Chiron 3 is about 60 points stronger than Chiron 2.
Now also available on Android:
Here’s a short list of changes:
- improved search (better parameters for pruning and move ordering)
- improved evaluation (in particular mobility and king safety)
- tuning of the evaluation
- new parallel search
- added support for Syzygybases
- added option “Tablebases Probe Depth” (default=6, lower it if you have a SSD)
- 6 pieces Scorpio bitbases can now be probed
- improved time management
- fixed a time management bug for X moves in Y minutes time controls (thanks to Gabor Szots)
- added option “Blockage Detection” to enable/disable pawn blockage detection
- added again support for CTG books
It comes with an improved Polyglot opening book developed by Arturo Ochoa.
Who has bought Chiron 2 in the last month will receive Chiron 3 for free.
Past Customers can buy it with a discount through the link received in an email.
Chiron has been chosen by Ubisoft for the chess mini game in its highly anticipated open world action game Watch Dogs. The player controls Aiden Pearce, a grey hat hacker obsessed with security and surveillance, and can play throughout the Chicago area normal chess games or try to solve puzzles to increase Aiden’s focus skill.
Of course, given the wide audience, a highly modified reduced strength version of Chiron was used.
Here are a couple of chess related screenshots from the game.
Arturo Ochoa has made and kindly offered an opening book for Chiron 2 customers.
It is a wide balanced Polyglot book (~76MB) made from a selection of human and computer games from 2013. You should have received an email with a link to download it.
My thanks to Arturo and my best wishes for Christmas!
Chiron 2 is about 40 Elo points stronger than Chiron 1.5 thanks to improvements in both search and evaluation functions.
Parallel search and time management have been improved too.
Several ChironBitbases have been integrated in the executable:
KBKP, KBPK, KNKP, KNPK, KPKP, KPPK, KPPKP, KQKB, KQKN, KQKP, KQKQ, KQKR, KRKB, KRKN, KRKP, KRKR
as well as 4 tablebases: KPK, KRK, KQK and KPKP with rammed pawns.
Besides that, some bugs have been fixed:
- Chiron would often crash with multi-socket systems
- Gaviota tablebases were rarely probed
I would like to thank Martin Thoresen and Ingo Bauer that have helped with testing.
Salvo Spitaleri released a new high quality book for Chiron 1.5 updated with the latest opening theory. You can freely download it here. My thanks to Salvo.
My best wishes for a Happy New Year!
Chiron 1.5 is about 30 Elo points stronger than Chiron 1.1a thanks to improvements in both search and evaluation functions.
Some ChironBitbases (.cbb) are included: KPK, KPPK, KPKP, KQKP, KPPKP.
The min split depth for parallel search can now be chosen automatically by setting the UCI option “Minimum Split Depth” to 0 or by setting “minsplitdepth” in chiron.ini to “auto”. It is already so by default.
Besides that, some minor bugs have been fixed:
- pawn blockage code was not handling correctly en-passant captures (thanks to R. Vida)
- hash table memory allocation was not working correctly when more than 2GB was set
Chiron has won the Computer Chess Tournament 14. The tournament, a 7 rounds Swiss, was organized by Peter Skinner and played online on February 25-26 on the Internet Chess Club. Time control was 50’+3″. There were 18 participants. Chiron finished first, one full point ahead of Komodo and Pandix. Thanks to Gary Belton for the book preparation and for his powerful i7 3930k Chiron ran on.
Chiron 1.1a is a bugfix version. Unfortunately there were still some bugs that needed to be fixed. Therefore, there are neither strength improvements nor new features.
Once again, thanks to all the users that reported bugs and helped with testing.
In particular these bugs have been fixed:
- tablebases mating problem
- parallel search was not working properly on multi-cpu systems
- pawn blockage detection function could make Chiron crash or score incorrectly some positions
- detection of number of cpu/cores was still wrong with some Intel cpu
- wrong pv output on fail highs
I take this opportunity to wish you a Merry Christmas and a Happy New Year!
Chiron 1.1 is a bugfix version. There are neither strength improvements nor new features.
Thanks to all the users that reported bugs and helped with testing.
In particular these bugs have been fixed:
- crashes when probing Nalimov tablebases and Shawul bitbases during parallel search
- incorrect parsing of UCI options regarding tablebases
- incorrect behaviour with UCI fixed depth search and infinite analysis
- disabling of the book after the first game with Chessbase GUIs
- wrong mate announcement with Winboard protocol when Chiron mates with black
- resigning after 6 moves with Winboard protocol when playing with Arena
- wrong castling with some Chess960 positions that lead to blunders and/or crashes
- wrong detection of the number of cpu and cores with Intel processors