Re: [ROOT] compiling root_v3 on LinuxPPC

From: Jiri Masik (masik@pc203b.fzu.cz)
Date: Wed Dec 20 2000 - 01:00:38 MET

Next message: Isard_Dunietz: "[ROOT] proper construction of TF1 objects."
Previous message: Fons Rademakers: "[ROOT] Re: TParticlePDG::Charge() problem?"
Maybe in reply to: cstrato@EUnet.at: "[ROOT] compiling root_v3 on LinuxPPC"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Damir Buskulic <buskulic@lapp.in2p3.fr> writes:

> This one is known. Simply compile without -O2. The optimisation is
> broken in some very specific cases with gcc2.95.2 under linux/ppc
[...]
> cstrato@EUnet.at wrote:
> > 
> > g++ -O2 -pipe -Wall -fsigned-char -fPIC -I/usr/X11/include -Iinclude
> > -DHAVE_CONFIG -o eg/src/TAttParticle.o -c eg/src/TAttParticle.cxx
> > {standard input}: Assembler messages:
> > {standard input}:23471: Error: operand out of range (-32828 not between
> > -32768 and 32767)

I prefer to build root with -O2 and run g++
 g++ -pipe -Wall -fsigned-char -fPIC -I/usr/X11/include -Iinclude
 -DHAVE_CONFIG -o eg/src/TAttParticle.o -c eg/src/TAttParticle.cxx

just on the two files which cause the trouble. The difference in speed
is a factor 1.5 for build with/without -O2. test/stress 1000 reports
160 rootmarks on my powerbook G3/400 then. I haven't seen anything
wrong with the optimized build apart from issues discussed in
http://root.cern.ch/root/roottalk/roottalk00/1459.html

        Jiri

Next message: Isard_Dunietz: "[ROOT] proper construction of TF1 objects."
Previous message: Fons Rademakers: "[ROOT] Re: TParticlePDG::Charge() problem?"
Maybe in reply to: cstrato@EUnet.at: "[ROOT] compiling root_v3 on LinuxPPC"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

This archive was generated by hypermail 2b29 : Tue Jan 02 2001 - 11:50:40 MET