location: Freehold, NJ

PostPosted: Mon Jul 10, 2006 10:32 pm    Post subject: Reply with quote

radfoj wrote:
Cleaners for unneeded dependencies like "dep -s" or "emerge -pv --depclean" offers me to unmerge virtual/libstdc++ and sys-libs/libstdc++-v3. Arent this packages needen anymore after switch to gcc-4.1?

I am considering it as cleaner mistake. Am I right?
Code:
# equery d libstdc++
[ Searching for packages depending on libstdc++... ]
sys-devel/gcc-4.1.1

You never needed them to begin with, unless you had stuff still compiled against the old libstdc++ (which is fixed by just recompiling it), or if a binary application depended upon their presence. We had them as a dep in the gcc ebuilds because it was the quick fix. Its finally be dropped though, and anything that actually needs them should depend upon the virtual explicitly.
_________________
Mark Loeser
http://www.halcy0n.com
Back to top
View user's profile Send private message
radfoj
Guru
Guru


Joined: 31 Dec 2004
Posts: 490
Location: Tísek, Czech Republic

PostPosted: Mon Jul 10, 2006 10:56 pm    Post subject: Reply with quote

Halcy0n, thanks for explanation. I unmerged them and tried open opera:
Code:
$ opera
/opt/opera/lib/opera/9.0-20060616.5/opera: error while loading shared libraries: libstdc++.so.5: cannot open shared object file: No such file or directory

and this is what you said, that binary packages that needs it, should depend upon the virtual explicitly. Opera ebuild should be corrected. Maybe others (acroread?, openoffice-bin?, vmware? ... and so).
Back to top
View user's profile Send private message
Seelenfeuer
n00b
n00b


Joined: 09 Nov 2005
Posts: 73

PostPosted: Sat Jul 15, 2006 11:58 am    Post subject: Reply with quote

Hi!
I dont know whether it was already mentioned cause im too lazy to read the whole thread, but with gcc-4.1.1 i cant compile ANY version of x11-libs/fox between 1.4.11 and 1.5.2 (havent tried versions prior to 1.4.11). with gcc-3.4.4 it works. i have some custom cflags and ldflags, but i cant remember whether i changed them since i installed gcc-4.1.1... just to mention it ;)
Regards
Seelenfeuer
_________________
Is this your only grace -
The art of conversation?

~Paradise Lost~
Back to top
View user's profile Send private message
kharan5876
n00b
n00b


Joined: 15 Jan 2006
Posts: 66

PostPosted: Wed Jul 19, 2006 8:43 pm    Post subject: Reply with quote

radfoj wrote:
Halcy0n, thanks for explanation. I unmerged them and tried open opera:
Code:
$ opera
/opt/opera/lib/opera/9.0-20060616.5/opera: error while loading shared libraries: libstdc++.so.5: cannot open shared object file: No such file or directory

and this is what you said, that binary packages that needs it, should depend upon the virtual explicitly. Opera ebuild should be corrected. Maybe others (acroread?, openoffice-bin?, vmware? ... and so).


Try emerging lib-compat and running it again.

I had to do this for a different package that was complaining about a missing libstdc++ library
Back to top
View user's profile Send private message
vovin
n00b
n00b


Joined: 07 May 2006
Posts: 16

PostPosted: Sat Aug 12, 2006 4:33 pm    Post subject: Reply with quote

nbench on Celeron M

gcc version 4.1.2 20060811 (prerelease)
Code:

BYTEmark* Native Mode Benchmark ver. 2 (10/95)
Index-split by Andrew D. Balsa (11/97)
Linux/Unix* port by Uwe F. Mayer (12/96,11/97)

TEST                : Iterations/sec.  : Old Index   : New Index
                    :                  : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT        :          792.64  :      20.33  :       6.68
STRING SORT         :          67.133  :      30.00  :       4.64
BITFIELD            :      2.6103e+08  :      44.78  :       9.35
FP EMULATION        :           98.28  :      47.16  :      10.88
FOURIER             :           17242  :      19.61  :      11.01
ASSIGNMENT          :          16.288  :      61.98  :      16.08
IDEA                :          2735.6  :      41.84  :      12.42
HUFFMAN             :          1318.4  :      36.56  :      11.67
NEURAL NET          :           18.71  :      30.06  :      12.64
LU DECOMPOSITION    :           784.8  :      40.66  :      29.36
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX       : 38.353
FLOATING-POINT INDEX: 28.829
Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
CPU                 : GenuineIntel Intel(R) Celeron(R) M CPU        420  @ 1.60GHz 1596MHz
L2 Cache            : 1024 KB
OS                  : Linux 2.6.17-gentoo-r5
C compiler          : 4.1.2
libc                :
MEMORY INDEX        : 8.871
INTEGER INDEX       : 10.131
FLOATING-POINT INDEX: 15.990
Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.


gcc version 3.4.6 (Gentoo 3.4.6-r2, ssp-3.4.6-1.0, pie-8.7.9)
Code:

YTEmark* Native Mode Benchmark ver. 2 (10/95)
Index-split by Andrew D. Balsa (11/97)
Linux/Unix* port by Uwe F. Mayer (12/96,11/97)

TEST                : Iterations/sec.  : Old Index   : New Index
                    :                  : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT        :            1303  :      33.42  :      10.97
STRING SORT         :          67.013  :      29.94  :       4.63
BITFIELD            :      2.4621e+08  :      42.23  :       8.82
FP EMULATION        :          78.977  :      37.90  :       8.74
FOURIER             :           17807  :      20.25  :      11.37
ASSIGNMENT          :          17.429  :      66.32  :      17.20
IDEA                :          2284.1  :      34.93  :      10.37
HUFFMAN             :            1497  :      41.51  :      13.26
NEURAL NET          :           19.46  :      31.26  :      13.15
LU DECOMPOSITION    :           696.2  :      36.07  :      26.04
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX       : 39.649
FLOATING-POINT INDEX: 28.369
Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
CPU                 : GenuineIntel Intel(R) Celeron(R) M CPU        420  @ 1.60GHz 1596MHz
L2 Cache            : 1024 KB
OS                  : Linux 2.6.17-gentoo-r5
C compiler          : 3.4.6
libc                :
MEMORY INDEX        : 8.893
INTEGER INDEX       : 10.718
FLOATING-POINT INDEX: 15.734
Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Portage & Programming All times are GMT
Goto page Previous  1, 2, 3, 4, 5, 6
Page 6 of 6

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum