September 2007 - PPL-devel - BUGSENG Mailman

Re: [PPL-devel] Handling of floating point numbers changed from GMP 4.1.4 to GMP 4.2.1
by Roberto Bagnara 12 Sep '07

12 Sep '07

Vincent Lefevre wrote: > On 2007-09-12 10:42:48 +0200, Roberto Bagnara wrote: >> The results are different depending on whether it is linked to GMP 4.1.4 >> or with GMP 4.2.1. For example, on my Fedora 7, x86_64 machine where I >> have GMP 4.1.4 in /usr/lib64 and GMP 4.2.1 in /usr/local/lib, I use the >> script >> >> g++ -I/usr/local/include -o bug.o -c bug.cc >> g++ -static -o bug-gmp-4.2.1 bug.o -L/usr/local/lib -lm -lgmpxx -lgmp >> g++ -static -o bug-gmp-4.1.4 bug.o -L/usr/lib64 -lm -lgmpxx -lgmp >> ./bug-gmp-4.2.1 >> ./bug-gmp-4.1.4 >> >> This prints >> >> real_coeff = 137171200400403985/1125899906842624 >> float = 121.83249999999999602 >> real_coeff = 137171200400403985/1125899906842624 >> float = 121.83250000000001023 >> >> The mismatch does not happen if one does not touch the rounding mode >> or if one sets it to FE_TONEAREST, FE_TOWARDZERO, FE_DOWNWARD. >> So, on this example, while GMP 4.1.4 results do not depend on the >> rounding mode, with GMP 4.2.1 the rounding mode matters. > > Were they compiled with the same compiler and the same compile switches? > If not, perhaps one uses SSE2 while the other one uses the traditional > FPU. Same compiler and same switches as per the script above. -- Prof. Roberto Bagnara Computer Science Group Department of Mathematics, University of Parma, Italy http://www.cs.unipr.it/~bagnara/ mailto:bagnara@cs.unipr.it

2 1

ppl/tests/Box Makefile.am bug1.cc
by Roberto Bagnara 12 Sep '07

12 Sep '07

CVSROOT: /cvs/ppl Module name: ppl Changes by: bagnara(a)cs.unipr.it 2007-09-12 17:49:01 Modified files: tests/Box : Makefile.am Added files: tests/Box : bug1.cc Log message: Test program exhibiting a bug somewhere. Patches: http://www.cs.unipr.it/cgi-bin/cvsweb.cgi/ppl/tests/Box/bug1.cc?cvsroot=ppl… http://www.cs.unipr.it/cgi-bin/cvsweb.cgi/ppl/tests/Box/Makefile.am.diff?cv…

1 0

[Fwd: Re: Handling of floating point numbers changed from GMP 4.1.4 to GMP 4.2.1]
by Roberto Bagnara 12 Sep '07

12 Sep '07

-------- Original Message -------- Subject: Re: Handling of floating point numbers changed from GMP 4.1.4 to GMP 4.2.1 Date: Wed, 12 Sep 2007 16:55:04 +0200 From: Vincent Lefevre <vincent(a)vinc17.org> To: gmp-discuss(a)swox.com References: <46E7A688.3030305(a)cs.unipr.it> On 2007-09-12 10:42:48 +0200, Roberto Bagnara wrote: > The results are different depending on whether it is linked to GMP 4.1.4 > or with GMP 4.2.1. For example, on my Fedora 7, x86_64 machine where I > have GMP 4.1.4 in /usr/lib64 and GMP 4.2.1 in /usr/local/lib, I use the > script > > g++ -I/usr/local/include -o bug.o -c bug.cc > g++ -static -o bug-gmp-4.2.1 bug.o -L/usr/local/lib -lm -lgmpxx -lgmp > g++ -static -o bug-gmp-4.1.4 bug.o -L/usr/lib64 -lm -lgmpxx -lgmp > ./bug-gmp-4.2.1 > ./bug-gmp-4.1.4 > > This prints > > real_coeff = 137171200400403985/1125899906842624 > float = 121.83249999999999602 > real_coeff = 137171200400403985/1125899906842624 > float = 121.83250000000001023 > > The mismatch does not happen if one does not touch the rounding mode > or if one sets it to FE_TONEAREST, FE_TOWARDZERO, FE_DOWNWARD. > So, on this example, while GMP 4.1.4 results do not depend on the > rounding mode, with GMP 4.2.1 the rounding mode matters. Were they compiled with the same compiler and the same compile switches? If not, perhaps one uses SSE2 while the other one uses the traditional FPU. -- Vincent Lefèvre <vincent(a)vinc17.org> - Web: <http://www.vinc17.org/> 100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/> Work: CR INRIA - computer arithmetic / Arenaire project (LIP, ENS-Lyon) _______________________________________________ gmp-discuss mailing list gmp-discuss(a)swox.com http://swox.com/mailman/listinfo/gmp-discuss -- Prof. Roberto Bagnara Computer Science Group Department of Mathematics, University of Parma, Italy http://www.cs.unipr.it/~bagnara/ mailto:bagnara@cs.unipr.it

1 0

[Fwd: Handling of floating point numbers changed from GMP 4.1.4 to GMP 4.2.1]
by Roberto Bagnara 12 Sep '07

12 Sep '07

-------- Original Message -------- Subject: Handling of floating point numbers changed from GMP 4.1.4 to GMP 4.2.1 Date: Wed, 12 Sep 2007 14:03:56 +0200 From: Torbjorn Granlund <tg(a)swox.com> To: gmp-discuss(a)swox.com ./bug-gmp-4.2.1 ./bug-gmp-4.1.4 This prints real_coeff = 137171200400403985/1125899906842624 float = 121.83249999999999602 real_coeff = 137171200400403985/1125899906842624 float = 121.83250000000001023 OK, so it seems GMP 4.1.4 did not comply to the documentation (or your compiler uses the wrong fp->unsigned instruction, not uncommon!). I don't recall this specifically, but I do recall that we rewrote much of this code, for speed and perhaps for correctness. IIRC, C is defined to always do truncation for float -> int assignments, so no matter the rounding mode, intvar = 1.999999; should put 1 in intvar. The mismatch does not happen if one does not touch the rounding mode or if one sets it to FE_TONEAREST, FE_TOWARDZERO, FE_DOWNWARD. So, on this example, while GMP 4.1.4 results do not depend on the rounding mode, with GMP 4.2.1 the rounding mode matters. I think you've got it backwards. (I tested this with GMP 4.2.2 on my x86_64 system and see no problems. I have not tried GMP 4.1.4, but if your output is not mixed up, it is GMP 4.1.4 that misbehaves for you.) Perhaps the answer is simply that GMP provides no guarantee when invoked with the rounding mode set to anything different from FE_TONEAREST, but I have not found that in the documentation. Instead, the documentation says, e.g., that double mpq_get_d (mpq_t op) Convert op to a double, truncating if necessary (ie.: rounding towards zero). Well, we don't provide a guarantee, but we do have very strong feelings about complying to our own documentation. :-) -- Torbjörn _______________________________________________ gmp-discuss mailing list gmp-discuss(a)swox.com http://swox.com/mailman/listinfo/gmp-discuss -- Prof. Roberto Bagnara Computer Science Group Department of Mathematics, University of Parma, Italy http://www.cs.unipr.it/~bagnara/ mailto:bagnara@cs.unipr.it

1 0

Handling of floating point numbers changed from GMP 4.1.4 to GMP 4.2.1
by Roberto Bagnara 12 Sep '07

12 Sep '07

When we installed GMP 4.2.1 on some machines, the regression testing procedure we use for the Parma Polyhedra Library started signaling a problem. It boils down to the following test program: #include <gmpxx.h> #include <iostream> #include <cstdio> #include <fenv.h> int main() { // Rounding upward. fesetround(FE_UPWARD); mpz_class num("137171200400403985"); mpz_class den("1125899906842624"); mpq_class rational(num, den); rational.canonicalize(); std::cout << "real_coeff = " << rational << "\n"; double floating = rational.get_d(); printf("float = %.20g\n", floating); return 0; } The results are different depending on whether it is linked to GMP 4.1.4 or with GMP 4.2.1. For example, on my Fedora 7, x86_64 machine where I have GMP 4.1.4 in /usr/lib64 and GMP 4.2.1 in /usr/local/lib, I use the script g++ -I/usr/local/include -o bug.o -c bug.cc g++ -static -o bug-gmp-4.2.1 bug.o -L/usr/local/lib -lm -lgmpxx -lgmp g++ -static -o bug-gmp-4.1.4 bug.o -L/usr/lib64 -lm -lgmpxx -lgmp ./bug-gmp-4.2.1 ./bug-gmp-4.1.4 This prints real_coeff = 137171200400403985/1125899906842624 float = 121.83249999999999602 real_coeff = 137171200400403985/1125899906842624 float = 121.83250000000001023 The mismatch does not happen if one does not touch the rounding mode or if one sets it to FE_TONEAREST, FE_TOWARDZERO, FE_DOWNWARD. So, on this example, while GMP 4.1.4 results do not depend on the rounding mode, with GMP 4.2.1 the rounding mode matters. Perhaps the answer is simply that GMP provides no guarantee when invoked with the rounding mode set to anything different from FE_TONEAREST, but I have not found that in the documentation. Instead, the documentation says, e.g., that double mpq_get_d (mpq_t op) Convert op to a double, truncating if necessary (ie.: rounding towards zero). We are uncertain about what to do: for our computations we need to set the rounding mode; for efficiency we would like to avoid setting it back and forth; and, on the other hand, we now have code that behaves differently depending on the version of GMP it is linked to. Any advice? All the best, Roberto -- Prof. Roberto Bagnara Computer Science Group Department of Mathematics, University of Parma, Italy http://www.cs.unipr.it/~bagnara/ mailto:bagnara@cs.unipr.it

1 0

ppl ./configure.ac Watchdog/configure.ac
by Roberto Bagnara 12 Sep '07

12 Sep '07

CVSROOT: /cvs/ppl Module name: ppl Changes by: bagnara(a)cs.unipr.it 2007-09-12 09:49:20 Modified files: . : configure.ac Watchdog : configure.ac Log message: Do not allow AC_PROG_CC and AC_PROG_CXX to affect CFLAGS and CXXFLAGS. Patches: http://www.cs.unipr.it/cgi-bin/cvsweb.cgi/ppl/configure.ac.diff?cvsroot=ppl… http://www.cs.unipr.it/cgi-bin/cvsweb.cgi/ppl/Watchdog/configure.ac.diff?cv…

1 0

ppl/src BD_Shape.templates.hh MIP_Problem.cc O ...
by Roberto Bagnara 12 Sep '07

12 Sep '07

CVSROOT: /cvs/ppl Module name: ppl Changes by: bagnara(a)cs.unipr.it 2007-09-12 00:04:16 Modified files: src : BD_Shape.templates.hh MIP_Problem.cc Octagonal_Shape.templates.hh Log message: Several temporary Coefficient variables are now allocated via TEMP_INTEGER. Patches: http://www.cs.unipr.it/cgi-bin/cvsweb.cgi/ppl/src/BD_Shape.templates.hh.dif… http://www.cs.unipr.it/cgi-bin/cvsweb.cgi/ppl/src/MIP_Problem.cc.diff?cvsro… http://www.cs.unipr.it/cgi-bin/cvsweb.cgi/ppl/src/Octagonal_Shape.templates…

1 0

ppl/src globals.cc
by Roberto Bagnara 11 Sep '07

11 Sep '07

CVSROOT: /cvs/ppl Module name: ppl Changes by: bagnara(a)cs.unipr.it 2007-09-11 23:50:04 Modified files: src : globals.cc Log message: Typo fixed. Patches: http://www.cs.unipr.it/cgi-bin/cvsweb.cgi/ppl/src/globals.cc.diff?cvsroot=p…

1 0

ppl/src BD_Shape.templates.hh Box.inlines.hh B ...
by Roberto Bagnara 11 Sep '07

11 Sep '07

1 0

ppl/src BD_Shape.templates.hh Octagonal_Shape. ...
by Roberto Bagnara 11 Sep '07

11 Sep '07

CVSROOT: /cvs/ppl Module name: ppl Changes by: bagnara(a)cs.unipr.it 2007-09-11 20:46:23 Modified files: src : BD_Shape.templates.hh Octagonal_Shape.templates.hh Log message: Useless initializations removed. For temporaries, use DIRTY_TEMP instead of local variables. Code formatting fixes. Patches: http://www.cs.unipr.it/cgi-bin/cvsweb.cgi/ppl/src/BD_Shape.templates.hh.dif… http://www.cs.unipr.it/cgi-bin/cvsweb.cgi/ppl/src/Octagonal_Shape.templates…

1 0