New open-source engine, Bit-Genie

[Gabor Szots]

About the stop and quit command, I have to do a little bit of research for it. The best solution right now seems to run search on another thread.

I don't think I'd be happy to see it reserve 2 threads of my 4-core machine.

[aryan1508]

About the stop and quit command, I have to do a little bit of research for it. The best solution right now seems to run search on another thread.

I don't think I'd be happy to see it reserve 2 threads of my 4-core machine.

Is there a better solution? I could read the stdin everyone time I check the time, but that would slow down the search a lot.

[RubiChess]

About the stop and quit command, I have to do a little bit of research for it. The best solution right now seems to run search on another thread.

I don't think I'd be happy to see it reserve 2 threads of my 4-core machine.

You will hardly find a single engine that REALLY uses only n OS threads when you setoption name Threads value n
E.g. Stockfish running in Windows shows 4 threads in Taskmanager when it uses only one thread for searching. And 5 threads when you use 2 threads for searching. And so on.
One of the threads usually reads the UCI input in a blocking "readline". This is what Aryan probably wants to implement.
Important is that only n threads stress the CPU when searching. But I must admit, there is some potential of "cheating" using "relaxed" threads for whatever.

Regards, Andreas

[aryan1508]

I've implemented it here https://github.com/Aryan1508/Bit-Genie/ ... rch_thread. What happens is A thread is used for the UCI loop since reading the stdin in C++ is 'blocking'. The other 'worker' thread demands cpu since it is used for searching

[Gabor Szots]

I've implemented it here https://github.com/Aryan1508/Bit-Genie/ ... rch_thread. What happens is A thread is used for the UCI loop since reading the stdin in C++ is 'blocking'. The other 'worker' thread demands cpu since it is used for searching

OK, I'm going to use that for my testing.

[Gabor Szots]

You will hardly find a single engine that REALLY uses only n OS threads when you setoption name Threads value n
E.g. Stockfish running in Windows shows 4 threads in Taskmanager when it uses only one thread for searching. And 5 threads when you use 2 threads for searching. And so on.
One of the threads usually reads the UCI input in a blocking "readline". This is what Aryan probably wants to implement.
Important is that only n threads stress the CPU when searching. But I must admit, there is some potential of "cheating" using "relaxed" threads for whatever.

Thanks Andreas. Maybe that's the reason why I have so many time forfeits in my tournaments.

[Guenther]

You will hardly find a single engine that REALLY uses only n OS threads when you setoption name Threads value n
E.g. Stockfish running in Windows shows 4 threads in Taskmanager when it uses only one thread for searching. And 5 threads when you use 2 threads for searching. And so on.
One of the threads usually reads the UCI input in a blocking "readline". This is what Aryan probably wants to implement.
Important is that only n threads stress the CPU when searching. But I must admit, there is some potential of "cheating" using "relaxed" threads for whatever.

Thanks Andreas. Maybe that's the reason why I have so many time forfeits in my tournaments.

Do you mean time forfeits with Bit-Genie?
I had zero time losses in over 400 games now in CuteChess - may I ask, if this is perhaps under Arena? ;-)

(quite sure he could add a more defensive time management for 'those kind of GUIs')


Stats from those 400 games all at tc of 40/120

Code: Select all

Score of Rustic_Alpha2-64 vs Bit-Genie_r107-64: 93 - 74 - 33 [0.547]
...      Rustic_Alpha2-64 playing White: 43 - 34 - 23  [0.545] 100
...      Rustic_Alpha2-64 playing Black: 50 - 40 - 10  [0.550] 100
...      White vs Black: 83 - 84 - 33  [0.497] 200
Elo difference: 33.1 +/- 44.3, LOS: 92.9 %, DrawRatio: 16.5 %
200 of 200 games finished.

Player: Rustic_Alpha2-64
   "Draw by 3-fold repetition": 10
   "Draw by adjudication": 6
   "Draw by fifty moves rule": 7
   "Draw by insufficient mating material": 8
   "Draw by stalemate": 2
   "Loss: Black mates": 34
   "Loss: White mates": 40
   "Win: Black mates": 49
   "Win: Black wins by adjudication: user decision": 1
   "Win: White mates": 43
Player: Bit-Genie_r107-64
   "Draw by 3-fold repetition": 10
   "Draw by adjudication": 6
   "Draw by fifty moves rule": 7
   "Draw by insufficient mating material": 8
   "Draw by stalemate": 2
   "Loss: Black mates": 49
   "Loss: Black wins by adjudication: user decision": 1
   "Loss: White mates": 43
   "Win: Black mates": 34
   "Win: White mates": 40

Code: Select all

Score of Loki_120-64 vs Bit-Genie_100-64: 85 - 82 - 33 [0.507]
...      Loki_120-64 playing White: 42 - 40 - 18  [0.510] 100
...      Loki_120-64 playing Black: 43 - 42 - 15  [0.505] 100
...      White vs Black: 84 - 83 - 33  [0.502] 200
Elo difference: 5.2 +/- 44.2, LOS: 59.2 %, DrawRatio: 16.5 %
200 of 200 games finished.

Player: Loki_120-64
   "Draw by 3-fold repetition": 15
   "Draw by adjudication": 1
   "Draw by fifty moves rule": 8
   "Draw by insufficient mating material": 7
   "Draw by stalemate": 2
   "Loss: Black mates": 40
   "Loss: White mates": 42
   "Win: Black mates": 43
   "Win: White mates": 42
Player: Bit-Genie_100-64
   "Draw by 3-fold repetition": 15
   "Draw by adjudication": 1
   "Draw by fifty moves rule": 8
   "Draw by insufficient mating material": 7
   "Draw by stalemate": 2
   "Loss: Black mates": 43
   "Loss: White mates": 42
   "Win: Black mates": 40
   "Win: White mates": 42

[Gabor Szots]

Do you mean time forfeits with Bit-Genie?
I had zero time losses in over 400 games now in CuteChess - may I ask, if this is perhaps under Arena?

Yes, under Arena, and not specific to Bit-Genie.

[Guenther]

Do you mean time forfeits with Bit-Genie?
I had zero time losses in over 400 games now in CuteChess - may I ask, if this is perhaps under Arena? ;-)

Yes, under Arena, and not specific to Bit-Genie.

So you mean more specific to Arena? But this is already known, sorry.
(uci filter and changing some graphic display options / animations might help, but I never user Arena for good reasons)

[mvanthoor]

Yes, under Arena, and not specific to Bit-Genie.

As Guenther says; enable the UCI-filter, so Arena disregards the engine output for the first 5-10 seconds of the search, and disable most of the graphics. If you don't, these things take too much time in Arena's main thread. When it then comes back to interpreting the engine's move, it "thinks" the engine is too late... where actually, it's Arena itself that is too late.

I had the same thing when I first started to test Rustic, and fixed it with the above settings: enable UCI-filter, disable as much graphics as possible. Later, I changed the engine's output:

- Always output the search stats if an iteration has finished.
- Only output intermediate search stats (such as currmove, etc) every 2 seconds.

This makes the engine send the first 7-8 plies almost at once, and then updates every 2 seconds. (I don't like to set it so it only starts sending updates after 2 seoncds, because in that case, I'll miss the first 7-9 plies in the engine output.)

Then I switched to CuteChess, because I can run 4 games at the same time (and thus have 8 engines loaded). Two engines in one game therefore share a single core, back and forth.