Releases: malu/asymptote
Releases · malu/asymptote
v0.8: Asymptote 0.8
Results vs v0.7: ELO | 79.03 +- 11.43 (95%) SPRT | 10.0+0.1s Threads=1 Hash=8MB LLR | 9.81 (-13.82, 13.82) [0.00, 5.00] Games | N: 2200 W: 909 L: 417 D: 874 https://chess.compactified.space/viewTest/267/ ELO | 90.80 +- 11.00 (95%) SPRT | 60.0+0.6s Threads=1 Hash=64MB LLR | 11.59 (-13.82, 13.82) [0.00, 5.00] Games | N: 2160 W: 871 L: 319 D: 970 https://chess.compactified.space/viewTest/268/
v0.7: Asymptote 0.7
This release is only slightly stronger than the previous 0.6.3 release. The most prominent changes are improved time management and introduction of a singular extension.
v0.6.3
v0.6.2: Asymptote 0.6.2
This release improves scaling to many cores.
v0.6.1: Asymptote 0.6.1
This is just a small bugfix release. I do not expect any gain in playing strength.
v0.6: Asymptote 0.6
This release fixes a serious bug in the calculation of the zobrist hash which was introduced in v0.5. Most of the time, Asymptote was unable to reuse results of previous searches because of this. This release also contains some improvements to almost all aspects of the engine. It improves: * move ordering * evaluation (especially king safety) * search (extensions, reductions, prunings) * small optimizations
v0.5
Asymptote 0.5 There have been 88 commits since the last release. The most notable changes are: * Asymptote now can search using multiple threads. * Asymptote now respects the uci stop command during search (except when doing a benchmark search). * Speed improvements * Complete overhaul of evaluation weights by Texel tuning * Improved king safety evaluation * Improved bishop evaluation * Improved repetition handling
v0.4.2: Asymptote 0.4.2
This release does not change the playing strength. It just slightly changes some output and the names of the resulting released binaries.
v0.4.1: Asymptote 0.4.1
This release fixes a bug in the detection of draws by insufficient material.
Asymptote v0.4
There have been 45 commits since v0.3. Besides code cleanups these commits include the following changes (in no particular order): * king safety considers safe checks * king safety is skipped in the end game * initial rook evaluation * several problems with the pawn promotions * improvements to uci output * reworked history history * evaluation considers pawn mobility * using transposition table in quiescence search * speed optimizations