the vesrion that I compiled from the source wins the tournament (1 0), no books, no exp files.... 4 rounds
it lost only one game with ShashChess 33.2
All twelve engines are downloaded AVX2 bin files, only the Tactical is compiled. Strange is that LittleBeats, Brainlearn, Raptora, Shash, Raid and BlueMarlin have BENCH tests almost DOUBLE than Tactical..... but they still lose on Tactical with default values.
It's a great result, in any case if you help me with the pgn, what opening book or pgn you used and interface will be very helpful, I will send you the tool to measure the bench.
And thank you for the valuable information about the compilation, the next update of TACTICAL and Cool Iris, will be done with the default commands for you as users.
Best regards.
PD. Give me about three hours to send you the tool, I'm not at home.
wolfie74 wrote: Thu Aug 24, 2023 2:24 pm
WOW I just cannot understand it,
the vesrion that I compiled from the source wins the tournament (1 0), no books, no exp files.... 4 rounds
it lost only one game with ShashChess 33.2
All twelve engines are downloaded AVX2 bin files, only the Tactical is compiled. Strange is that LittleBeats, Brainlearn, Raptora, Shash, Raid and BlueMarlin have BENCH tests almost DOUBLE than Tactical..... but they still lose on Tactical with default values.
wolfie74 wrote: Thu Aug 24, 2023 2:24 pm
WOW I just cannot understand it,
the vesrion that I compiled from the source wins the tournament (1 0), no books, no exp files.... 4 rounds
it lost only one game with ShashChess 33.2
All twelve engines are downloaded AVX2 bin files, only the Tactical is compiled. Strange is that LittleBeats, Brainlearn, Raptora, Shash, Raid and BlueMarlin have BENCH tests almost DOUBLE than Tactical..... but they still lose on Tactical with default values.
The more i go deep in the chess engines the more puzzled I am. I used the build tester tool to measure the strength of the engines, and
Tactical although won the previous tournament with flying colours, is (according to the test tool) weakest engine.
11 tested engines, sorted by strenght:
On the other hand, will it really be necessary to add a "live book"? A good training experience file is enough.
You can train two engines (One for your OTB games and the second experience file for vs engine games). Since I don't have proof on hand of how many users use the "live book" tool, I'm not going to add it.
wolfie74 wrote: Thu Aug 24, 2023 2:24 pm
Strange is that LittleBeats, Brainlearn, Raptora, Shash, Raid and BlueMarlin have BENCH tests almost DOUBLE than Tactical..... but they still lose on Tactical with default values.
Bench does not really matter. Sometimes a tiny change can change the bench a lot. Sometimes it is almost the same even if there are big changes. Latest Stockfish and its derivatives has usually lower bench because of reduced classical evaluation.
I was under impression that NPS and strenght are connected. i.e. higher NPS = higher strenght.
NPS can be measured by BENCH, but how can you measure the strenght? only by tournaments?