Perft & speed

[Grzegorz Sidorowicz]

I was comparing my perft results to Sjeng 11 and when I did it on
PIV 2GHz I couldn't find nothing to worry...

Initial position, depth 6:
Sjeng 37.93 s
My engine 36.77 s

...but when I did it on Athlon 1700+ results made me feel gloomy

Sjeng 27.03 s
My engine 41.56 s

After all I don't know what I am doing wrong
What kind of operations are working faster on Athlon??
Had you similar problem?

[Roman Hartmann]

Hi,
sounds to me like a compiler issue. You might want tro try some other compiler switches in order to improve the performance of your engine on the AMD as well.

Roman

[Grzegorz Sidorowicz]

Yes it is probably related with compiler, but change compiler it is bad
solution for me. I'm rather interested in discovery causes in my code.
I have my own habits in programming and difference in increase speed between my new engine and Armageddon are on the same level on each processor. However Armageddon is a little faster on Athlon.
I have some suspicions related with places where my move generators put moves (table, structure, kind of table, kind of structure), but I think it is not only this reason. My programs have been written in C++. Who now maybe this fact have influence on compiler. I haven't idea