> Aborted
> 
> That's it. Any ideas what it means and how I should compile the povray?

	You might try configuring povray so as to include debug information
and run it under a debugger:

./configure COMPILED_BY=you --enable-debug
make
gdb ./unix/povray
run test.pov options
bt

	and report the result.

	- NC

Post a reply to this message

> gdb ./unix/povray
> run test.pov options
> bt
>
>  and report the result.

Thanks for advice, here the output below:

<... the same output as in previous message...>
Program received signal SIGABRT, Aborted.
0xa000000000010641 in ?? ()
(gdb) bt
#0  0xa000000000010641 in ?? ()
#1  0x2000000000635190 in raise () from /lib/tls/libc.so.6.1
#2  0x2000000000637a60 in abort () from /lib/tls/libc.so.6.1
#3  0x20000000000580e0 in (anonymous namespace)::_default_terminate_handler
()
   from /local/programs/icc/9.1.045/lib/libcxa.so.6
#4  0x2000000000057aa0 in std::terminate () from
/local/programs/icc/9.1.045/lib/libcxa.so.6
#5  0x2000000000059530 in Terminate??unw () from
/local/programs/icc/9.1.045/lib/libcxa.so.6
#6  0x2000000000059c70 in __gxx_personality_v0??unw () from
/local/programs/icc/9.1.045/lib/libcxa.so.6
#7  0x20000000000599b0 in __intel_personality_routine??unw () from
/local/programs/icc/9.1.045/lib/libcxa.so.6
#8  0x20000000005c4140 in _Unwind_RaiseException () from /lib/libgcc_s.so.1
#9  0x20000000000594d0 in __cxa_throw??unw () from
/local/programs/icc/9.1.045/lib/libcxa.so.6
#10 0x400000000033ca10 in _ZN12POVMS_ObjectC9ERS_ (this=0xfffffffffffffff1,
source=@0x60000fffffff8820) at povmscpp.cpp:473
#11 0x4000000000340ea0 in POVMS_Object::operator() (this=0x60000fffffff8820)
at povmscpp.h:207
#12 0x4000000000340e40 in POVMS_MessageReceiver::ReceiveHandler () at
povmscpp.cpp:1031
#13 0x2000000000637a60 in abort () from /lib/tls/libc.so.6.1
(gdb)

And for idb also the same:

<... the same output as in previous message...>
Thread received signal ABRT
stopped at [0xa000000000010641]
(idb) where
>0  0xa000000000010641
#1  0x2000000000635190 in raise(...) in /lib/tls/libc.so.6.1
#2  0x2000000000637a60 in abort(...) in /lib/tls/libc.so.6.1
#3  0x20000000000580e0 in
_ZN37_GLOBAL__N__13_exception_cpp_a3e89b6426_default_terminate_handlerEv(...)
in /local/programs/icc/9.1.045/lib/libcxa.so.6
#4  0x2000000000057aa0 in _ZSt9terminatev(...) in
/local/programs/icc/9.1.045/lib/libcxa.so.6
#5  0x2000000000059530 in Terminate(...) in
/local/programs/icc/9.1.045/lib/libcxa.so.6
#6  0x2000000000059c70 in __gxx_personality_v0(...) in
/local/programs/icc/9.1.045/lib/libcxa.so.6
#7  0x20000000000599b0 in __intel_personality_routine(...) in
/local/programs/icc/9.1.045/lib/libcxa.so.6
#8  0x20000000005c4140 in _Unwind_RaiseException(...) in /lib/libgcc_s.so.1
#9  0x20000000000594d0 in __cxa_throw(...) in
/local/programs/icc/9.1.045/lib/libcxa.so.6
#10 0x400000000033ca10 in
((POVMS_Object*)0x7fffffff00000000)->POVMS_Object::POVMS_Object(source=&
class POVMS_Object { ... }) "povmscpp.cpp":473
#11 0x4000000000340ea0 in
((POVMS_Object*)0x7fffffff00000000)->POVMS_Object::operator()()
"povmscpp.cpp"
#12 0x4000000000340e40 in ((POVMS_Object*)<bad
value>)->POVMS_Object::operator()() "povmscpp.cpp":1031
#13 0x2000000000637a60 in abort(...) in /lib/tls/libc.so.6.1
(idb)

Post a reply to this message

> Thanks for advice, here the output below:
> And for idb also the same:

	This looks like the compiler is generating bad code.  Did you try
compiling with gcc and/or without optimizations?

	- NC

Post a reply to this message

Nicolas Calimet <pov### [at] freefr> wrote:
> > Thanks for advice, here the output below:
> > And for idb also the same:
>
>  This looks like the compiler is generating bad code.  Did you try
> compiling with gcc and/or without optimizations?
>
>  - NC

Yes, I tried all options and cases I could imagine. GCC works fine, either
with optimizations or without it. However ICC doesn't whatever option or
switches combination I used. I suppose the problem is in the libcxa intel
library. Still trying to investigate this hypothesis.

BTW, when the multithreading version of the POV-Ray will be availabe for
UNIX machines like mine?

Post a reply to this message

> I suppose the problem is in the libcxa intel
> library. Still trying to investigate this hypothesis.

Yes! Finally I solved the problem. It was really a problem of
incompatibility between gcc and intel libraries. So, to obtain working
binaries the
LDFLAGS="-cxxlib-icc -lcxa" should be supplied for the configure.

Result is pretty good, look at gcc vs icc timings:

Total Scene Processing Times
  Parse Time:    0 hours  0 minutes  0 seconds (0 seconds)
  Photon Time:   0 hours  0 minutes  4 seconds (4 seconds)
  Render Time:   0 hours  1 minutes 46 seconds (106 seconds)
  Total Time:    0 hours  1 minutes 50 seconds (110 seconds)

Total Scene Processing Times
  Parse Time:    0 hours  0 minutes  1 seconds (1 seconds)
  Photon Time:   0 hours  0 minutes  4 seconds (4 seconds)
  Render Time:   0 hours  1 minutes 13 seconds (73 seconds)
  Total Time:    0 hours  1 minutes 18 seconds (78 seconds)

Almost 1.5x speedup!

Post a reply to this message

RedAndr <Red### [at] gmailcom> wrote:
> Almost 1.5x speedup!

  Which version of gcc? Did you compile a 32-bit or a 64-bit binary with it?

-- 
                                                          - Warp

Post a reply to this message

> Which version of gcc?
gcc (GCC) 3.3.3 (SuSE Linux)

> Did you compile a 32-bit or a 64-bit binary with it?
Only 64-bit. It wouldn't even possible to compile 32-bit binaries on
Itanium.

Post a reply to this message

RedAndr <Red### [at] gmailcom> wrote:
> > Which version of gcc?
> gcc (GCC) 3.3.3 (SuSE Linux)

  I think gcc4 has better support for 64-bit code.

-- 
                                                          - Warp

Post a reply to this message

>   I think gcc4 has better support for 64-bit code.
Ok, thanks. I will try it, however I am not sure it can compete with Intel
compiler anyway.

Post a reply to this message

> Yes! Finally I solved the problem. It was really a problem of
> incompatibility between gcc and intel libraries. So, to obtain working
> binaries the
> LDFLAGS="-cxxlib-icc -lcxa" should be supplied for the configure.

	Since you're using rather old GCC libraries together with the latest-or-so
ICC compiler, it is worth trying to install a GCC 4.1 or even 4.2 (as suggested
by Warp for speed reasons) and see whether this library incompatibility remains
or not.

	- NC

Post a reply to this message