New engine releases 2020

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Guenther
Posts: 3278
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Full name: Guenther Simon
Contact:

Re: New engine releases 2020

Post by Guenther » Mon Jan 13, 2020 11:57 am

ECE 20.1 released yesterday
http://lesgames.altervista.org/
https://rwbc-chess.de/chronology.htm
--------------------------------------------------
The troll explosion at talkchess:
https://docs.google.com/spreadsheets/d/ ... KSptBx9AUs

RubiChess
Posts: 124
Joined: Fri Mar 30, 2018 5:20 am

Re: New engine releases 2020

Post by RubiChess » Mon Jan 13, 2020 1:00 pm

Norbert Raimund Leisner wrote:
Mon Jan 13, 2020 9:08 am
RubiChess 1.6.1 https://github.com/Matthies/RubiChess/releases
v1.6.1.1 exists too but without a Windows binary

MarvinChess 3.5.0 a8 https://github.com/bmdanielsson/marvin-chess/releases (with source code)
I'm sorry for the confusing Version numbering.
1.6.1 was prepared in a hurry to compete in Grahams div1 tornament. It is not good enough for a 1.7 version imo.
After that another patch passed that I wanted to be included in the TCEC Tournament Version. So I added another digit to the version.
TCEC Plays on Linux so no Windows binary is needed.

Everybody who wants to test should use the 1.6.1 that plays at Grahams.

Regards, Andreas

Guenther
Posts: 3278
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Full name: Guenther Simon
Contact:

Re: New engine releases 2020

Post by Guenther » Mon Jan 13, 2020 1:07 pm

Norbert Raimund Leisner wrote:
Mon Jan 13, 2020 9:08 am
RubiChess 1.6.1 https://github.com/Matthies/RubiChess/releases
v1.6.1.1 exists too but without a Windows binary

MarvinChess 3.5.0 a8 https://github.com/bmdanielsson/marvin-chess/releases (with source code)
RubiChess 1.61 binaries seem to be there? Compiled at 2020-01-11.

---

I have compiled two binaries (pop/nopop) for MarvinChess 3.50a8, but I had to make a tiny change in utils.c otherwise
my compiler would complain. So it would be on your own risk, if someone wanna try them.
I can upload them, if at least one user makes a request for it. In that case I would be interested in the
speed diff from cmd for the startposition, or if the pop version runs at all - cannot test it here...

There were a lot of changes since the last version from July 2019, the last few ones were especially for TCEC17 being
able to use up to 512 threads and 128G hash.
I guess though that Martin will create a proper release soon anyway, after the qualification round of TCEC seems already over.

Code: Select all

uci
id name Marvin 3.5.0-a8
id author Martin Danielsson
option name Hash type spin default 32 min 1 max 131072
option name Clear Hash type button
option name OwnBook type check default true
option name Ponder type check default false
option name SyzygyPath type string default <empty>
option name Threads type spin default 1 min 1 max 512
option name LogLevel type spin default 0 min 0 max 2
uciok
ucinewgame
go depth 15
info depth 1 seldepth 0 nodes 1 time 0 nps 0 tbhits 0 hashfull 0 score cp 22 pv b1a3
info depth 1 seldepth 0 nodes 2 time 0 nps 0 tbhits 0 hashfull 0 score cp 75 pv b1c3
info depth 1 seldepth 0 nodes 3 time 0 nps 0 tbhits 0 hashfull 0 score cp 87 pv g1f3
info depth 1 seldepth 0 nodes 13 time 0 nps 0 tbhits 0 hashfull 0 score cp 92 pv e2e3
info depth 2 seldepth 1 nodes 44 time 0 nps 0 tbhits 0 hashfull 0 score cp -9 pv e2e3 d7d5
info depth 2 seldepth 1 nodes 74 time 0 nps 0 tbhits 0 hashfull 0 score cp 0 pv g1f3 g8f6
info depth 3 seldepth 2 nodes 237 time 0 nps 0 tbhits 0 hashfull 0 score cp 87 pv g1f3 d7d5 d2d4
info depth 4 seldepth 3 nodes 771 time 0 nps 0 tbhits 0 hashfull 0 score cp 0 pv g1f3 d7d6 d2d3 g8f6
info depth 5 seldepth 5 nodes 2294 time 10 nps 229000 tbhits 0 hashfull 0 score cp 61 pv g1f3 g8f6 d2d3 d7d5 c1e3
info depth 6 seldepth 7 nodes 5147 time 10 nps 514000 tbhits 0 hashfull 0 score cp 51 upperbound
info depth 6 seldepth 7 nodes 6387 time 20 nps 319000 tbhits 0 hashfull 0 score cp 31 upperbound
info depth 6 seldepth 8 nodes 8381 time 20 nps 419000 tbhits 0 hashfull 0 score cp 0 pv b1c3 b8c6 g1f3 g8f6 d2d4 d7d5
info depth 6 seldepth 9 nodes 10363 time 30 nps 345000 tbhits 0 hashfull 1 score cp 5 pv e2e4 d7d5 b1c3 d5e4 c3e4 g8f6 d2d3 f6e4 d3e4 d8d1 e1d1
info depth 7 seldepth 11 nodes 12267 time 30 nps 408000 tbhits 0 hashfull 1 score cp 17 lowerbound
info depth 7 seldepth 11 nodes 12868 time 30 nps 428000 tbhits 0 hashfull 1 score cp 37 lowerbound
info depth 7 seldepth 12 nodes 15107 time 40 nps 377000 tbhits 0 hashfull 2 score cp 51 pv d2d4 g8f6 g1f3 d7d5 h2h3 h7h6 b1c3
info depth 8 seldepth 12 nodes 25523 time 60 nps 425000 tbhits 0 hashfull 5 score cp 41 upperbound
info depth 8 seldepth 12 nodes 31204 time 70 nps 445000 tbhits 0 hashfull 6 score cp 21 upperbound
info depth 8 seldepth 12 nodes 33147 time 70 nps 473000 tbhits 0 hashfull 6 score cp 0 pv e2e4 d7d5 e4d5 g8f6 d2d4 e7e6 g1f3 e6d5
info depth 9 seldepth 12 nodes 44290 time 90 nps 492000 tbhits 0 hashfull 7 score cp 11 lowerbound
info depth 9 seldepth 12 nodes 45605 time 90 nps 506000 tbhits 0 hashfull 7 score cp 31 lowerbound
info depth 9 seldepth 12 nodes 53680 time 100 nps 536000 tbhits 0 hashfull 8 score cp 49 pv e2e4 d7d5 e4d5 g8f6 d2d4 d8d5 b1c3 d5e6 c1e3 b8c6
info depth 10 seldepth 13 nodes 99733 time 180 nps 554000 tbhits 0 hashfull 15 score cp 39 upperbound
info depth 10 seldepth 15 nodes 153144 time 270 nps 567000 tbhits 0 hashfull 23 score cp 44 pv e2e4 d7d5 e4e5 c7c5 g1f3 b8c6 d2d4 c5d4 c2c3 c8f5 c3d4
info depth 11 seldepth 15 nodes 178044 time 310 nps 574000 tbhits 0 hashfull 26 score cp 44 pv e2e4 d7d5 e4e5 c7c5 g1f3 b8c6 d2d4 c5d4 c2c3 c8f5 c3d4
info depth 11 seldepth 15 nodes 218211 time 380 nps 574000 tbhits 0 hashfull 35 score cp 54 lowerbound
info depth 11 seldepth 15 nodes 233355 time 400 nps 583000 tbhits 0 hashfull 36 score cp 44 pv e2e4 d7d5 e4e5 c7c5 g1f3 b8c6 d2d4 c5d4 c2c3 c8f5 c3d4
info depth 12 seldepth 16 nodes 305918 time 520 nps 588000 tbhits 0 hashfull 47 score cp 44 pv e2e4 d7d5 e4e5 c7c5 f1b5 c8d7 b5d7 b8d7 g1f3 e7e6 e1g1 d5d4 d2d3 g8e7 b1d2
info depth 13 seldepth 16 nodes 527053 time 900 nps 585000 tbhits 0 hashfull 84 score cp 54 lowerbound
info depth 13 seldepth 20 nodes 855779 time 1440 nps 594000 tbhits 0 hashfull 139 score cp 49 pv e2e4 e7e6 d2d4 d7d5 b1c3 g8f6 e4e5 f6e4 g1f3 f8b4 c1d2 b8c6 h2h3 e4c3 d2c3 b4c3 b2c3
info depth 14 seldepth 21 nodes 1518803 time 2550 nps 595000 tbhits 0 hashfull 244 score cp 59 lowerbound
info depth 14 seldepth 21 nodes 1582642 time 2660 nps 594000 tbhits 0 hashfull 252 score cp 59 pv e2e4 e7e6 g1f3 d7d5 e4e5 c7c5 c2c3 b8c6 d2d4 g8h6 c1h6 g7h6 f1e2 c5d4 e1g1 d8b6 c3d4 b6b2
info depth 15 seldepth 22 nodes 2574029 time 4320 nps 595000 tbhits 0 hashfull 385 score cp 59 pv e2e4 e7e6 g1f3 d7d5 e4e5 c7c5 c2c3 b8c6 d2d4 g8h6 c1h6 g7h6 f1e2 c5d4 e1g1 d8b6 c3d4 b6b2
info depth 15 currmove g1f3 currmovenumber 2
info depth 15 currmove b1c3 currmovenumber 3
info depth 15 currmove d2d4 currmovenumber 4
info depth 15 currmove c2c3 currmovenumber 5
info depth 15 currmove e2e3 currmovenumber 6
info depth 15 currmove h2h3 currmovenumber 7
info depth 15 currmove d2d3 currmovenumber 8
info depth 15 currmove c2c4 currmovenumber 9
info depth 15 currmove g1h3 currmovenumber 10
info depth 15 currmove f2f3 currmovenumber 11
info depth 15 currmove f2f4 currmovenumber 12
info depth 15 currmove b1a3 currmovenumber 13
info depth 15 currmove a2a3 currmovenumber 14
info depth 15 currmove b2b4 currmovenumber 15
info depth 15 currmove g2g3 currmovenumber 16
info depth 15 currmove h2h4 currmovenumber 17
info depth 15 currmove b2b3 currmovenumber 18
info depth 15 currmove a2a4 currmovenumber 19
info depth 15 currmove g2g4 currmovenumber 20
bestmove e2e4
https://rwbc-chess.de/chronology.htm
--------------------------------------------------
The troll explosion at talkchess:
https://docs.google.com/spreadsheets/d/ ... KSptBx9AUs

RubiChess
Posts: 124
Joined: Fri Mar 30, 2018 5:20 am

Re: New engine releases 2020

Post by RubiChess » Mon Jan 13, 2020 5:06 pm

Norbert Raimund Leisner wrote:
Mon Jan 13, 2020 9:08 am
RubiChess 1.6.1 https://github.com/Matthies/RubiChess/releases
v1.6.1.1 exists too but without a Windows binary

MarvinChess 3.5.0 a8 https://github.com/bmdanielsson/marvin-chess/releases (with source code)
After beeing asked for a 1.6.1.1 binary at Github by a yellow panther (Norbert?) I finally decided to upload one. But as I mentioned, it shouldn't be a visible difference in strength.

Andreas

Damir
Posts: 2251
Joined: Mon Feb 11, 2008 2:53 pm
Location: Denmark
Full name: Damir Desevac

Re: New engine releases 2020

Post by Damir » Mon Jan 13, 2020 7:11 pm

I thought there is only a Black Panther,,,, :) :) :D

voffka
Posts: 83
Joined: Sat Jun 30, 2018 8:58 pm
Location: Ukraine
Full name: Volodymyr Shcherbyna

Re: New engine releases 2020

Post by voffka » Tue Jan 14, 2020 12:59 am

Image

Igel 2.3.1 at https://github.com/vshcherbyna/igel/releases/tag/2.3.1 (binaries for Raspberry Pi, Linux and Windows). This version is the one playing in TCEC S17 L2 and has significant strength improvements over Igel 2.3.0.

What's new:

- Re-write the root search routine
- Re-write aspiration loop with a smaller aspiration window
- Do not cutoff qsearch on pv moves in qsearch
- Handle repetitions at the qsearch entance
- More strict pruning conditions for quiet moves
- Proper way to pass position to child threads in SMP mode to handle repetitions
- Unified condition to killer moves as non-tactical moves
- Perft tests for six different positions

Known issues:

- Ponder mode is not operational. It will be fixed in next versions.

Relative strength compared to previous versions of Igel in 60s+0.6s time control:

Code: Select all

os=windows
hash=256
threads=1
tc=all/60+0.6
Score of Igel 2.3.1 64 POPCNT vs Igel 2.3.0 64 POPCNT: 1304 - 435 - 2242  [0.609] 3981
Elo difference: 77.08 +/- 7.06
Relative strength compared to Amoeba 3.0.w64p-l in CCRL 40/4 time control:

Code: Select all

os=windows
hash=256
threads=1
tc=40/102
Score of Igel 2.3.1 64 POPCNT vs Amoeba 3.0.w64p-l: 7 - 0 - 9  [0.719] 16
Elo difference: 162.99 +/- 109.83

joechess
Posts: 24
Joined: Sun Oct 30, 2016 10:02 pm

Re: New engine releases 2020

Post by joechess » Tue Jan 14, 2020 12:20 pm

The latest marvin keeps crashing for me in windows 7 with a msvcrt.dll error

Guenther
Posts: 3278
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Full name: Guenther Simon
Contact:

Re: New engine releases 2020

Post by Guenther » Tue Jan 14, 2020 1:15 pm

joechess wrote:
Tue Jan 14, 2020 12:20 pm
The latest marvin keeps crashing for me in windows 7 with a msvcrt.dll error
For me too. Good that I have created my own compile, which works here :)
(Win7 SSE4.1 no popcount hardware)
https://rwbc-chess.de/chronology.htm
--------------------------------------------------
The troll explosion at talkchess:
https://docs.google.com/spreadsheets/d/ ... KSptBx9AUs

joechess
Posts: 24
Joined: Sun Oct 30, 2016 10:02 pm

Re: New engine releases 2020

Post by joechess » Tue Jan 14, 2020 2:21 pm

Guenther wrote:
Tue Jan 14, 2020 1:15 pm
joechess wrote:
Tue Jan 14, 2020 12:20 pm
The latest marvin keeps crashing for me in windows 7 with a msvcrt.dll error
For me too. Good that I have created my own compile, which works here :)
(Win7 SSE4.1 no popcount hardware)
Any chance you could post it up pretty please?

Norbert Raimund Leisner
Posts: 1617
Joined: Tue May 20, 2008 2:57 pm
Location: Augsburg - Germany

Re: New engine releases 2020

Post by Norbert Raimund Leisner » Tue Jan 14, 2020 3:17 pm

Martin made an own Marvin 3.5.0 a8 compile here: https://github.com/bmdanielsson/marvin- ... naries.zip (with popcnt and the normal edition for older cpus)

Post Reply