Vajolet2 2.8 release

[elcabesa]

With pleasure I'd like to announce the release of Vajolet2 2.8.

Notable features of this release are the ability to play Fisher Random Chess and a totally rewritten syzygy probing code.

The new syzygy code is totally rewritten and integrated in Vajolet2, it take inspiration from Costalba stockfish syzygy code and it's very much faster than my previous integration of fathom library. I hope it has very few bug.

Since a lot of new code is present, please let me know of any problem or nuisance you find

Vajolet can be downloaded on github Release page

From my own testing I hope in an increment of 15 elo points.

Below the long release note from github page:

Code: Select all

New feature:
chess FRC
new syzygy code

bug Fixes:
Multi Pv fix
validation of TTmove in qSearch
PVline follower in qSearch
futility in Pv nodes

Improvements:
changed max mate depth to a more reasonable number (200 ply)
improve endgame scaling coderefactor time managemrn

CI/CD:
change dir structure

[Damir]

Many thanks for the new engine. Can you tell us the estimated elo gain ?

[Paloma]

Look post above, he hope +15 ELO

[elcabesa]

probably 15 Elo points

[kramnik]

Interesting. I was waiting for an update of this program.

There is an arbitrary code on its GitHub repository: https://github.com/elcabesa/vajolet/blo ... gw-w64.bat, this works only on the developer's computer.

[elcabesa]

There is an arbitrary code on its GitHub repository: https://github.com/elcabesa/vajolet/blo ... gw-w64.bat, this works only on the developer's computer.

yes i know.

[LocutusOfPenguin]

trying to compile it with RaspPi.

First: I had to remove
-msse4.2 -m64 -mpopcnt

thats old - in former versions this was a problem too > provide a tartget for Rasp?


but with new version i also get these errors:

Building file: src/command.cpp
Invoking: GCC C++ Compiler
g++ -std=c++1z -O3 -DNDEBUG -pedantic -Wall -Wextra -c -fmessage-length=0 -Isrc -MMD -MP -MF"src/command.d" -MT"src/command.o" -o "src/command.o" "src/command.cpp"
In file included from src/syzygy/tbpairs.h:28:0,
from src/syzygy/tbtable.h:27,
from src/syzygy/tbtableDTZ.h:24,
from src/syzygy/tbtables.h:28,
from src/syzygy/syzygy.h:26,
from src/command.cpp:35:
src/syzygy/sparseEntry.h: In member function ‘uint32_t SparseEntry::getBlock()’:
src/syzygy/sparseEntry.h:30:66: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
uint32_t getBlock() {return *reinterpret_cast<uint32_t *>(_block);}
^
src/syzygy/sparseEntry.h: In member function ‘uint16_t SparseEntry::getOffset()’:
src/syzygy/sparseEntry.h:31:68: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
uint16_t getOffset() {return *reinterpret_cast<uint16_t *>(_offset);}
^
src/command.cpp: In member function ‘void UciManager::impl::_go(std::istringstream&)’:
src/command.cpp:629:42: error: expected ‘)’ before ‘;’ token
if( Move m = _moveFromUci(_pos, token); m != Move::NOMOVE )
^
src/command.cpp:629:42: warning: suggest braces around empty body in an ‘if’ statement [-Wempty-body]
src/command.cpp:629:4: warning: this ‘if’ clause does not guard... [-Wmisleading-indentation]
if( Move m = _moveFromUci(_pos, token); m != Move::NOMOVE )
^~
src/command.cpp:629:44: note: ...this statement, but the latter is misleadingly indented as if it is guarded by the ‘if’
if( Move m = _moveFromUci(_pos, token); m != Move::NOMOVE )
^
src/command.cpp:629:44: error: ‘m’ was not declared in this scope
makefile:146: die Regel für Ziel „src/command.o“ scheiterte
make: *** [src/command.o] Fehler 1

[elcabesa]

I'm compiling it with gcc 8.1 or clang9. I don't know what's the minimum version of gcc you can use but sometime I use modern c++ feature like "init statement inside if". It's standard in c++17 and should be implmented in gcc 7.

[LocutusOfPenguin]

must be gcc7

i tried it with cross-compile on ubuntu (gcc version 7.4.0 (Ubuntu/Linaro 7.4.0-1ubuntu1~18.04.1)
no more compile errors, but the resulting bin file gives me an illegal machine command" (on rasp zero for sure).
Jürgen

[Gabor Szots]

Thanks for the release, Marco.

While it may be a minor issue, I would like to ask you to clarify whether its version is 2.8 or 2.8.0. You announce it here as 2.8 but the engine itself says 2.8.0.