On-line engine blitz tourney May

[Joost Buijs]

The monthly online engine blitz tournament for May will take place on:

Saturday, May 10, 2:00 PM EST (Boston time) 20:00 CET (Amsterdam time)

To connect, use the following command:

winboard -zp -ics -icshost nightmare-chess.nl -icshelper timeseal -fcp ENGINE.exe -fd ENGINEFOLDER -autoKibitz -keepAlive 30
(For UCI engines, add -fUCI)

If you don’t have an account, please submit a request to admin@nightmare-chess.nl.

Our tournament manager will be Eendje (Lisebeth @ ICC).

The format will be 9 rounds, Swiss system, 5/1 time control.

Lisebeth prefers to start the tournament on the hour, so please log in 15 minutes early. Late entries are not supported.

[Joost Buijs]

It was fun as usual, in the end LC0 on an AMD AI 9 HX 370 mobile processor won the tournament.
I didn't expect LC0 to win, I just wanted to test the new mini-PC I bought a couple of days ago, to see if it's stable or not.
I'm interested in doing some experiments with the integrated NPU, that's why I bought it, LC0 ran on the integrated GPU.
Isa had engine troubles, eendje had to forfeit him after a few rounds, no problem, these things sometimes happen.

Code: Select all

:Tourney Players: Round 9 of 9 
:
:     Name              Rating Score Perfrm Upset  Results 
:     ----------------- ------ ----- ------ ------ ------- 
:  1 +lczAIHX           [2726]  6.5  [2842] [ 336] +20w =03w +06b =12w +05b =02w =07b =04w +14b 
:  2 +rofChade          [2956]  6.0  [2848] [   0] +10w +13w =12b =05w =04b =01b =03w =09b +08w 
:  3 +ArasanX           [2893]  6.0  [2817] [  31] =14w =01b =04w +08b +07w =13b =02b =05w +09w 
:  4 +NightmareX        [2883]  6.0  [2871] [  41] +15w =06w =03b +09b =02w =05w =12b =01b +13w 
:  5 +PatriciaRB        [2732]  6.0  [2842] [ 407] +16b +17w =09w =02b -01w =04b +13w =03b +12w 
:  6 +atomGoldbar       [2639]  5.5  [2533] [ 122] +19w =04b -01w =10b +14w -07b =08w +16w +17b 
:  7 +Viper             [2514]  5.5  [2572] [ 317] +21w -12b +11w =14b -03b +06w =01w =13b +19b 
:  8 +WaDuuttie         [2481]  5.5  [2617] [ 433] +22w -09b +16w -03w +11b +15w =06b +12w -02b
:  9 -Marvin            [2806]  5.0  [2683] [  89] +11w +08w =05b -04w =15b =12w +17b =02w -03b 
: 10 +Dirty             [2391]  5.0  [2333] [ 124] -02b =14b =15w =06w =16b =20w =19b +22w +21b 
: 11 +atomFloyd         [2175]  5.0  [2312] [  97] -09b +22w -07b +21w -08w +18b =15b =19w +20b 
: 12 +Goldbar           [2835]  4.5  [2667] [  84] +25b +07w =02w =01b =13w =09b =04w -08b -05b 
: 13 +atomNightmare     [2686]  4.5  [2623] [ 177] +18w -02b +25w +17w =12b =03w -05b =07w -04b 
: 14 +EveAnn            [2373]  4.5  [2408] [ 339] =03b =10w +18b =07w -06b =19w =16b +20w -01w 
: 15 +Ares              [2370]  4.5  [2348] [ 228] -04b +19w =10b +25b =09w -08b =11w -17w +23b 
: 16 +rpiArminius       [2169]  4.5  [2312] [ 341] -05w +23b -08b +18w =10w =17b =14w -06b +22b
: 17 +Arminius          [2425]  4.0  [2323] [   0] +23w -05b +20w -13b =19b =16w -09w +15b -06w 
: 18 +Barbarossa        [2149]  4.0  [2045] [   0] -13b -20b -14w -16b +23w -11w +22b +21w +24b 
: 19 +DogPC             [2040]  4.0  [2261] [ 717] -06b -15b +23w +20b =17w =14b =10w =11b -07w 
: 20 +feligres          [2157]  3.5  [2129] [ 117] -01b +18w -17b -19w +21b =10b +23w -14b -11w 
: 21 +Abbess            [1981]  3.0  [1954] [   0] -07b -25w +22b -11b -20w +23b +24w -18b -10w 
: 22 +Skiull            [1917]  3.0  [1960] [ 310] -08b -11b -21w +23b +25w +24b -18w -10b -16w 
: 23 +atomSargon        [1708]  1.0  [1751] [   0] -17b -16w -19b -22w -18b -21w -20b +24w -15w 
: 24 +_BYE_             [   0]  0.0  [1538] [   0] -22w -21b -23b -18w
: 25 -Isa               [forf]  1.0  [2117] [   0] -12w +21b -13b -15w -22b 
:
:     Average Rating    2426.4

[Modern Times]

Quite a close tournament - 4 engines tied for 2nd place. I didn't expect PatriciaRB to be in that group, pretty good showing for it. It doesn't support Ponder, but I don't think that mattered too much.

[mar]

Isa had engine troubles, eendje had to forfeit him after a few rounds, no problem, these things sometimes happen.

my fault - UCI pondering bug, already fixed

polyglot doesn't wait for bestmove from engine after pondermiss (stop) and sends new position and go immediately;
this not something I expected. now we send bestmove after a ponder search no matter what (polyglot option SyncStop would have helped,
but it's rather difficult to figure out what's wrong during the tournament)

[Joost Buijs]

Quite a close tournament - 4 engines tied for 2nd place. I didn't expect PatriciaRB to be in that group, pretty good showing for it. It doesn't support Ponder, but I don't think that mattered too much.

Patricia seems to be very strong, his performance rating is good too. I watched a few games, tactically very strong. It's a new engine for me, besides this tournament I've been doing exactly zero with computer-chess for the past 5 years, so my knowledge about it is a bit rusty.

It's also amazing that LC0 did so well on a mobile processor with just 10.000 N/s, without EGTB, and small book of only a few moves.

[Joost Buijs]

Isa had engine troubles, eendje had to forfeit him after a few rounds, no problem, these things sometimes happen.

my fault - UCI pondering bug, already fixed

polyglot doesn't wait for bestmove from engine after pondermiss (stop) and sends new position and go immediately;
this not something I expected. now we send bestmove after a ponder search no matter what (polyglot option SyncStop would have helped,
but it's rather difficult to figure out what's wrong during the tournament)

It's difficult to fix something during a tournament, I'm always afraid that the server will crash, the 'Lasker Chess Server' has many flaws, and is very prone to buffer overruns.

Nightmare also sends the best move after a ponder-miss, it has no problems with 'stop, position and go' immediately. When it receives 'stop' during ponder-search it sends best-move and ponder-move, and waits for the next command from polyglot (which probably already is in the receiving buffer).

The only problem Nightmare has with pondering is that it sometimes sends a wrong ponder-move to polyglot, this is still something that needs to be fixed.

[mar]

Nightmare also sends the best move after a ponder-miss, it has no problems with 'stop, position and go' immediately. When it receives 'stop' during ponder-search it sends best-move and ponder-move, and waits for the next command from polyglot (which probably already is in the receiving buffer).

yes - this is the proper way to do things - so now we sync at stop instead (=wait for the search to finish and send bestmove), before only an atomic flag was set on stop which had caused problems

in general, I have to say that such online tournaments test the engine to its limits (apart from being fun of course)
it's necessary to play with a book to conserve time initially, time management has to be robust since it's blitz,
in order to do well one has to support pondering and smp, there's so automatic adjudication either so some engines have
to play until mate if the opponent doesn't resign
we also found a dumb mate-related bug by applying TT mate correction twice when saving to hash table, causing a half-point loss in the last tourney

[flok]

Something went wrong over here: polyglot chose to use only 1 thread with Dog?

[Joost Buijs]

Something went wrong over here: polyglot chose to use only 1 thread with Dog?

Did you set the number of threads in polyglot.ini?

Ususally I don't have problems with it, but I always check if it does what I want it to do.