Re: GNAT compiler switches and optimization

[Martin Krischik <krischik@users.sourceforge.net>]

tkrauss wrote:

> Unfortunately the Ada code takes 3-4 times as long.

Well, I collected the various suggestions made and created three test
programs [1,2,3] where one is indeed faster then the next.


GNAT 4.1.1

./Linux-x86_64-Release/Test_Array_1 800
 1.99202E+02 1.96854E+02
 1.88844E+02 1.84498E+02
Time:          18.615262000

./Linux-x86_64-Release/Test_Array_2 800
 1.99202E+02 1.96854E+02
 1.88844E+02 1.84498E+02
Time:           5.457141000

./Linux-x86_64-Release/Test_Array_3 800
 1.99202E+02 1.96854E+02
 1.88844E+02 1.84498E+02
Time:           4.509063000

GNAT GPL 2006 (20060522-34)

./Linux-x86_64-Release/Test_Array_1 800
 1.99202E+02 1.96854E+02
 1.88844E+02 1.84498E+02
Time:          13.950607000

./Linux-x86_64-Release/Test_Array_2 800
 1.99202E+02 1.96854E+02
 1.88844E+02 1.84498E+02
Time:           8.010260000

./Linux-x86_64-Release/Test_Array_3 800
 1.99202E+02 1.96854E+02
 1.88844E+02 1.84498E+02
Time:           7.101502000

As options I used my trusted "Release" option which you can see here [4].

Martin

[1]http://svn.sourceforge.net/viewvc/wikibook-ada/trunk/demos/Source/test_array_1.adb?view=markup
[2]http://svn.sourceforge.net/viewvc/wikibook-ada/trunk/demos/Source/test_array_2.adb?view=markup
[3]http://svn.sourceforge.net/viewvc/wikibook-ada/trunk/demos/Source/test_array_3.adb?view=markup
[4]http://svn.sourceforge.net/viewvc/wikibook-ada/trunk/demos/GNAT/wikibook_ada.gpr?view=markup

-- 
mailto://krischik@users.sourceforge.net
Ada programming at: http://ada.krischik.com