Thus the regulation equation needs to be exponential/logarithmic, not linear. Openings: Because meanwhile the high draw-rates made it impossible to measure Elo-progress in regression-tests, here my UHOV38mvsbig+090+119 openings are used (part of my Anti Draw openings download).
A setting of 100 should stop the hits completely or reduce it to just an occasional hit. GUI: Cutechess-cli (GUI ends game, when a 5-piece endgame is on the board) Tablebases: None for engines, 5 Syzygy for cutechess-cli. So the probe depth setting needs to offer more user control at the upper end of the 0-100 range. User needs to (or may wish to) adapt/experiment with this setting to discover what's best for a particular computer system or analysis position. Even with a fairly modern computer, it can get overwhelmed.īut that's not really the point. The slowdown with Stockfish 5 (recent development versions) is *not* minimal when analyzing certain chess positions for several hours or days in infinite analysis mode, which is the only way I use these chess engines. I don't like to be argumentative (keep going in circles), but will try to state it a bit differently. The Syzygy 5piece tablebase occupies 1 gigabyte of hard disc space, the 6piece tablebase 149 gigabytes, and the 7piece tablebase 18 terabytes. Their size increases rapidly with the number of pieces.
#5 PIECE SYZYGY ENDGAME TABLEBASES FREE#
It needs to be much more exponential(!), not so linear. Tablebases for 3, 4, 5, 6 and 7 pieces can be downloaded free on the Internet. Ronald de Man (or somebody) needs to make Stockfish's syzygy probe depth scaling similar to Komodo. The tablebase is able to tell you whether your position is a win, loss. In other words they're about a power of 2 apart. Syzygy endgame tablebase is a massive database that contains pre-calculated exhaustive analysis of chess endgame positions. Of course the default setting of 1 works OK - gives high hits in early endgames, and occasional hits in opening positions.Īt present a syzygy probe depth setting of 100 for Stockfish 5 is approximately the same as a setting of 10 for Komodo 8. It becomes impossible for users to tweak that setting to find the "sweet spot" for a particular situation on the chessboard. In simple endgames the number of hits grows too large and kn/s speed suffers, even using a SSD. It's reduced somewhat, apparently in linear fashion, but not enough. Unfortunately it doesn't work that way with Stockfish 5 (latest development versions) at the high end of the scale!? Setting probe depth to 99 doesn't cut off or reduce the tablebase hits significantly in typical early endgame positions. The default setting of 2 gives high hits in early endgames, and occasional hits even in opening positions. A setting of about 10 starts giving fairly high hits in this situation with Komodo. I must reduce the setting to about 25 to start seeing significant tablebase hits in typical early endgames. Setting probe depth to 99 results in no tablebase hits whatsoever in my simple tests with early endgame positions. I believe that with correct scaling, setting probe depth to the maximum (99-100) should cut off tablebase access completely, or slow it down to just a very few hits. That's almost the same range but they behave very differently. Komodo 8 has a probe depth scale of 1-99.
#5 PIECE SYZYGY ENDGAME TABLEBASES FULL#
See the COPYING file for the full license text.I wasn't aware of a problem with Stockfish until I tried some other engines with the 6-piece syzygy tablebases.įor example Stockfish 5 (development versions) has a syzygy probe depth scale of 1-100. Lila-tablebase is licensed under the GNU Affero General Public License 3.0 (or any later version at your
Thanks to Bojun Guo for generating and sharing 7-piece tables. Thanks to Ronald de Man for his Syzygy endgame tables. GET /atomic GET /antichess Acknowledgements Claims draw by 50-move rule as soon as possible and ends the mainline.Įrror 404 if not a tablebase position or required tables not present.