Best free chess engines under linux
Submitted by
on Sat, 05/09/2009 at 8:49am.
There are many free chess engines that run on linux, and I decided to test some of them and determine which one is the strongest. It is not as if there are no results available already on the internet, and it is not like I wanted to decide which engine should I try to beat because they are all light years ahead of me. What I would like to know is which engine I can trust the most if I analyse a game, and also I like to kibitz high level games. Plus, it is fun to run a private tournament, especially since I don't have to provide any money for the contestants.
I tested what I think are the strongest chess engines available under linux: Toga II, Spike Turin, Glaurung, and Fruit. I did try crafty, but it is not in the same league as the aforementioned engines.
I set out to do the same a couple of moths ago, but at that time I used linux under a virtual machine, in 32 bit mode, and with not enough memory, so I decided to repeat my little experiment on a real PC. This time I am using a 64-bit version of Ubuntu 9.04, and except for Spike Turin, all of the listed chess engines are available through package manager. Many of the strongest chess engines may not be available on repositories for other linux distributions, but that is not a big problem. They can be downloaded from here (Toga II, Fruit), here (Glaurung), and here (Spike Turin 1.2), and easily installed. Also, since all of these engines except for Spike are UCI compliant, one needs to install polyglot. In addition to that, it is good to install opening books .
Aside from these chess engines being very powerful and free, under linux you also enjoy the benefit of using a 64 bit OS. Otherwise, most of the PCs are sold with 32 bit OS despite the fact that CPUs have been 64 bit for quite some time.
Let's see the results, then. In the first tournament, I set the time control to 5 minutes using a shell script that calls xboard, and each engine plays 40 matches against any other engine. The results are listed in the table below, where comma separated values represent win, loss, and draw (e.g., 16, 10, 14 means 16 wins, 10 losses, and 14 draws).
|
Engine
|
1
|
2
|
3
|
4
|
Score
|
|
1. Toga II 1.4.1 SE
|
x
|
16, 10, 14
|
23, 9, 8
|
25, 4, 11
|
57.5
|
|
2. Spike Turin 1.2
|
10, 16, 14
|
x
|
13, 12, 15
|
9,13,18
|
14.5
|
|
3. Glaurung 2.2
|
9,23, 8
|
12, 13, 15
|
x
|
18, 11, 11
|
9
|
|
4. Fruit 2.1
|
4, 25, 11
|
13, 9, 18
|
11, 18, 11
|
x
|
-4
|
Obviously, Toga II is the strongest engine in blitz games. I did not expect to find a winner with such a large margin, as for example this site has all these engines with the same rating. Even if the numbers listed there are not very accurate, the differences between the engines (in terms of strength) should be small. To see how these engines perform in longer games, I limited the number of games to only 2 for each pair, and set the time control to 60 minutes. While these results are statistically less reliable (small sample size), they are an indicator that these engines are pretty close performance-wise. The results are listed in the next table.
|
Engine
|
1
|
2
|
3
|
4
|
Score
|
|
1. Glaurung
|
x
|
0, 0, 2
|
1, 0, 1
|
1, 0, 1
|
4
|
|
2.Toga II 1.4.1 SE
|
0, 0, 2
|
x
|
1, 1, 0
|
2, 0, 0
|
3
|
|
3. Spike Turin 1.2
|
0, 1, 1
|
1, 1, 0
|
x
|
1, 1, 0
|
-0.5
|
|
4. Fruit 2.1
|
0, 1, 1
|
0, 2, 0
|
1, 1, 0
|
x
|
-2.5
|
In conclusion, you will not make a mistake if you chose any of these engines as they are all very strong and quite close to the strongest commercially available engines. And if the past is any guide, they will only get stronger.