Houdini 5 vs Lastest SF Dev test

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

Tdunbug
Posts: 47
Joined: Mon Mar 07, 2016 12:46 am

Houdini 5 vs Lastest SF Dev test

Post by Tdunbug »

I have completed a 22000 game test between Houdini 5 and latest SF dev.

For those who want to do number crunching and data analysis I have copied a link to the pgn file. If you have problem accessing it please let me know. Otherwise copy and paste link into browser

http://tinyurl.com/hobbbju
JJJ
Posts: 1346
Joined: Sat Apr 19, 2014 1:47 pm

Re: Houdini 5 vs Lastest SF Dev test

Post by JJJ »

Stockfish above H5 by 36 elo with +-3 error bar at 40 secondes and 50 ms increment.

I think others test shows less gap between them at longer time control.

In anycase, it is a very nice score for H5.
arunsoorya1309
Posts: 217
Joined: Tue Nov 03, 2015 4:55 pm
Full name: Arun Soorya

Re: Houdini 5 vs Lastest SF Dev test

Post by arunsoorya1309 »

thanks for the pgn ... why dnt u put a small note on the result .. so that people dont have to download the pgn to know how many games are won/lost.
S.Taylor
Posts: 8514
Joined: Thu Mar 09, 2006 3:25 am
Location: Jerusalem Israel

Re: Houdini 5 vs Lastest SF Dev test

Post by S.Taylor »

Tdunbug wrote:I have completed a 22000 game test between Houdini 5 and latest SF dev.

For those who want to do number crunching and data analysis I have copied a link to the pgn file. If you have problem accessing it please let me know. Otherwise copy and paste link into browser

http://tinyurl.com/hobbbju
Very interesting to see what turns up after 22000 games.
Just one thing, I only hope they are long time control games and on very fast hardware. Then we will really give them a good chance to show their full potential and we can study their long term plans and strategies and to see how they pull them off. 22000 games? ok, brilliant.
Tdunbug
Posts: 47
Joined: Mon Mar 07, 2016 12:46 am

Re: Houdini 5 vs Lastest SF Dev test

Post by Tdunbug »

arunsoorya1309 wrote:thanks for the pgn ... why dnt u put a small note on the result .. so that people dont have to download the pgn to know how many games are won/lost.
1 Stockfish 121116 64 POPCNT +800 1 12493.5/22644
2 Houdini 5 Pro x64-popc +800 0 10150.5/22644

with an ELO diff of 37
arunsoorya1309
Posts: 217
Joined: Tue Nov 03, 2015 4:55 pm
Full name: Arun Soorya

Re: Houdini 5 vs Lastest SF Dev test

Post by arunsoorya1309 »

thanks a lot .... any plans to run a 20 core match
Tdunbug
Posts: 47
Joined: Mon Mar 07, 2016 12:46 am

Re: Houdini 5 vs Lastest SF Dev test

Post by Tdunbug »

running a small 1k match @ 3 min+1 tc
barbasg
Posts: 10
Joined: Sun Jun 08, 2014 11:55 am

Re: Houdini 5 vs Lastest SF Dev test

Post by barbasg »

nice terence ,very nice
Tdunbug
Posts: 47
Joined: Mon Mar 07, 2016 12:46 am

Re: Houdini 5 vs Lastest SF Dev test

Post by Tdunbug »

I decided to run a 5min +2 sec match of 1000+ games
Hardware:

dual 10 core(20 cores HT disabled) Intel 2670 v2 @ 2.7ghz
512 mb ram each
each using the same private book

at about the quarter point now

XEON-SERVER, Blitz 5m+2s 163 games


1 Houdini 5 Pro x64-popc +31 +26/=71/-16 54.42% 61.5/113
2 Stockfish 121116 64 POP -31 +16/=71/-26 45.58% 51.5/113
beram
Posts: 1187
Joined: Wed Jan 06, 2010 3:11 pm

Re: Houdini 5 vs Lastest SF Dev test

Post by beram »

Tdunbug wrote:I decided to run a 5min +2 sec match of 1000+ games
Hardware:

dual 10 core(20 cores HT disabled) Intel 2670 v2 @ 2.7ghz
512 mb ram each
each using the same private book

at about the quarter point now

XEON-SERVER, Blitz 5m+2s 163 games


1 Houdini 5 Pro x64-popc +31 +26/=71/-16 54.42% 61.5/113
2 Stockfish 121116 64 POP -31 +16/=71/-26 45.58% 51.5/113
Did a test myself with privat book 25 lines on i5 4200m @2500Mhz 2 cores, PB off SF 21116 wins 57,5% (from http://chess.ultimaiq.net/stockfish.html)

Code: Select all

SF211116 - H5, Blitz 3m+2s 
                                    
1   Stockfish 211116 64 BMI2   +53  +22/=71/-7 57.50%   57.5/100
2   Houdini 5 x64-pext         -53  +7/=71/-22 42.50%   42.5/100