Amoeba 2.5

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

abulmo2
Posts: 492
Joined: Fri Dec 16, 2016 11:04 am
Location: France
Full name: Richard Delorme

Amoeba 2.5

Post by abulmo2 »

I am pleased to announce a new version of Amoeba :
https://github.com/abulmo/amoeba/releases/tag/v2.5
It should be significantly stronger than version 2.4 (I hope more than 50 elo).

Under Windows, it comes in 3 flavors:
- amoeba-win64.exe
- amoeba-win64-nopopcount.exe
- amoeba-win32.exe

The first one is 25% faster (~20 elo stronger) than the second one, and 78% faster (~50 elo stronger) than the last one. Amoeba indeed depends heavily on a fast popcount instruction.

Under Linux, I did not build a 32 bit version. On the same machine, a Linux version is probably 20 % faster than a Windows version (10% because of the OS, & 10% because of the compiler).

I will add binaries for Mac OS/X at the end of the week.
Richard Delorme
User avatar
Werner
Posts: 3037
Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle

Re: Amoeba 2.5

Post by Werner »

Thanks Richard for the new engine.
Werner
User avatar
Graham Banks
Posts: 45747
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Amoeba 2.5

Post by Graham Banks »

Thanks Richard. Nice gain. :)
gbanksnz at gmail.com
abulmo2
Posts: 492
Joined: Fri Dec 16, 2016 11:04 am
Location: France
Full name: Richard Delorme

Re: Amoeba 2.5

Post by abulmo2 »

abulmo2 wrote:It should be significantly stronger than version 2.4 (I hope more than 50 elo).
Well, we have got another example of self play at very fast control that does not translate well when playing against a gauntlet of opponents at longer time controls. Version 2.5 should be only 20 elo stronger than 2.4.
Richard Delorme
User avatar
Werner
Posts: 3037
Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle

Re: Amoeba 2.5

Post by Werner »

Hi Richard,
we are just testing your engine for CEGT 40/20.
After 360 games of 1000 we have around +50 :-)
....but the start rating is often too high and will probably fall.

regards
Werner