While waiting for major updates from major strong engines, I utilized the idle time to test the latest Stockfish released last February 26. This is just to probe whether there are some major regressions and see if the pattern of the result correspond with their ranking.
The test was done with Arena 3.5 GUI using 2 cores and 128 MB of hash. The computer used was an AMD quad core with 4GB of RAM which is the main computer utilized in the rating list production. Instead of a self-test, I choose the two major rivals Houdini and Komodo to be the sparring partners in the tournament. Time control was 1 minute base + 1 seconds increment and each will face each other 100 times in a round robin.
Stockfish 14022619 Test Tournament - 2 cores 128MB hash
|1||Stockfish_14022619_x64||119.5/200||· ·· ·· ··||34-16-50||38-17-45||10770.25|
|2||Houdini 4 Pro x64||100.0/200||16-34-50||· ·· ·· ··||35-17-48||9649.00|
|3||Komodo TCEC x64||80.5/200||17-38-45||17-35-48||· ·· ·· ··||8820.25|
300 games played / Tournament finished
Tournament start: 2014.02.28, 00:58:57
Latest update: 2014.02.28, 13:37:10
Level: Blitz 1/1
Hardware: AMD Phenom(tm) II X4 945 Processor with 1.8 GB Memory
Operating system: Windows 7 Ultimate Professional Service Pack 1 (Build 7601) 64 bit
Table created with: Arena 3.5
The result showed that Stockfish 14022619 won the test tournament with 119.5/200, 19.5 points margin higher than the second placer Houdini 4 Pro. Komodo was in its usual rank at number 3. Comparing the performance of the latest Stockfish against the one that was in the rating list a week earlier, it showed no improvement and no perceptible regression. The minor patches in a weeks time has no particular effect in the strength of Stockfish 14022619.
There seemed to be slow and minor improvements in the past weeks. It may have been caused by the disagreement over the manner of implementing the patches. When individual contributors have differing styles and temperaments, conflicts may occur. It is particularly hard for the leader to appease everyone and eventually forking may occur. That would be the likely outcome if one contributor wants that his ideas will always be implemented irregardless of the policies. It's alright, let it be. It will make somebody happy.
Download the test tournament games here.