octave-bug-tracker
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Octave-bug-tracker] [bug #37308] JIT build fails some numerical tests;


From: Arun Giridhar
Subject: [Octave-bug-tracker] [bug #37308] JIT build fails some numerical tests; non-JIT build is OK.
Date: Mon, 10 Sep 2012 03:25:51 +0000
User-agent: Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:15.0) Gecko/20100101 Firefox/15.0

URL:
  <http://savannah.gnu.org/bugs/?37308>

                 Summary: JIT build fails some numerical tests; non-JIT build
is OK.
                 Project: GNU Octave
            Submitted by: arungiridhar
            Submitted on: Mon 10 Sep 2012 03:25:51 AM GMT
                Category: Interpreter
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Incorrect Result
                  Status: None
             Assigned to: None
         Originator Name: 
        Originator Email: 
             Open/Closed: Open
         Discussion Lock: Any
                 Release: dev
        Operating System: GNU/Linux

    _______________________________________________________

Details:

Hello,

The 20120909 dev version passes tests with "make check" as expected when
configured with --disable-jit, but fails a few numerical tests with
--enable-jit, keeping everything else the same. The numerical differences are
very small, but sometimes return complex instead of real and consequently
break functions like fzero.

Here is a "diff -c" of the two test/fntests.log files (I have replaced the
"/home/path/to/src" with "$OCTAVESRCDIR", everything else is the same):


*** build-nojit/test/fntests.log        2012-09-09 21:31:01.059421438 -0400
--- build/test/fntests.log      2012-09-09 22:05:36.971477029 -0400
***************
*** 77,82 ****
--- 77,96 ----
  >>>>> processing $OCTAVESRCDIR/octave/libinterp/dldfcn/tsearch.cc
  >>>>> processing $OCTAVESRCDIR/octave/libinterp/interp-core/oct-map.cc
  >>>>> processing $OCTAVESRCDIR/octave/libinterp/interp-core/pt-jit.cc
+   ***** assert (id (1+1i), 1+1i)
+ !!!!! test failed
+ assert (id (1 + 1i),1 + 1i) expected
+  1 + 1i
+ but got
+  1.0000 + 1.0000i
+ values do not match
+ shared variables 
+   scalar structure containing the fields:
+ 
+     id =
+ 
+ @(x) x
+ 
  >>>>> processing $OCTAVESRCDIR/octave/libinterp/interp-core/sparse-xpow.cc
  >>>>> processing $OCTAVESRCDIR/octave/libinterp/interpfcn/data.cc
  >>>>> processing $OCTAVESRCDIR/octave/libinterp/interpfcn/defaults.cc
***************
*** 105,110 ****
--- 119,142 ----
  >>>>> processing $OCTAVESRCDIR/octave/libinterp/octave-value/ov-cx-diag.cc
  >>>>> processing
$OCTAVESRCDIR/octave/libinterp/octave-value/ov-fcn-handle.cc
  >>>>> processing
$OCTAVESRCDIR/octave/libinterp/octave-value/ov-fcn-inline.cc
+   ***** assert (feval (fn, 6), 37)
+ !!!!! test failed
+ assert (feval (fn, 6),37) expected
+  37
+ but got
+   3.7000e+01 + 1.9763e-323i
+ complex != real
+ shared variables   scalar structure containing the fields:
+     fn = f(x) = x.^2 + 1
+   ***** assert (fn (6), 37)
+ !!!!! test failed
+ assert (fn (6),37) expected
+  37
+ but got
+   3.7000e+01 + 1.9763e-323i
+ complex != real
+ shared variables   scalar structure containing the fields:
+     fn = f(x) = x.^2 + 1
  >>>>> processing
$OCTAVESRCDIR/octave/libinterp/octave-value/ov-flt-re-mat.cc
  >>>>> processing $OCTAVESRCDIR/octave/libinterp/octave-value/ov-int16.cc
  >>>>> processing $OCTAVESRCDIR/octave/libinterp/octave-value/ov-int32.cc
***************
*** 353,360 ****
--- 385,402 ----
  >>>>> processing $OCTAVESRCDIR/octave/scripts/optimization/fminbnd.m
  >>>>> processing $OCTAVESRCDIR/octave/scripts/optimization/fminsearch.m
  >>>>> processing $OCTAVESRCDIR/octave/scripts/optimization/fminunc.m
+   ***** assert (fminunc (@(x) x^2, 1, optimset ("FunValCheck", "on")), 0,
eps)
+ !!!!! test failed
+ fminunc: non-real value encountered
  >>>>> processing $OCTAVESRCDIR/octave/scripts/optimization/fsolve.m
  >>>>> processing $OCTAVESRCDIR/octave/scripts/optimization/fzero.m
+   ***** assert (fzero (@(x) x^(1/3) - 1e-8, [0,1], opt0), 1e-24, 1e-22*eps)
+ !!!!! test failed
+ fzero: not a valid initial bracketing
+ shared variables   scalar structure containing the fields:
+     opt0 =
+       scalar structure containing the fields:
+         TolX = 0
  >>>>> processing $OCTAVESRCDIR/octave/scripts/optimization/lsqnonneg.m
  >>>>> processing $OCTAVESRCDIR/octave/scripts/optimization/optimset.m
  >>>>> processing $OCTAVESRCDIR/octave/scripts/optimization/pqpnonneg.m
***************
*** 1107,1110 ****
  $OCTAVESRCDIR/octave/scripts/statistics/tests/z_test.m
  $OCTAVESRCDIR/octave/scripts/statistics/tests/z_test_2.m
  $OCTAVESRCDIR/octave/scripts/testfun/runtests.m
! $OCTAVESRCDIR/octave/build-nojit/scripts/plot/gnuplot_binary.m
--- 1149,1152 ----
  $OCTAVESRCDIR/octave/scripts/statistics/tests/z_test.m
  $OCTAVESRCDIR/octave/scripts/statistics/tests/z_test_2.m
  $OCTAVESRCDIR/octave/scripts/testfun/runtests.m
! $OCTAVESRCDIR/octave/build/scripts/plot/gnuplot_binary.m


Exact configure options for the JIT version (broken up to multiple lines for
readability):

../configure --enable-jit CC="gcc" CXX="g++"
CFLAGS="-march=native -mtune=native -O3 -funroll-loops
    -ftracer -pthread"
CXXFLAGS="-march=native -mtune=native -O3 -funroll-loops
    -ftracer -pthread"
FFLAGS="-O3"
LLVM_CONFIG=`which llvm-config`


Exact configure options for the non-JIT version (broken up to multiple lines
for readability):

../configure --enable-jit=no CC="gcc" CXX="g++"
CFLAGS="-march=native -mtune=native -O3 -funroll-loops
    -ftracer -pthread"
CXXFLAGS="-march=native -mtune=native -O3 -funroll-loops
    -ftracer -pthread"
FFLAGS="-O3"
LLVM_CONFIG=`which llvm-config`


Output of "uname -a":

Linux myname-laptop 3.2.0-30-generic-pae #48-Ubuntu SMP Fri Aug 24 17:14:09
UTC 2012 i686 i686 i386 GNU/Linux


Output of "llvm-config --version":

3.2svn






    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?37308>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/




reply via email to

[Prev in Thread] Current Thread [Next in Thread]