what happened to scorpio NN in TCEC?

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

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Daniel Shawul
Posts: 3724
Joined: Tue Mar 14, 2006 10:34 am
Location: Ethiopia
Contact:

Re: what happened to scorpio NN in TCEC?

Post by Daniel Shawul » Sun Nov 18, 2018 11:20 am

Guenther wrote:
Sun Nov 18, 2018 11:06 am
Daniel Shawul wrote:
Sun Nov 18, 2018 10:49 am

According to the protocol, engines can take upto 1 hour to do initialization.

Daniel
I don't know why they don't use a compile which allows a longer start time (if this is the main problem)? It is an easy change.
I remember very well that LC0 also had that problem with too long start time in the first months.

OTH, 1 hour of course would be no fun and no one would ever use engines really taking up
one hour for initialising.
The real network initialization takes up 30 seconds on the GPU but apparently that is too much for cutechess.

The blitz version used an approach where the initialization of networks is done after a "new" or "go" command is sent.
That means this time is counted towards scorpio's play time and will work as long as the time control is longer than the NN loading time.
The workaround should have worked but I don't think they used the correct version judging from the nemorino-scorpio log file -- it still started loading networks immediately without waiting for go/new.

This workaround works fine with cutechess-cli in my test with a fake 2 min delayed loading time as shown below.

Code: Select all

8 >scorpio-nn1(0): xboard
9 >scorpio-nn1(0): protover 2
9 >scorpio-nn2(1): xboard
9 >scorpio-nn2(1): protover 2
14 <scorpio-nn1(0): feature done=0
14 >scorpio-nn1(0): accepted done
14 <scorpio-nn2(1): feature done=0
14 >scorpio-nn2(1): accepted done
25 <scorpio-nn1(0): ht 4194304 X 16 = 64.0 MB
25 <scorpio-nn1(0): eht 524288 X 8 = 8.0 MB
25 <scorpio-nn1(0): pht 32768 X 24 = 0.8 MB
25 <scorpio-nn1(0): treeht 335539200 X 40 = 12799.8 MB
25 <scorpio-nn2(1): ht 4194304 X 16 = 64.0 MB
25 <scorpio-nn2(1): eht 524288 X 8 = 8.0 MB
25 <scorpio-nn2(1): pht 32768 X 24 = 0.8 MB
25 <scorpio-nn2(1): treeht 335539200 X 40 = 12799.8 MB
65 <scorpio-nn1(0): processors [16]
66 <scorpio-nn2(1): processors [16]
129 <scorpio-nn1(0): EgbbProbe 4.3 by Daniel Shawul
129 <scorpio-nn2(1): EgbbProbe 4.3 by Daniel Shawul
180 egbbs loaded !      ding egbbs....
318 <scorpio-nn1(0): Loading neural network : nets/net-0.pb
180 egbbs loaded !      ding egbbs....
318 <scorpio-nn2(1): Loading neural network : nets/net-26.pb
320 <scorpio-nn2(1): Loading graph on /cpu:0
320 <scorpio-nn1(0): Loading graph on /cpu:0
388 <scorpio-nn1(0): Neural network loaded !	
388 <scorpio-nn1(0): loading_time = 0s
394 <scorpio-nn2(1): Neural network loaded !	
394 <scorpio-nn2(1): loading_time = 0s
30065 <scorpio-nn1(0): feature name=1 myname="Scorpio_2.8.9 MCTS+NN"
30066 >scorpio-nn1(0): accepted name
30066 >scorpio-nn1(0): accepted myname
30066 <scorpio-nn1(0): feature sigint=0 sigterm=0
30066 >scorpio-nn1(0): rejected sigint
30066 >scorpio-nn1(0): rejected sigterm
30066 <scorpio-nn1(0): feature setboard=1 draw=0 colors=0
30066 >scorpio-nn1(0): accepted setboard
30066 >scorpio-nn1(0): rejected draw
30066 >scorpio-nn1(0): rejected colors
30066 <scorpio-nn1(0): feature smp=0 memory=0
30066 >scorpio-nn1(0): accepted smp
30066 >scorpio-nn1(0): accepted memory
30066 <scorpio-nn1(0): feature option="log -check 0"
30066 >scorpio-nn1(0): accepted option
30066 <scorpio-nn1(0): feature option="clear_hash -button"
30066 >scorpio-nn1(0): accepted option
30066 <scorpio-nn1(0): feature option="resign -spin 800 100 30000"
30066 >scorpio-nn1(0): accepted option
30066 <scorpio-nn1(0): feature option="cores -spin 1 1 4096"
30066 >scorpio-nn1(0): accepted option
30066 <scorpio-nn1(0): feature option="ht -spin 64 1 131072"
30066 >scorpio-nn1(0): accepted option
30066 <scorpio-nn1(0): feature option="eht -spin 8 1 16384"
30066 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="pht -spin 1 1 256"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="egbb_path -path /home/daniel/egbb/"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="nn_path -path nets/net-0.pb"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="egbb_cache_size -spin 32 1 16384"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="egbb_load_type -spin 3 0 3"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="egbb_depth_limit -spin 6 0 64"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="egbb_ply_limit_percent -spin 75 0 100"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="n_devices -spin 1 1 128"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="device_type -combo *CPU /// GPU "
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="delay -spin 1 0 1000"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="float_type -combo FLOAT /// *HALF  /// INT8 "
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="smp_type -combo *YBW /// ABDADA /// SHT "
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="smp_depth -spin 8 1 10"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="use_singular -check 1"
30067 >scorpio-nn1(0): accepted option
30067 <scorpio-nn1(0): feature option="use_probcut -check 0"
30067 >scorpio-nn1(0): accepted option
30068 <scorpio-nn1(0): feature option="singular_margin -spin 32 0 1000"
30068 >scorpio-nn1(0): accepted option
30068 <scorpio-nn1(0): feature option="probcut_margin -spin 195 0 1000"
30068 >scorpio-nn1(0): accepted option
30068 <scorpio-nn1(0): feature option="aspiration_window -spin 10 0 100"
30068 >scorpio-nn1(0): accepted option
30068 <scorpio-nn1(0): feature option="UCTKmin -spin 0 0 100"
30068 >scorpio-nn1(0): accepted option
30068 <scorpio-nn1(0): feature option="UCTKmax -spin 30 0 100"
30068 >scorpio-nn1(0): accepted option
30068 <scorpio-nn1(0): feature option="reuse_tree -check 1"
30068 >scorpio-nn1(0): accepted option
30068 <scorpio-nn1(0): feature option="backup_type -combo *MINMAX AVERAGE MIX MINMAX_MEM AVERAGE_MEM MIX_MEM CLASSIC MIX_VISIT"
30068 >scorpio-nn1(0): accepted option
30068 <scorpio-nn1(0): feature option="frac_alphabeta -spin 0 0 100"
30068 >scorpio-nn1(0): accepted option
30068 <scorpio-nn1(0): feature option="frac_freeze_tree -spin 100 0 100"
30068 >scorpio-nn1(0): accepted option
30068 <scorpio-nn1(0): feature option="frac_abrollouts -spin 20 0 100"
30069 >scorpio-nn1(0): accepted option
30069 <scorpio-nn1(0): feature option="frac_abprior -spin 30 0 100"
30069 >scorpio-nn1(0): accepted option
30069 <scorpio-nn1(0): feature option="frac_width -spin 100 0 1000"
30069 >scorpio-nn1(0): accepted option
30069 <scorpio-nn1(0): feature option="mcts_strategy_depth -spin 30 0 100"
30069 >scorpio-nn1(0): accepted option
30069 <scorpio-nn1(0): feature option="alphabeta_depth -spin 16 1 100"
30069 >scorpio-nn1(0): accepted option
30069 <scorpio-nn1(0): feature option="evaluate_depth -spin 0 0 100"
30069 >scorpio-nn1(0): accepted option
30069 <scorpio-nn1(0): feature option="virtual_loss -spin 3 0 1000"
30069 >scorpio-nn1(0): accepted option
30069 <scorpio-nn1(0): feature option="visit_threshold -spin 800 0 1000"
30069 >scorpio-nn1(0): accepted option
30069 <scorpio-nn1(0): feature option="montecarlo -check 1"
30069 >scorpio-nn1(0): accepted option
30069 <scorpio-nn1(0): feature option="treeht -spin 12799 0 131072"
30069 >scorpio-nn1(0): accepted option
30069 <scorpio-nn1(0): feature done=1
30069 >scorpio-nn1(0): accepted done
30069 <scorpio-nn2(1): feature name=1 myname="Scorpio_2.8.9 MCTS+NN"
30069 >scorpio-nn2(1): accepted name
30069 >scorpio-nn2(1): accepted myname
30069 <scorpio-nn2(1): feature sigint=0 sigterm=0
30070 >scorpio-nn2(1): rejected sigint
30070 >scorpio-nn2(1): rejected sigterm
30070 <scorpio-nn2(1): feature setboard=1 draw=0 colors=0
30070 >scorpio-nn2(1): accepted setboard
30070 >scorpio-nn2(1): rejected draw
30070 >scorpio-nn2(1): rejected colors
30070 <scorpio-nn2(1): feature smp=0 memory=0
30070 >scorpio-nn2(1): accepted smp
30070 >scorpio-nn2(1): accepted memory
30070 <scorpio-nn2(1): feature option="log -check 0"
30070 >scorpio-nn2(1): accepted option
30070 <scorpio-nn2(1): feature option="clear_hash -button"
30070 >scorpio-nn2(1): accepted option
30070 <scorpio-nn2(1): feature option="resign -spin 800 100 30000"
30070 >scorpio-nn2(1): accepted option
30070 <scorpio-nn2(1): feature option="cores -spin 1 1 4096"
30070 >scorpio-nn2(1): accepted option
30070 <scorpio-nn2(1): feature option="ht -spin 64 1 131072"
30070 >scorpio-nn2(1): accepted option
30070 <scorpio-nn2(1): feature option="eht -spin 8 1 16384"
30070 >scorpio-nn2(1): accepted option
30070 <scorpio-nn2(1): feature option="pht -spin 1 1 256"
30070 >scorpio-nn2(1): accepted option
30070 <scorpio-nn2(1): feature option="egbb_path -path /home/daniel/egbb/"
30070 >scorpio-nn2(1): accepted option
30070 <scorpio-nn2(1): feature option="nn_path -path nets/net-26.pb"
30070 >scorpio-nn2(1): accepted option
30070 <scorpio-nn2(1): feature option="egbb_cache_size -spin 32 1 16384"
30071 >scorpio-nn2(1): accepted option
30071 <scorpio-nn2(1): feature option="egbb_load_type -spin 3 0 3"
30071 >scorpio-nn2(1): accepted option
30071 <scorpio-nn2(1): feature option="egbb_depth_limit -spin 6 0 64"
30071 >scorpio-nn2(1): accepted option
30071 <scorpio-nn2(1): feature option="egbb_ply_limit_percent -spin 75 0 100"
30071 >scorpio-nn2(1): accepted option
30071 <scorpio-nn2(1): feature option="n_devices -spin 1 1 128"
30071 >scorpio-nn2(1): accepted option
30071 <scorpio-nn2(1): feature option="device_type -combo *CPU /// GPU "
30071 >scorpio-nn2(1): accepted option
30071 <scorpio-nn2(1): feature option="delay -spin 1 0 1000"
30071 >scorpio-nn2(1): accepted option
30071 <scorpio-nn2(1): feature option="float_type -combo FLOAT /// *HALF  /// INT8 "
30071 >scorpio-nn2(1): accepted option
30071 <scorpio-nn2(1): feature option="smp_type -combo *YBW /// ABDADA /// SHT "
30071 >scorpio-nn2(1): accepted option
30071 <scorpio-nn2(1): feature option="smp_depth -spin 8 1 10"
30072 >scorpio-nn2(1): accepted option
30072 <scorpio-nn2(1): feature option="use_singular -check 1"
30072 >scorpio-nn2(1): accepted option
30072 <scorpio-nn2(1): feature option="use_probcut -check 0"
30072 >scorpio-nn2(1): accepted option
30072 <scorpio-nn2(1): feature option="singular_margin -spin 32 0 1000"
30072 >scorpio-nn2(1): accepted option
30072 <scorpio-nn2(1): feature option="probcut_margin -spin 195 0 1000"
30072 >scorpio-nn2(1): accepted option
30072 <scorpio-nn2(1): feature option="aspiration_window -spin 10 0 100"
30072 >scorpio-nn2(1): accepted option
30072 <scorpio-nn2(1): feature option="UCTKmin -spin 0 0 100"
30072 >scorpio-nn2(1): accepted option
30072 <scorpio-nn2(1): feature option="UCTKmax -spin 30 0 100"
30072 >scorpio-nn2(1): accepted option
30072 <scorpio-nn2(1): feature option="reuse_tree -check 1"
30072 >scorpio-nn2(1): accepted option
30072 <scorpio-nn2(1): feature option="backup_type -combo *MINMAX AVERAGE MIX MINMAX_MEM AVERAGE_MEM MIX_MEM CLASSIC MIX_VISIT"
30072 >scorpio-nn2(1): accepted option
30072 <scorpio-nn2(1): feature option="frac_alphabeta -spin 0 0 100"
30073 >scorpio-nn2(1): accepted option
30073 <scorpio-nn2(1): feature option="frac_freeze_tree -spin 100 0 100"
30073 >scorpio-nn2(1): accepted option
30073 <scorpio-nn2(1): feature option="frac_abrollouts -spin 20 0 100"
30073 >scorpio-nn2(1): accepted option
30073 <scorpio-nn2(1): feature option="frac_abprior -spin 30 0 100"
30073 >scorpio-nn2(1): accepted option
30073 <scorpio-nn2(1): feature option="frac_width -spin 100 0 1000"
30073 >scorpio-nn2(1): accepted option
30073 <scorpio-nn2(1): feature option="mcts_strategy_depth -spin 30 0 100"
30073 >scorpio-nn2(1): accepted option
30073 <scorpio-nn2(1): feature option="alphabeta_depth -spin 16 1 100"
30073 >scorpio-nn2(1): accepted option
30073 <scorpio-nn2(1): feature option="evaluate_depth -spin 0 0 100"
30073 >scorpio-nn2(1): accepted option
30073 <scorpio-nn2(1): feature option="virtual_loss -spin 3 0 1000"
30073 >scorpio-nn2(1): accepted option
30073 <scorpio-nn2(1): feature option="visit_threshold -spin 800 0 1000"
30073 >scorpio-nn2(1): accepted option
30074 <scorpio-nn2(1): feature option="montecarlo -check 1"
30074 >scorpio-nn2(1): accepted option
30074 <scorpio-nn2(1): feature option="treeht -spin 12799 0 131072"
30074 >scorpio-nn2(1): accepted option
30074 <scorpio-nn2(1): feature done=1
30074 >scorpio-nn2(1): accepted done
Started game 1 of 8000 (scorpio-nn1 vs scorpio-nn2)
30074 >scorpio-nn1(0): new
30074 >scorpio-nn1(0): force
30074 >scorpio-nn1(0): setboard rq2kb1r/p2p1ppp/1pb1pn2/8/2P1P3/P1N5/1PQ1BPPP/R1B1K2R w KQkq - 0 1
30074 >scorpio-nn1(0): level 0 1000 0.1
30074 >scorpio-nn1(0): post
30074 >scorpio-nn1(0): easy
30074 >scorpio-nn1(0): computer
30074 >scorpio-nn1(0): name scorpio-nn2
30074 >scorpio-nn2(1): new
30074 >scorpio-nn2(1): force
30074 >scorpio-nn2(1): setboard rq2kb1r/p2p1ppp/1pb1pn2/8/2P1P3/P1N5/1PQ1BPPP/R1B1K2R w KQkq - 0 1
30074 >scorpio-nn2(1): level 0 1000 0.1
30074 >scorpio-nn2(1): post
30075 >scorpio-nn2(1): easy
30075 >scorpio-nn2(1): computer
30075 >scorpio-nn2(1): name scorpio-nn1
30075 >scorpio-nn1(0): time 6000000
otim 6000000
30075 >scorpio-nn1(0): go
30250 <scorpio-nn2(1): Hello scorpio-nn1!
30253 <scorpio-nn1(0): Hello scorpio-nn2!
30254 <scorpio-nn1(0): # [sv = 800 , hply = 0]
30323 <scorpio-nn1(0): 63 -25 7 93  Bc1-e3 Bf8-d6 Ke1-c1
30374 <scorpio-nn1(0): 64 8 12 177  Bc1-e3 Bf8-d6 Ke1-c1
30438 <scorpio-nn1(0): 65 34 18 265  Ke1-g1 Bf8-c5 Bc1-e3 Bc5xe3 f2xe3
30493 <scorpio-nn1(0): 66 50 23 345  Bc1-e3 Bf8-d6 Ke1-c1
30543 <scorpio-nn1(0): 67 59 29 426  Ke1-g1 Bf8-c5 Bc1-e3 Bc5xe3 f2xe3
30614 <scorpio-nn1(0): 68 67 36 536  Ke1-g1 Bf8-c5 Bc1-e3 Bc5xe3 f2xe3
30685 <scorpio-nn1(0): 69 71 43 632  Ke1-f1 Bf8-c5 Bc1-e3 Bc5xe3 f2xe3
30737 <scorpio-nn1(0): 70 73 48 718  g2-g3 Qb8-c7 Ke1-g1
30800 <scorpio-nn1(0): # Move  Value  Policy   V+P    Visits                  PV
30800 <scorpio-nn1(0): #  1    0.60    0.70    1.30      44   Ke1-g1 Bf8-c5 Bc1-e3 Ke8-g8
30800 <scorpio-nn1(0): #  2    0.60    0.65    1.25      36   Bc1-e3 Bf8-d6 Ke1-c1
30800 <scorpio-nn1(0): #  3    0.60    0.62    1.23      33   h2-h3 Bf8-d6 Ke1-g1
30800 <scorpio-nn1(0): #  4    0.61    0.62    1.23      34   b2-b3 Bf8-d6 Bc1-b2
30800 <scorpio-nn1(0): #  5    0.60    0.62    1.22      34   Bc1-d2 Bf8-d6 Ke1-c1
30800 <scorpio-nn1(0): #  6    0.61    0.62    1.22      35   f2-f3 Bf8-c5 b2-b3
30800 <scorpio-nn1(0): #  7    0.44    0.61    1.05       5   Bc1-g5 h7-h6 Bg5-e3 Bf8-d6
30800 <scorpio-nn1(0): #  8    0.60    0.61    1.21      31   Be2-f3 Bf8-d6 Bc1-e3
30800 <scorpio-nn1(0): #  9    0.60    0.60    1.20      29   f2-f4 Bf8-c5 Be2-f3
30800 <scorpio-nn1(0): # 10    0.61    0.59    1.19      39   g2-g3 Qb8-c7 Ke1-g1
30800 <scorpio-nn1(0): # 11    0.60    0.59    1.19      29   b2-b4 Bf8-d6 Bc1-e3
30800 <scorpio-nn1(0): # 12    0.60    0.58    1.19      32   h2-h4 Bf8-d6 Ke1-g1
30800 <scorpio-nn1(0): # 13    0.60    0.58    1.18      32   Rh1-f1 Qb8xh2 Be2-f3 Bf8-d6
30800 <scorpio-nn1(0): # 14    0.61    0.57    1.18      37   Ke1-d1 Qb8-d6 Bc1-d2 Ra8-d8 h2-h3
30800 <scorpio-nn1(0): # 15    0.60    0.57    1.17      34   Be2-d3 Bf8-d6 Bc1-e3
30800 <scorpio-nn1(0): # 16    0.60    0.56    1.17      30   Ra1-b1 Bf8-d6 Bc1-e3
30800 <scorpio-nn1(0): # 17    0.60    0.55    1.15      31   Qc2-d3 Qb8-c7 Ke1-g1
30800 <scorpio-nn1(0): # 18    0.60    0.55    1.16      33   Be2-d1 Bf8-d6 Bc1-e3
30800 <scorpio-nn1(0): # 19    0.60    0.54    1.14      29   g2-g4 Bf8-d6 g4-g5

The nemorino-scorpio log file after the restart, you can see it is still loading the NN immediately. With the workaround it should have waited
until the first go/new command. It is most likely because the wrong version is used ...

Code: Select all

Indexing opening suite...
27 >Nemorino 5.05(0): uci
30 >ScorpioNN 2.8.9(1): xboard
30 >ScorpioNN 2.8.9(1): protover 2
65 <Nemorino 5.05(0): Initialization Time: 24ms
67 <Nemorino 5.05(0): id name Nemorino 5.05
67 <Nemorino 5.05(0): id author Christian Guenther
67 <Nemorino 5.05(0): option name BookFile type string default book.bin
67 <Nemorino 5.05(0): option name Clear Hash type button
68 <Nemorino 5.05(0): option name Contempt type spin default 0 min -1000 max 1000
68 <Nemorino 5.05(0): option name Hash type spin default 32 min 1 max 16384
68 <Nemorino 5.05(0): option name MoveOverhead type spin default 0 min 0 max 60000
69 <Nemorino 5.05(0): option name MultiPV type spin default 1 min 1 max 216
69 <Nemorino 5.05(0): option name OwnBook type check default false
69 <Nemorino 5.05(0): option name Ponder type check default false
69 <Nemorino 5.05(0): option name SyzygyPath type string
69 <Nemorino 5.05(0): option name SyzygyProbeDepth type spin default 0 min 0 max 129
69 <Nemorino 5.05(0): option name Threads type spin default 44 min 1 max 128
69 <Nemorino 5.05(0): option name UCI_Chess960 type check default false
69 <Nemorino 5.05(0): option name UCI_Opponent type string
69 <Nemorino 5.05(0): uciok
69 >Nemorino 5.05(0): setoption name Hash value 16384
69 >Nemorino 5.05(0): setoption name OwnBook value false
69 >Nemorino 5.05(0): setoption name SyzygyPath value C:/Syzygy/
69 >Nemorino 5.05(0): setoption name Threads value 43
69 >Nemorino 5.05(0): isready
90 <Nemorino 5.05(0): info string Found 510 tablebases
109 <Nemorino 5.05(0): readyok
1121 <ScorpioNN 2.8.9(1): feature done=0
1121 >ScorpioNN 2.8.9(1): accepted done
1121 <ScorpioNN 2.8.9(1): ht 4194304 X 16 = 64.0 MB
1121 <ScorpioNN 2.8.9(1): eht 524288 X 8 = 8.0 MB
1121 <ScorpioNN 2.8.9(1): pht 32768 X 24 = 0.8 MB
1121 <ScorpioNN 2.8.9(1): treeht 335539200 X 40 = 12799.8 MB
1197 <ScorpioNN 2.8.9(1): processors [128]
1197 <ScorpioNN 2.8.9(1): EgbbProbe 4.3 by Daniel Shawul
1224 <ScorpioNN 2.8.9(1): Loading egbbs....0 egbbs loaded !      
1224 <ScorpioNN 2.8.9(1): Loading neural network : C:/Users/TCEC/Engines/scorpio/nets-epd-10m/net-6x64.uff
1224 <ScorpioNN 2.8.9(1): Loading graph on /gpu:0
31831 <ScorpioNN 2.8.9(1): Neural network loaded !	
31831 <ScorpioNN 2.8.9(1): loading_time = 30s
31927 <ScorpioNN 2.8.9(1): feature name=1 myname="Scorpio_2.8.9 MCTS+NN"
31927 >ScorpioNN 2.8.9(1): accepted name
31927 >ScorpioNN 2.8.9(1): accepted myname
31928 <ScorpioNN 2.8.9(1): feature sigint=0 sigterm=0
31928 >ScorpioNN 2.8.9(1): rejected sigint
31928 >ScorpioNN 2.8.9(1): rejected sigterm
Daniel

Krzysztof Grzelak
Posts: 811
Joined: Tue Jul 15, 2014 10:47 am

Re: what happened to scorpio NN in TCEC?

Post by Krzysztof Grzelak » Sun Nov 18, 2018 11:29 am

And maybe change to ChessGUI instead cutechess-cli.

elcabesa
Posts: 806
Joined: Sun May 23, 2010 11:32 am
Contact:

Re: what happened to scorpio NN in TCEC?

Post by elcabesa » Sun Nov 18, 2018 2:10 pm

this is the command just added on TCEC chat by kanchess

scorpio
Nightbot : Scorpio did not crash yet and the game will eventually be replayed

it seems a problem between cpu and gpu server, from what I understood.

so please stop taking offense and start behave politely :)

User avatar
CMCanavessi
Posts: 834
Joined: Thu Dec 28, 2017 3:06 pm
Location: Argentina

Re: what happened to scorpio NN in TCEC?

Post by CMCanavessi » Sun Nov 18, 2018 7:09 pm

lucasart wrote:
Sun Nov 18, 2018 6:15 am
CMCanavessi wrote:
Sat Nov 17, 2018 11:47 pm
lucasart wrote:
Sat Nov 17, 2018 11:40 pm
This tournament is becoming a bit of a mess:

They were supposed to restart without Scorpio, but they just restarted and forgot the "without Scorpio" part…
They did, but they aborted it after 3 games and re-started it with Scorpio in again, apparently with the same version they used for teh blitz test games (scorpio did 80 games without crashing even once)

What about Pirarucu? It had 3 crashes, and they said 3 crashes were disqualifying…
From what I've read, it crashed 2 times, the 3rd was a problem with something else

Again this will stall the live every time Pirarucu or Scorpio crashes (annoying), and will pollute results (unfair). Some engines will get lucky and "win" against Pirarucu due tocrashes, where they otherwise would have lost or drawn, some will not be gifted with opponent crashes.

As for Scorpio, it cannot pollute the results, because it would lose every single game regardless.
Let's hope now that things have stabilized no crashes will appear.
What has stabilized ? Nothing has been fixed. The same causes will have the same effect. Scorpio has already crashed in its 1st game. Pirarucu will crash too, unless it gets very lucky.
Scorpio hasn't crashed yet in the "new" tournament (after the restart). The loss in game 2 was due to a connection issue between the GPU-based computer and the normal one, as reported by TCEC admins (and the game will be replayed). We'll see how it goes.
Follow my tournament and some Leela gauntlets live at http://twitch.tv/ccls

Post Reply