This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: gcc 4.2.4 vs llvm-2.3/llvm-gcc-4.2.1
- From: dominiq at lps dot ens dot fr (Dominique Dhumieres)
- To: fortran at gcc dot gnu dot org
- Date: Sat, 21 Jun 2008 00:02:29 +0200
- Subject: Re: gcc 4.2.4 vs llvm-2.3/llvm-gcc-4.2.1
> if any of these flags alone can produce this speed
> reduction between gfortran 4.2.4 and 4.3.1 on the
> induct test.
Try '--param min-vect-loop-bound=2' and you should go from
~90s to ~60s. You can also test '-finline-limit=600' and look
at the times for channel and fatigue. Note that tfft is quite
sensitive to the cache size (probably the same is true for channel).
Darwin9.
On a MacBook Core2Duo 2.16Ghz today 4.4.0 trunk, I get:
================================================================================
Date & Time : 20 Jun 2008 23:01:33
Test Name : pbharness
Compile Command : gfc %n.f90 -m64 -O3 -ffast-math -funroll-loops -ftree-loop-linear -fomit-frame-pointer -finline-limit=600 --param min-vect-loop-bound=2 -o %n
Benchmarks : ac aermod air capacita channel doduc fatigue gas_dyn induct linpk mdbx nf protein rnflow test_fpu tfft
Maximum Times : 300.0
Target Error % : 0.200
Minimum Repeats : 2
Maximum Repeats : 5
Benchmark Compile Executable Ave Run Number Estim
Name (secs) (bytes) (secs) Repeats Err %
--------- ------- ---------- ------- ------- ------
ac 4.52 50712 12.26 5 0.2948
aermod 96.81 1217096 30.41 3 0.1211
air 6.49 73208 8.32 5 0.7620
capacita 3.34 64520 55.34 5 0.2514
channel 2.51 46848 2.28 2 0.0220
doduc 13.85 187688 42.61 2 0.0375
fatigue 5.59 84840 10.51 5 0.3918
gas_dyn 5.94 700392 10.03 5 1.1128
induct 11.40 168872 35.87 2 0.1018
linpk 1.70 42496 28.14 2 0.0124
mdbx 3.58 64768 15.02 2 0.0100
nf 23.92 157288 32.25 2 0.0031
protein 10.54 114096 46.10 2 0.0271
rnflow 10.98 171336 36.92 2 0.0406
test_fpu 10.11 158280 12.63 2 0.1148
tfft 1.31 26392 2.83 2 0.0353
Geometric Mean Execution Time = 17.08 seconds
Dominique