Re: Is there an Ada compiler whose Ada.Numerics.Generic_Elementary_Functions.Log(Base=>10, X=>variable) is efficient?

["John B. Matthews" <nospam@nospam.invalid>]

In article 
<alpine.LNX.2.00.1002151339400.17315@Bluewhite64.example.net>,
 Colin Paul Gloster <Colin_Paul_Gloster@ACM.org> wrote:

> Thanks for running it. The Ada program for timing had 500 statements 
> in the body of the loop. I reproduced only the first and last 
> verbatim: I showed a  bash  (Bourne Again SHell) line for producing 
> all 500 statements in a comment in-between the the first statement 
> and the last statement. You ran a program with 498 of the statements 
> missing.

Ah, source level compression!

> Anyway, the answer produced by the program is not so much of concern 
> as the relative speeds of different implementations. Did  g++  
> produce a faster result for you than GNAT? It did for me for many 
> versions of GCC today on a different platform than I used in the 
> beginning of this thread...

Retesting both 500+ line programs produces results similar to yours:

$ make test
Darwin: gcc 4.3.4
rm -f *.o *.ali b~* core
rm -f *.s temp.txt logada logcpp 
gnatmake logada -cargs -O3 -gnatnp -bargs -shared -largs -ffast-math 
-dead_strip
gcc -c -O3 -gnatnp logada.adb
gnatbind -shared -x logada.ali
gnatlink logada.ali -shared-libgcc -ffast-math -dead_strip
g++ -O3 logcpp.cc -o logcpp
time ./logcpp
6.35785e+08        0.57 real         0.56 user         0.00 sys
time ./logada
 6.35785378608776E+08
       28.15 real        28.10 user         0.02 sys

-- 
John B. Matthews
trashgod at gmail dot com
<http://sites.google.com/site/drjohnbmatthews>