Symbolic vs tscp: more match results

[sje]

Symbolic vs tscp: more match results

The 1,000 game matches played at 1 min plus 1 sec earlier seem to proven crash/freeze/leak resistance; I've now switched to running matches at 5 minutes per game and with the tablebase files on an SSD. The idea is to better approximate tournament conditions.

In a 100 game match at 5 minutes/game time control played yesterday, Symbolic scored 77-11-12 (+275 elo). The PGN is here: https://dl.dropboxusercontent.com/u/316 ... 927.pgn.gz

[sje]

Symbolic got a few minor changes and then played another 100 game match against tscp using a 5 minute/game time control. The result was 78-8-14 (+301 elo).

In many of the games, tscp had the better position out of the opening book, but eventually Symbolic's transposition capability and SSD-resident tablebase files forced tscp to snatch defeat from the jaws of victory.

The gzip PGN file is here: https://dl.dropboxusercontent.com/u/316 ... 928.pgn.gz

[hgm]

With only 8% losses the results with 100 games become virtually random; you won't he able to determine anymore how much better Symbolic is. You should switch to a stronger opponent. Fairy-Max 4.8V and after that KingSlayer/Simple might be good next steps. The latter will be more stable, (Fairy-Max always finishes the iteration, which makes it prone to time losses on promotions and such).

[sje]

I agree in that it's time to move to a stronger opponent, preferably one who can score 50% to 75% vs Symbolic. Do you have a link to the latest Unix compatible source version of FairyMax?

I could also use my CookieCat or my Myopic, as both are open source although Myopic isn't very strong.

[hgm]

From http://hgm.nubati.net/cgi-bin/gitweb.cgi you can always download the latest 'snapshot' of all my open-source projects as a tar ball. The Fairy-Max sources include a Makefile, and you can build it with "make install". A slightly stronger engine is in the simple.git section there; this project is so fresh that I have not included a Makefile yet. But you can compile with

gcc -O2 simple.c -o simple

[sje]

From http://hgm.nubati.net/cgi-bin/gitweb.cgi you can always download the latest 'snapshot' of all my open-source projects as a tar ball. The Fairy-Max sources include a Makefile, and you can build it with "make install". A slightly stronger engine is in the simple.git section there; this project is so fresh that I have not included a Makefile yet. But you can compile with

gcc -O2 simple.c -o simple

The make install didn't work very well: https://dl.dropboxusercontent.com/u/316 ... nstallBugs

[sje]

One problem with the Makefile is that "cp -u <src> <dest>" is Linux specific; there is no -u option on other Unixes like OpenBSD (Mac OS/X).

[hgm]

Oh, I did not know that. I just copied it from some other Makefile. I looked it up on the man page, and all the -u does is suppress the copying if the destination file already exists and is newer. Which would not be very useful.

I will remove the -u.

[sje]

There's also:

Code: Select all

ld: warning: option -s is obsolete and being ignored

[hgm]

OK, I pushed a rejuvenated version of the Makefile, which obeys the GNU standard better. (It installs now in the /usr/local sub-tree by default, rather than /usr). The -u is gone. This one should work without problems.