[6265] in www-talk@info.cern.ch
Linux-Activists - GCC Channel digest. 94-9-20-8:31
daemon@ATHENA.MIT.EDU (Linux Activists)
Thu Oct 20 12:48:22 1994
Date: Thu, 20 Oct 1994 17:44:35 +0100
Errors-To: postmaster@www0.cern.ch
Errors-To: postmaster@www0.cern.ch
Reply-To: linux-activists@niksula.hut.fi
From: "Linux Activists" <linux-activists@niksula.hut.fi>
To: Multiple recipients of list <www-talk@www0.cern.ch>
Topics:
ELF timing
Re: ELF patches for XFree86 3.1
100% ELF
----------------------------------------------------------------------
From: Kayvan Sylvan <kayvan@Sylvan.COM>
Subject: ELF timing
Date: Wed, 19 Oct 94 17:57 PDT
>>>>> "HJL" == H J Lu <hjl@nynexst.com> writes:
>> The new ELF binaries run *much* slower than the a.out binaries, based
>> on my preliminary tests. Anyone else see stuff like this?
HJL> Just a question: Do you have the ELF library in core all the time?
HJL> Please compile your login shell in the shared ELF format first.
I couldn't recompile bash at all (I sent you the bug report), so I
compiled zsh-2.3.1 in both a.out and ELF formats. Each shell then runs
the script which runs the simple g++ program (this way, I can be sure
that the appropriate library is in core).
I used the same compile options on all the various builds.
Here's what happens:
kayvan@satyr$ time /home/src/zsh.elf repeat.zsh 250 Vec.ELF
33.01user 50.45system 1:33.17elapsed 89%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+0minor)pagefaults 0swaps
kayvan@satyr$ time /home/src/zsh.a.out repeat.zsh 250 Vec.a.out
3.38user 33.24system 0:40.11elapsed 91%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+0minor)pagefaults 0swaps
kayvan@satyr[~/src/g++/bjarne]:532$
The ELF binaries are still slower. :-(
---Kayvan
Kayvan Sylvan | Sylvan Associates | Proud Dad of:
kayvan@Sylvan.COM | Training, Consulting | Katherine Yelena (8/8/89)
PGP Key available. | NLP Master Practitioner | Robin Gregory (2/28/92)
"The trust and respect of a child is an honor to be earned, not demanded."
------------------------------
From: riepe@ifwsn4.ifw.uni-hannover.de (Michael Riepe)
Subject: Re: ELF patches for XFree86 3.1
Date: Thu, 20 Oct 94 03:06:01 +0100
> Is there any chance you've put the ELF XFree libraries up for ftp anywhere?
> I don't have enough space to build them. :(
I have uploaded the patches and shared libraries to our local FTP
server. You can get them from
ftp://ifwsn4.ifw.uni-hannover.de/private/linuxelf/
Maybe someone in the US can mirror this, it's less than 1MB...
So long,
Michael.
--
Michael Riepe <riepe@ifwsn4.ifw.uni-hannover.de>
Universit"at Hannover
Institut f"ur Fertigungstechnik und Spanende Werkzeugmaschinen
Schlosswender Str. 5 30159 Hannover (Germany)
... beware the storm that gathers here ("The Prophet's Song")
------------------------------
From: hjl@nynexst.com (H.J. Lu)
Subject: 100% ELF
Date: Thu, 20 Oct 94 0:42:06 EDT
>
> From: adam@adam.yggdrasil.com (Adam J. Richter)
> Subject: Ready for 100% ELF?
> Date: Tue, 18 Oct 1994 18:52:00 +0200
>
>
>
> Question: is there anything left in Linux that requires non-ELF
> binaries. In particular, do we still need non-ELF binaries to build:
>
> (a) the kernel, and
Yes. Kernel has to be compiled in a.out. I think it is the time to
add ELF support for kernel compilation.
> (b) dynamicly loadable libraries for the Andrew system.
Are you saying "loadable"? Eric is working on dlopen et al. If they
are not written in C++ or not using dlclose (), they will be supported
when we have dl stuff. We are still looking for a good solution for
the shared C++ libraries to work with dlopen/dlclose.
>
> I know that andrew can be built under SCO Unix, so I would
> assume that there already is an ELF or COFF version of the andrew
> dynamically loadable libraries somewhere in the andrew source tree.
> I'm just wondering if people have actually done either of these
> things yet.
>
H.J.
------------------------------
End of GCC Digest
*****************
-------
