Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Reducing Rybka 1.0 Size

Author: Marc D

Date: 08:41:29 12/06/05

Go up one level in this thread


On December 06, 2005 at 11:05:10, F. Huber wrote:

>On December 06, 2005 at 10:33:14, Marc D wrote:
>
>>Hi
>>
>>I wonder if reducing Rybkas exe size with upx would result in strength loss
>>or higher nodes per second.
>>
>>I have shrinked the size down to 751 kb from original over 2,5 mb.
>>
>>So far it seems to work without problems but i have to do further testing.
>>
>>Besides that ,i don't know if it is allowed to compress the exe.
>>If anyone knows this matter please inform me.
>>
>>Best
>>Marc
>
>Hello Marc,
>
>all three answers to your questions are NO, i.e. there´s of course neither
>any strength loss nor any speedup (nps), because such a (UPX-)compressed
>EXE-file is simply de-compressed immediately when it´s started. After this
>de-compression the ´full´ engine is executed in memory, so there´s absolutely
>no difference while running (and of course de-compression only needs fractions
>of a second).
>Also I couldn´t think of any problem with compressing a program - otherwise
>even making ZIP-files would be illegal. ;-)
>
>But back again to compression with UPX:
>Why would you like to do this? Because of the 2.5 MB size? Is this really a
>problem nowadays with 100-300 GB harddisks?
>
>And there might be another problem with such UPX-files - I don´t know if it´s
>the case with the Rybka engine (probably not), but I remember that there has
>been a _big_ problem with an earlier UPX-compressed Arena-version:
>The original (uncompressed) version needed about 8-9MB RAM after the program
>has been started, but the UPX-compressed version used about 24-25MB! :-(
>It took some time until Martin Blume found out, that the problem actually was
>this UPX-compression - this de-compression seems to reserve RAM for all
>program resources whether being actually used or not (later while running).
>So he didn´t use UPX any longer for Arena 1.1.
>
>As I said, I´m not sure if this would be the case for Rybka (you could test
>it with any RAM-manager), but I really don´t see any advantages.
>Conclusion (in German): laß besser die Finger davon! ;-)
>
>Regards,
>Franz.


thanks franz.

i didn't think of the fact that the exe gets decompressed when running.


regards
marc



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.