New engine releases 2019

[Guenther]

You can still get dev builds from https://ci.appveyor.com/project/christi ... no/history by clicking the desired build, selecting the configuration, and then selecting artifacts. The failed builds still seem to have executables, with a version number one version behind.

Thanks for the info Tony! I didn't realize he had appveyor set for dev builds.
It seems the binary vesions are one off since one version just had a changed readme file.
(appveyor history version 5.22 remained version 5.21 and so on)

[jorose]

Winter 0.7 has been released. Hopefully I will find time to make a proper release post this time.

To avoid any confusion, this Winter version is NN based. All versions of Winter from 0.6.2 onward are NN based. Furthermore there is no support for GPU, so it should be run like any other CPU based engine. For now there is also no external weights file, this may change if the network becomes bigger than it is.

I am hoping this version proves to be around +50 over the previous release. The speed penalty seems to be higher on Windows then on Linux and I didn't test on older CPUs, so I wouldn't be surprised if it ends up being Elo neutral.

Winter also supports contempt now and features an Armageddon setting. Default contempt is 0 for now, but a small contempt of 10 will likely become the default at some point in the future as it seems to have little to no rating cost while strongly reducing the draw rate. Setting the Armageddon setting to true gained roughly 50 Elo in Armageddon mode over the baseline.

I intend to start more actively maintaining the readme.

https://github.com/rosenthj/Winter

[Norbert Raimund Leisner]

KnightX 1.96 http://technochess.free.fr/

[Norbert Raimund Leisner]

Bagatur Chess 1.8a https://github.com/bagaturchess/Bagatur ... ads/Engine

[Gabor Szots]

KnightX 1.96 http://technochess.free.fr/

I have just noticed that KnightX can now be used as a UCI engine as well. It seems it has been so from version 1.94.

[fabianVDW]

I have a new version of FabChess: https://github.com/fabianvdW/FabChess/releases

This version is significantly stronger than the previous version.
Tests against Version1.11 at TC 120s + 1.2s:

Changelog:
- A bug was fixed with the transposition table. This is where most of the elo of this patch comes from. This bug led to blundering wins and draws away pretty often, especially in longer timecontrols.
- Slight tweaks to LMR
- Added PST for Queen and Rooks
Other:
- Evaluation Wiki page was updated accordingly

Future Updates will contain:
- More UCI functionality
- SMP support
- EGTB support
- and of course more playing strength
- when developement slows down, more wiki pages will be added. For now, significant changes are still planned in the structure, which would render the wiki outdated pretty soon

I estimate the strength to be somewhere around 2800 now for both CCRL scales.
Thanks to all the testers having tested previous versions of FabChess <3

[Gabor Szots]

ChessBrainVB 3.74 TCEC

https://github.com/RZulu54/ChessBrainVB/releases

[voffka]

Igel 1.9.1 at https://github.com/vshcherbyna/igel/releases/tag/1.9.1 (binaries for Raspberry Pi, Linux, macOS and Windows).

What's new:

- improve LMR
- fix time control mode "nodes limit"
- fix problem when a bestmove may become 0 in case tb search failed
- fix bug #96: incorrect evaluation of position 4k3/8/8/8/5n2/1R6/6P1/5K2 w KQkq -
- fix bug #97: incorrect evaluation of king exposed
- expose evaluation of a position by implementing "eval" command option

Strength should be around Winter 0.7 in CCRL 40/4 time control:

Code: Select all

Score of Igel 1.9.1 64 POPCNT vs Winter 0.7 BMI2: 15 - 16 - 21  [0.490] 52

Since Winter 0.7 is not yet tested by CCRL it is hard to tell how strong it will be. I am cautiously optimistic and hope it will be around 3000.

[Guenther]

KnightX 1.96 http://technochess.free.fr/

I have just noticed that KnightX can now be used as a UCI engine as well. It seems it has been so from version 1.94.

Code: Select all

v1.96, September 2019
	positional evaluation improvements
	tree searching improvements
	no more use the Nalimov tablebases (was not working)

I also notice from the readme that it does not use nalimov tbs anymore, 'because they did not work'?
I am quite sure though they were working in the past at least.

[Gabor Szots]

KnightX 1.96 http://technochess.free.fr/

I have just noticed that KnightX can now be used as a UCI engine as well. It seems it has been so from version 1.94.

Code: Select all

v1.96, September 2019
	positional evaluation improvements
	tree searching improvements
	no more use the Nalimov tablebases (was not working)

I also notice from the readme that it does not use nalimov tbs anymore, 'because they did not work'?
I am quite sure though they were working in the past at least.

That is confusing. From the same readme:

Knightx 1.96
Copyright 2019 Christophe Jolly
The tablebases code is by Eugene Nalimov

Also, the sample cfg file still contains commands for setting up the EGTB environment.

Besides, UCI mode seems broken, setting up a 6-piece ending UCI mode stops at depth 7 while WB mode shows depth 59.