New engine releases 2019

[Damir]

I suppose nobody here is dumb enough to use it.....

[Graham Banks]

Bagatur 1.9

https://github.com/bagaturchess/Bagatur ... ads/Engine

[abulmo2]

I suppose nobody here is dumb enough to use it.....

Dumb, but fast
If you want to see high nodes per second on a single score, you should give it a try.

[Guenther]

A new open source chess engine was sent to CCRL in December, but was not announced yet.
It's name is Halogen by Kieren Pearson from Australia.
(still trying to figure out what is the family name?)

A little dilemma for the entry in the chronology, it had 'releases' since June, but was unknown before and in the past
we (Frank, Thomas, Leo, myself) only recorded dates of official first releases or when being found by someone...

Last official version 2.73, but there is already a version 2.82 in the 'binaries' folder

https://github.com/KierenP/Halogen2

[Gabor Szots]

A new open source chess engine was sent to CCRL in December, but was not announced yet.
It's name is Halogen by Kieren Pearson from Australia.
(still trying to figure out what is the family name?)

A little dilemma for the entry in the chronology, it had 'releases' since June, but was unknown before and in the past
we (Frank, Thomas, Leo, myself) only recorded dates of official first releases or when being found by someone...

Last official version 2.73, but there is already a version 2.82 in the 'binaries' folder

https://github.com/KierenP/Halogen2

I'd start with the date of version 2.7.3.
I don't understand your problem with the family name. If the author then Pearson. If the engine then Halogen (not Halogen2, I asked him to clarify that).

[vincent]

Little Wing v0.6.0
https://vinc.cc/blog/2019/12/22/little- ... -released/
https://vinc.cc/binaries/littlewing-v0.6.0/

[Guenther]

A new open source chess engine was sent to CCRL in December, but was not announced yet.
It's name is Halogen by Kieren Pearson from Australia.
(still trying to figure out what is the family name?)

A little dilemma for the entry in the chronology, it had 'releases' since June, but was unknown before and in the past
we (Frank, Thomas, Leo, myself) only recorded dates of official first releases or when being found by someone...

Last official version 2.73, but there is already a version 2.82 in the 'binaries' folder

https://github.com/KierenP/Halogen2

I'd start with the date of version 2.7.3.
I don't understand your problem with the family name. If the author then Pearson. If the engine then Halogen (not Halogen2, I asked him to clarify that).

I did not know if Pearson or Kieren is the family name - Google says both is possible also in Australia ;-)
The rest was known to me as I had read the CCRL forum.

[voffka]

Igel 2.2.0 at https://github.com/vshcherbyna/igel/releases/tag/2.2.0 (binaries for Raspberry Pi, Linux, macOS and Windows). This is a TCEC S17 preparation release that contains important performance improvements as well as various bugfixes.

What's new:

- Fix futility pruning and use it only after at least one legal move has been played
- Fix 32bit overflow in syzygy probing code that caused bugs when probing 7 piece positions
- Fix syncronisation issue that caused "connection stalled" in Linux OS
- Fix overflow bug that prevented using transposition table bigger than 32 GB
- Improve performance in SMP mode and avoid copying the pawn hash table for each worker
- Improve performance in SMP mode when clearing large hash size (more than 64 Gb)
- Improve performance in SMP mode and solve cache coherency issues when calculating node/tbhits count
- Improve performance of tuner and perform multithreaded texel tuning
- Improve performance and use hardware instruction to count leading zeroes (lsb/poplsb)
- Improve performance in SMP mode and start worker threads before the main thread
- Improve performance in SMP mode and skip random depths by worker threads (Laser/Ethereal method)
- Improve performance in Linux OS and use -flto compiler option (around 10% of increase of NPS)

Relative strength against Igel 2.1.0:

Code: Select all

tc=all/10+0.1
hash=256
Score of Igel 2.2.0 64 POPCNT vs Igel 2.1.0 64 POPCNT: 438 - 212 - 533  [0.596] 1183
Elo difference: 67.20 +/- 14.71

Some CCRL 40/4 4CPU testing against Strelka 5.5 x64 (CCRL 40/4 elo 3141)

Code: Select all

Score of Igel 2.1.2 next8 64 POPCNT vs Strelka 5.5 x64: 4 - 2 - 5  [0.591] 11
Elo difference: 63.88 +/- 164.19

P.S. Currently there is only Windows compile available in the package. I will compile the usual packages (macOS, Linux, Raspberry) during the week.

[Gabor Szots]

RuyDos 1.1.11

https://bitbucket.org/alonamaloh/ruydos/downloads/

Let's see if it can play with 6-man syzygy's now (used to crash).

[Gabor Szots]

RuyDos 1.1.11

https://bitbucket.org/alonamaloh/ruydos/downloads/

Let's see if it can play with 6-man syzygy's now (used to crash).

It can't.