fluxroot / pulse Goto Github PK

Hi,

I had an issue with it not linking (using gcc (g++) 4.9.2-10). That was solved by adding apropriate -pthread here and there (see pull request).
Other problem is that it won't start at all:

root@platform-one:/usr/src/pulse# gdb /usr/local/./pulse-1.6.1-cpp
GNU gdb (GDB) 7.4.1-debian
...
Reading symbols from /usr/local/pulse-1.6.1-cpp...done.
(gdb) r
Starting program: /usr/local/pulse-1.6.1-cpp
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/arm-linux-gnueabihf/libthread_db.so.1".
[New Thread 0xb6f342d0 (LWP 19778)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xb6f342d0 (LWP 19778)]
0x00065c44 in execute_native_thread_routine ()
(gdb) bt
#0 0x00065c44 in execute_native_thread_routine ()
#1 0x0008385c in start_thread ()
#2 0x000b5338 in ?? ()
#3 0x000b5338 in ?? ()

Backtrace stopped: previous frame identical to this frame (corrupt stack?)

The main thread seems to be running though:
Thread 1 (Thread 0x135000 (LWP 19775)):
#0 0x00085ed0 in read () at ../sysdeps/unix/syscall-template.S:81
#1 0x0009cbf8 in _IO_new_file_underflow ()
#2 0x0009f19c in _IO_default_uflow ()
#3 0x0009f030 in __uflow ()
#4 0x0009b7f0 in getc ()
#5 0x000450c8 in __gnu_cxx::stdio_sync_filebuf<char, std::char_traits >::underflow() ()
#6 0x00027438 in std::basic_istream<char, std::char_traits >& std::getline<char, std::char_traits, std::allocator >(std::basic_istream<char, std::char_traits >&, std::basic_string<char, std::char_traits, std::allocator >&, char) ()
#7 0x0001a054 in pulse::Pulse::run() () at /usr/include/c++/4.9/bits/basic_string.h:2814
#8 0x00011984 in main () at /usr/src/pulse/src/main/cpp/main.cpp:24

Hello,

I let my engine play a lot of games vs. pulse 1.6 cpp. The games are always long & close, lots of remis (both engines don't care too much about 3-fold repetition).

Though today pulse took quite a big hit by a single move. It might be bad luck beyond the horizon, though problems start at move 8 in following pgn (annotated by lichess).

Greetings from Biel/Bienne,
Bernhard

[Event “Mini Tournament"]
[Date "2020.06.16"]
[Round "3"]
[White "SNAPSHOT (2200)"]
[Black "Pulse-1.6.1-cpp (2200)"]
[Result "1-0"]
[WhiteElo "?"]
[BlackElo "?"]
[Variant "Standard"]
[TimeControl "-"]
[ECO "D21"]
[Opening "Queen's Gambit Accepted: Normal Variation"]
[Termination "Unknown"]
[Annotator "lichess.org"]

1. d4 d5 2. c4 dxc4 3. Nf3 { D21 Queen's Gambit Accepted: Normal Variation } Qd6?! { (0.24 → 0.93) Inaccuracy. Nf6 was best. } (3... Nf6 4. e3 e6 5. Bxc4 c5 6. O-O cxd4 7. exd4 Be7 8. Nc3) 4. e3 b5? { (0.79 → 2.16) Mistake. c5 was best. } (4... c5 5. Na3 e6 6. Nxc4 Qc7 7. Bd2 Nc6 8. Be2 Nf6 9. O-O) 5. Be2?! { (2.16 → 1.19) Inaccuracy. a4 was best. } (5. a4 Qa6 6. b3 e6 7. bxc4 b4 8. c5 Qb7 9. a5 a6 10. Nbd2 Bd7 11. Ne5 Nf6) 5... Nf6?! { (1.19 → 1.93) Inaccuracy. a6 was best. } (5... a6 6. a4 Bb7 7. O-O e6 8. b3 b4 9. Bxc4 c5 10. Bb2 cxd4 11. Nxd4 Nf6 12. Nd2) 6. Nc3?! { (1.93 → 0.85) Inaccuracy. a4 was best. } (6. a4 Qa6 7. O-O e6 8. b3 Be7 9. Ne5 O-O 10. bxc4 b4 11. c5 Qb7 12. a5 c6) 6... b4 7. Na4 Qd5 8. Ne5 Nc6?? { (1.16 → 4.25) Blunder. Nbd7 was best. } (8... Nbd7 9. Nxc4) 9. Bf3 Ne4 10. Nc5 f5 11. Qa4 a5? { (3.82 → 6.84) Mistake. e6 was best. } (11... e6) 12. Nxc6 Nxc5 13. dxc5 Qd7 14. O-O e5 15. Rd1 Bd6 16. cxd6 cxd6 17. Bd5 Bb7 18. b3 h5 19. Nxa5 Qxa4 20. bxa4 Rxa5 21. Bxb7 Kd7 22. Bf3 Rxa4 23. Be2 b3 24. Bb2 Ke7 25. axb3 Rxa1 26. Rxa1 cxb3 27. Ra7+ Kf6 28. Rd7 Ke6 29. Rxg7 Rc8 30. Bxh5 Rb8 31. Bd1 Rb5 32. Ra7 Rb6 33. Ra3 d5 34. g3 Kd6 35. Rxb3 Rxb3 36. Bxb3 e4?! { (66.76 → Mate in 11) Checkmate is now unavoidable. f4 was best. } (36... f4 37. h4) 37. h4 Kc5 38. h5 Kb4 39. Bxd5 { White wins. } 1-0

i copied your newest source code (by git clone) and compiled it (on Linux) .. this gave me a proper binary but the uci command shows v2.0.0 !? I expect v1.7.3 (indeed, your asset file has this version).

Hello

Are there any examples available for this library? like checking if a move is valid or basic setup..

I was testing 1.7.1 and it will always lose on time. This didn't happen with 1.6.1 which worked just fine.
I am using a 1 minute + 1 second time control with Arena.

Hello, this is a really nice engine,
Unfortunately it has a little bug, which happen's regularly against opponent's that ca run out of time and is easy to reproduce:

./pulse
go wtime 100 btime 0
terminate called after throwing an instance of 'std::exception'
  what():  std::exception
Aborted

This is the code in search.cpp on line 199 and 201:

if (whiteTimeLeft < 1) throw std::exception();
if (blackTimeLeft < 1) throw std::exception();

I think, those lines are not necessary and can be removed entirely (or the condition < 0).
Then Pulse would be very stable.

Best Regards,
Oliver

i just let Pulse v1.7.3 (compiled on Linux 64-bit) play with the Princhess engine, latest v0.5.1 asset and dev-version .. in both games Pulse was clearly winning but let the game slip into a draw by 3-fold repetition !? That's a pitty and a huge shortcoming :