SSDF

Comments

to the Swedish Rating List 2/2019, July 8

On this July ratinglist of 2019, we can present eight new entrants.

First out is the new leader in the rating list, and the first to break the 3500- barrier in our rating list. It's Stockfish 10 by Marco Costalba, Tord Romstad and Joona Kiiski. As with the former versions, we have used the free opening book from Sedat Canbaz for the testing, and this time it has been the "Aggressive 4.0 by Fauzi.abk". After 577 games it now leads the list by a margin of 36 points in front of its older version, and with the rating of 3529!

Second out and still early on in the testing, is the latest program by Mark Lefler and Larry Kaufman. It has placed itself third in this list with the rating of 3482. This is 28 points stronger than the last Komodo-version we have tested. As in the testing of the 12.3-version, we have used Erdogan Guenes opening book. This version is the "normal" version which can be compared to the older versions we have tested.

We have also tested the Monte Carlo Tree Search-version of Komodo, named MCTS, for the first time. We have used the same opening book for this testing. Even though it's also early on in the testing, it has achieved a strong rating of 3424 after 205 games played. It trails the normal 13.02-version by 58 points, but are stronger than most other programs outside the latest Stockfish/Komodo range and places itself on the eight spot in the ratinglist.

Last one out of the programs which we have tested on our 1800X-platform, is John Stanback's - Wasp 3.5 program. It has achieved a rating of 3235 after the first 480 games played, and this is 22 points stronger than Wasp 3. Contrary to the other two Wasp-versions we have tested (2.01 and 3), Wasp 3.5 now includes an own opening book, which we have used for the testing.

As can be seen from the rating list, we now also have enough games to be able to present four new additions to the Q6600 hardware level. First one is Stockfish 9, which has gotten a rating of 3445 after 431 games played. This is 48 points lower than the version we have tested on the AMD Ryzen 7 1800X hardware level, but is also the strongest program we have ever tested on the Q6600 platform!

With a slim margin of only 2 points lower than Stockfish 9, we have now the Komodo 12.3 on our Q6600 hardware. It has managed to get a rating of 3443 after the first 280 games. This is only 11 points lower than the same version on the 1800X and 110 points in front of the 11.01 version on the Q6600!

Next up is the program of Jon Dart - Arasan 21.2, which we now have 152 games with. It's also an early bird which barely made it to this list, but its rating after the first couple of matches are 3250. This rating is 30 points lower than the same version on our 1800X platform, but a whopping 249 points stronger than the 17.2 version which we also tested on our Q6600 level!

Last one out and extremely in time for this list, was Wasp 3.5 on our Q6600. It has achieved a rating of 3174 after the first 112 games, which is 61 points behind the 1800X version, and 109 points stronger than the last Wasp-version (2.01) we have tested on the Q6600.

Notable gains/losses from the last list was: Arasan 21.2 1800X, which gained 11 points, Booot 6.3.1 Q6600 gained 8 points and Booot 6.3.1 1800X gained 5 points. The programs which lost most points was: Mephisto London 68000 which lost 23 points, The Baron 3.43 1800X lost 7 points and Stockfish 8 Q6600/Wasp 3 1800X lost 5 points each.

For the moment we are manually testing both the Millenium ChessGenius Exclusive, the Millenium The King Exclusive and also the previously untested oldie Mephisto London 68020. We will be able to introduce the ChessGenius Exclusive on our next rating list which we plan to have out in the end of summer or as soon as this computer have enough games.

Lars Sandin