Closed GoogleCodeExporter closed 9 years ago
Original comment by gigaherz
on 27 Feb 2009 at 3:30
As far as forceinlines, you may have missed that __forceinline is redefined in
PS2Etypes:
#define __forceinline __attribute__((always_inline,unused))
I can move the stat.h include, though I haven't really had any issues with it.
As far as the latter goes, I think I see the issue. The pcsx2 team, as a whole,
decided to remove x64 support from Linux a while back, and judging from the gcc
output you quoted, you're compiling on a 64 bit platform.
The 64 bit code in the plugins has never been removed, but by the same token
hasn't
been tested on a 64 bit platform since. My reccommendation would be to set up a
32
bit chroot, and compile and run pcsx2 from in there.
This guide may help:
http://www.gentoo.org/proj/en/base/amd64/howtos/index.xml?part=1&chap=2
Though you are probably better off using stage files from here:
http://www.funtoo.org/
Original comment by arcum42@gmail.com
on 27 Feb 2009 at 4:07
Hi arcum,
I resolved the last one (#3) wasn't a 64 bit issue. But rather was
zerospu2/Linux/Linux.cpp I had to move Linux.h near the top. I think the
problem is
mainly differences in versions of gcc as some required weird shuffling. I'm now
stuck
at soundtouch.
Also, if it __forceinline has been redefined, i wonder why my compiler didn't
catch it.
Original comment by rak...@gmail.com
on 27 Feb 2009 at 4:13
You are using the build.sh file, right (sh build.sh all)? With the
rearrangement of
the build system, I actually was having a lot of trouble with SoundTouch, and
ultimately, set the build.sh file to build SoundTouch, then manually copy the
libSoundTouch.a file into the zerospu2 folder.
I'm still looking for a better solution then that, but it seems to work.
And I'm not sure on why your compiler didn't catch that. I'm using gcc 4.3.3.
There is, however, one function that it inlined that won't inline in ZeroGS.
That
might be part of it. I'll fix that at least, and rearrange headers a bit.
Original comment by arcum42@gmail.com
on 27 Feb 2009 at 4:24
Correct, I am using sh build.sh all
I'm using gcc 4.1.2 (4.3.3 is still experimental in portage). Let me set up my
32bit
chroot and i'll get back to you guys.
Original comment by rak...@gmail.com
on 27 Feb 2009 at 5:07
Do you guys notice any difference between gcc 4.3 with -march=core2 and
-msse4.1 ?
ie. will the core 2 duo optimizations and sse4.1 optimizations make a signifcant
difference in performance?
Original comment by rak...@gmail.com
on 27 Feb 2009 at 4:09
Hi, you guys still need to move the headers in zerospu2/Linux/Linux.cpp (mainly
move
Linux.h include to the top)
Also, I'm getting stuck here:
gcc -DPACKAGE_NAME=\"pcsx2\" -DPACKAGE_TARNAME=\"pcsx2\" -DPACKAGE_VERSION=\"0.9.5\"
-DPACKAGE_STRING=\"pcsx2\ 0.9.5\" -DPACKAGE_BUGREPORT=\"zerofrog@gmail.com\"
-DPACKAGE=\"pcsx2\" -DVERSION=\"0.9.5\" -DSVN_REV=\"Revision:\ 627\" -DNDEBUG=1
-DPCSX2_FORCESSE3=1 -DPCSX2_FORCESSE4=1 -DENABLE_NLS=1 -I. -I./.. -I./../../
-I./../../../common/include -I./../../../3rdparty -pipe -msse -O3
-fno-guess-branch-probability -fno-dse -fno-tree-dse -fpermissive -Xlinker
-zmuldefs
-MT ix86_tools.o -MD -MP -MF .deps/ix86_tools.Tpo -c -o ix86_tools.o ix86_tools.cpp
cc1plus: error: unrecognized command line option "-fno-dse"
make[2]: *** [ix86_tools.o] Error 1
make[2]: Leaving directory `/home/pcsx2/pcsx2-read-only/pcsx2/x86/ix86'
make[1]: *** [install-recursive] Error 1
make[1]: Leaving directory `/home/pcsx2/pcsx2-read-only/pcsx2/x86'
make: *** [install-recursive] Error 1
Error with building pcsx2
Original comment by rak...@gmail.com
on 27 Feb 2009 at 8:03
Original issue reported on code.google.com by
rak...@gmail.com
on 27 Feb 2009 at 3:24