Linux friendly engines at all levels

[flok]

Embla is primarily developed on linux.
2.0.0 is ~2174 elo

[hgm]

Fairy-Max is fully Linux compliant, and is even available as Debian package or from the repositories of most Linux distros as binary package.

[brtzsnr]

You can pick one of the older versions of Zurichess. Binaries here: http://www.zurichess.xyz/downloads/

The last version is close to 2900, but previous versions are in the desired Elo range.

[abulmo2]

Also, several old engines have been compiled for Linux by Jim Ablett and are available here:

http://kirr.homeunix.org/chess/engines/ ... ENGINES%20++/

[Volodya]

Dear all,

Thank you for all your replies. I was pleasantly surprised to see the response rate! This looks like a very vibrant community indeed.

I am working through your replies and, so far, I believe I only had one issue: I did not have the LLVM gold plugin for DUMB (I am trying to keep the installs as light as possible).

Thanks,

V.

[xr_a_y]

Dear all,

Thank you for all your replies. I was pleasantly surprised to see the response rate! This looks like a very vibrant community indeed.

I am working through your replies and, so far, I believe I only had one issue: I did not have the LLVM gold plugin for DUMB (I am trying to keep the installs as light as possible).

Thanks,

V.

You can try to compile without -flto

[Volodya]

Dear all,

Thank you for all your replies. I was pleasantly surprised to see the response rate! This looks like a very vibrant community indeed.

I am working through your replies and, so far, I believe I only had one issue: I did not have the LLVM gold plugin for DUMB (I am trying to keep the installs as light as possible).

Thanks,

V.

You can try to compile without -flto

Thanks for the suggestion. It indeed works, modulo the deletion of the appropriate commands in the makefile. This is something I did not want to do at first because it feels as an alteration to the engine intended by the developer. Anyway, since it's done, here is the makefile I used for successful compilation (in case this can help anyone):

Code: Select all

#
# Makefile to compile dumb
#

SRC = dumb.d util.d board.d eval.d move.d search.d

DC = ldc2

#lcd2
FFLAGS = -O3 -release -singleobj -w -dw -dip25
DFLAGS = -O0 -g -gc -dip25
PFLAGS = -O1 -w -dw -dip25
CFLAGS = $(PFLAGS) -cov

# executable target
EXE = dumb

pgo:$(SRC)
	$(MAKE) clean
	$(DC) $(FFLAGS) $(SRC)

debug:$(SRC)
	$(DC) $(DFLAGS) $(SRC)

cov:$(SRC)
	$(DC) $(CFLAGS) $(SRC)

clean:
	$(RM) *.o *.profraw *.profdata *.lst

[Volodya]

Might I suggest though that you use something like Stockfish or Komodo at time odds? You want to spend. CPU time on your program, afterall.

Thanks for the suggestions, Evert. May I ask you to precise the above sentences? I did not quite get what you meant. I do have Stockfish 8 installed, but I have a long way to go (if that is even feasible) before being able to get anything useful from games against such fine engines, even in one of their old versions.

V.

[hgm]

What he means is that you can use a strong engine (say Stockfish) to simulate a much weaker opponent by allowing it to think only a tiny fraction of the time a truly weak engine would have had to think. It would save you nearly half the CPU time during test games if the opponent would move 100 times faster than the engine under test.

[abulmo2]

Thanks for the suggestion. It indeed works, modulo the deletion of the appropriate commands in the makefile. This is something I did not want to do at first because it feels as an alteration to the engine intended by the developer.

You can also use the binary provided in the github repository. Avoiding both flto and pgo will slow down Dumb by about 8% and make it a little bit weaker.