On 06/12/15 9:35 PM, Alex wrote:
Ok, lets conclude this post for now. Did some comparison runs with the
original C++ code. Missed this at the beginning.
The results so far are as following:
Here is the original code in C++.
http://benchmarksgame.alioth.debian.org/u64q/program.php?test=binarytrees&lang=gpp&id=6
With modifications to be able to run it on my mac os machine this
results in the code available here:
http://pastebin.com/G5cYESdX
compilation done with
g++ -c -pipe -O3 -fomit-frame-pointer -march=native -fopenmp main.cpp -o
main.o && g++ main.o -o main.run -fopenmp -lboost_system
Here you can find the last version of my D code:
http://dpaste.dzfl.pl/8c8ab00699b5
compilation done with
dmd -release -O -boundscheck=off -inline main.d
time comparison, just with
time ./main
yields
for C++
real 0m8.255s
user 0m7.342s
sys 0m0.905s
for D
real 0m35.356s
user 0m35.149s
sys 0m0.154s
so the overall factor is round about 5.
Thanks for commenting to everyone! If anybody has further ideas - all of
them would be appreciated :)
The original site is not interested in any further languages to be
tested, so my experiment ends here for now...
Why is TreeNode not final?
Also yours does not use threads in any way.
If you cannot add multithreading on D, remove it from c++ before
comparing. They are not equal.
