[Bug fortran/43463] large compile time regressions for Polyhedron 2005 benchmarks

["howarth at nitro dot med dot uc dot edu" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #3 from howarth at nitro dot med dot uc dot edu  2010-03-21 18:40 -------
Using built-in specs.
Target: x86_64-apple-darwin10.3.0
Configured with: ../gcc-4.4.4/configure --prefix=/sw --prefix=/sw/lib/gcc4.4
--mandir=/sw/share/man --infodir=/sw/share/info
--enable-languages=c,c++,fortran,objc,java --with-gmp=/sw
--with-libiconv-prefix=/sw --with-ppl=/sw --with-cloog=/sw --with-system-zlib
--x-includes=/usr/X11R6/include --x-libraries=/usr/X11R6/lib
--disable-libjava-multilib --enable-checking=release
Thread model: posix
gcc version 4.4.4 20100320 (prerelease) (GCC) 

shows...

================================================================================
Date & Time     : 21 Mar 2010 13:31:33
Test Name       : gfortran_lin_O3
Compile Command : gfortran -ffast-math -funroll-loops -msse3 -O3 %n.f90 -o %n
Benchmarks      : ac aermod air capacita channel doduc fatigue gas_dyn induct
linpk mdbx nf protein rnflow test_fpu tfft
Maximum Times   :     2000.0
Target Error %  :      0.100
Minimum Repeats :    10
Maximum Repeats :   100

   Benchmark   Compile  Executable   Ave Run  Number   Estim
        Name    (secs)     (bytes)    (secs) Repeats   Err %
   ---------   -------  ----------   ------- -------  ------
          ac      0.79       10000     11.24      10  0.0063
      aermod     36.50       10000     21.49      10  0.0064
         air      2.10       10000      5.70      10  0.0564
    capacita      1.34       10000     35.27      10  0.0140
     channel      0.74       10000      1.75      11  0.0928
       doduc      4.32       10000     27.50      10  0.0118
     fatigue      1.64       10000      8.46      10  0.0541
     gas_dyn      2.39       10000      5.08      15  0.0971
      induct      3.51       10000     12.81      10  0.0122
       linpk      0.72       10000     15.43      10  0.0618
        mdbx      1.40       10000     11.28      10  0.0471
          nf      3.01       10000     29.38      17  0.0977
     protein      3.38       10000     35.30      10  0.0102
      rnflow      4.44       10000     23.42      10  0.0160
    test_fpu      4.13       10000      8.74      10  0.0189
        tfft      0.45       10000      1.90      12  0.0253

Geometric Mean Execution Time =      11.47 seconds

================================================================================

and for gcc trunk...

Using built-in specs.
COLLECT_GCC=gcc-4
COLLECT_LTO_WRAPPER=/sw/lib/gcc4.5/libexec/gcc/x86_64-apple-darwin10.3.0/4.5.0/lto-wrapper
Target: x86_64-apple-darwin10.3.0
Configured with: ../gcc-4.5-20100320/configure --prefix=/sw
--prefix=/sw/lib/gcc4.5 --mandir=/sw/share/man --infodir=/sw/share/info
--enable-languages=c,c++,fortran,objc,obj-c++,java --with-gmp=/sw
--with-libiconv-prefix=/sw --with-ppl=/sw --with-cloog=/sw --with-mpc=/sw
--with-system-zlib --x-includes=/usr/X11R6/include --x-libraries=/usr/X11R6/lib
--disable-libjava-multilib --enable-checking=release
Thread model: posix
gcc version 4.5.0 20100321 (experimental) (GCC) 

================================================================================
Date & Time     : 21 Mar 2010 12:28:57
Test Name       : gfortran_lin_O3
Compile Command : gfortran -ffast-math -funroll-loops -msse3 -O3 %n.f90 -o %n
Benchmarks      : ac aermod air capacita channel doduc fatigue gas_dyn induct
linpk mdbx nf protein rnflow test_fpu tfft
Maximum Times   :     2000.0
Target Error %  :      0.100
Minimum Repeats :    10
Maximum Repeats :   100

   Benchmark   Compile  Executable   Ave Run  Number   Estim
        Name    (secs)     (bytes)    (secs) Repeats   Err %
   ---------   -------  ----------   ------- -------  ------
          ac      0.90       10000     10.60      10  0.0706
      aermod     39.17       10000     20.96      10  0.0060
         air      2.26       10000      5.68      12  0.0254
    capacita      1.71       10000     33.17      10  0.0097
     channel      0.56       10000      1.83      10  0.0326
       doduc      4.59       10000     27.75      10  0.0040
     fatigue      1.54       10000      8.07      10  0.0647
     gas_dyn      2.85       10000      4.39      16  0.0850
      induct      3.64       10000     12.65      10  0.0056
       linpk      0.71       10000     15.39      10  0.0233
        mdbx      1.51       10000     11.33      10  0.0061
          nf      1.89       10000     27.56      48  0.0935
     protein      3.47       10000     37.03      10  0.0112
      rnflow      5.16       10000     23.67      10  0.0166
    test_fpu      4.18       10000      8.67      10  0.0148
        tfft      0.49       10000      1.88      10  0.0389

Geometric Mean Execution Time =      11.24 seconds

================================================================================

which reduces the compile time regressions down to...

benchmark           gcc45 compile time/gcc44 compile time
ac                                    1.14
aermod                          1.07
air                                   1.08
capacita                         1.28
channel                          0.76
doduc                             1.06
fatigue                            0.94
gas_dyn                          1.19
induct                             1.04
linpk                                0.99
mdbx                              1.08
nf                                     1.35
protein                            1.03
rnflow                             1.16
test_fpu                          1.01
tfft                                   1.09

So the bulk of the compile time regression is due to not using release
checking.
This increased compile time penalty for not using release checking should be
mentioned
in the release notes (since it is a significant change from prior releases). 


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43463