the order of elements returned by std::partition is implementation defined (since not stable) and could depend on the version of libstdc++ linked.
As std::stable_partition was tested to be too slow (http://tests.stockfishchess.org/tests/view/585cdfd00ebc5903140c6082).
Instead combine partition with our custom implementation of insert_sort, which fixes this issue.
Implementation based on a patch by mstembera (http://tests.stockfishchess.org/tests/view/58d4d3460ebc59035df3315c), which suggests some benefit by itself.
Higher depth moves are all sorted (INT_MIN version), as in current master.
It is worth noting that an attempt to only increase the bonus passed STC but failed LTC, and
an attempt to remove the cap without increasing the bonus is still running at STC, but will probably fail after more than 100k.
By adding pos.non_pawn_material(pos.side_to_move()) as a precondition in step 13,
which is already in use in Futility Pruning (child node) and Null Move Pruning for similar reasons.
Pawn endgames, especially those with only 1 or 2 pawns, are simply heavily influenced by zugzwang situations.
Since we are using a bitbase for KPK endgames, I see no reason to accept buggy evals as shown in #760
a single Xeon Phi can present itself as a single numa node with up to 288 threads (4 threads per hardware core).
Tested to work as expected with a Xeon Phi CPU 7230 up to 256 threads.
joergoster [Fri, 17 Mar 2017 21:41:08 +0000 (14:41 -0700)]
Pawns count imbalance table
Instead of having a continuous increasing bonus for our number of pawns when calculating imbalance, use a separate lookup array with tuned values.
Idea by GuardianRM.
GuardianRM [Fri, 24 Feb 2017 05:26:59 +0000 (21:26 -0800)]
Pieces protecting king
Initial protective idea by Snicolet for knight, for other pieces too
Patch add penalties and bonuses for pieces, depending on the distance from the own king
snicolet [Sun, 19 Feb 2017 22:25:05 +0000 (14:25 -0800)]
Keep pawns on both flanks
Positions with pawns on only one flank tend to be more drawish. We add
a term to the initiative bonus to help the attacking player keep pawns
on both flanks.
STC: yellowish run stopped after 257137 games
LLR: -0.92 (-2.94,2.94) [0.00,5.00]
Total: 257137 W: 46560 L: 45511 D: 165066
FauziAkram [Sun, 19 Feb 2017 21:56:17 +0000 (13:56 -0800)]
Variable tuning
A tuning patch which cover the following changes:
increase the importance of queen and rook mobility in endgame and
decrease it in mg, since if we use the heavy pieces too early in the game
we will just make opponent develop their pieces by threatening ours.
King Psqt:
1)King will be encouraged more to stay in the first ranks in the MG
2)and will be encouraged more to go to the middle of the board/last ranks in the EG
Bishop scale better in EG
Logical changes on various psqt tables
1/6 of the changes of the last tuning session on mobility tables
VoyagerOne [Wed, 25 Jan 2017 14:32:10 +0000 (09:32 -0500)]
Penalty for a quiet ttMove that fails low
Also the penalty/bonus function is misleading, we
should simply change it to stat_bonus(depth) for
bonus and -stat_bonus(depth+ ONE_PLY) for extra
penalty.
This submitted version (using if (abs(mg + eg) > 1500) )
seems more logical than the following other green simplification (using if (abs(mg)>1500))
since it can happen than mg_value is > eg_value (about 20% of the time)
and the submitted version seems stronger at LTC
After we have taken into account all cheap evaluation
terms, we check whether the score exceeds a given threshold.
If this is the case, we return a scaled down evaluation.
Stéphane Nicolet [Wed, 11 Jan 2017 17:11:17 +0000 (18:11 +0100)]
Check for overflow in Score * int multiplication (#969)
Add asserts to check for overflow in Score * int multiplication.
There is no overflow in current master, but it would be easy to
create one as the scale of the current eval does not leave many
spare bits. For instance, adding the following unused variables
in master at the end of evaluate() (line 882 of evaluate.cpp)
overflows:
New shell scripts for testing, used for travis CI (#957)
Perform more complex verification and validation.
- signature.sh : extract and optionally compare Bench/Signature/Node count.
- perft.sh : verify perft counts for a number of positions.
- instrumented.sh : run a few commands or uci sequences through valgrind/sanitizer instrumented binaries.
- reprosearch.sh : verify reproducibility of search.
These script can be used from directly from the command line in the src directory.
Marco Costalba [Sat, 7 Jan 2017 09:13:49 +0000 (10:13 +0100)]
Drop Stats c'tors
Now taht we correctly value-initialize Thread objects,
we don't need c'tors anymore because tables will be
zero-initialized by the compier when Thread object
is instanced.
It can be used uninitialized in time management.
Fixes all valgrind errors on './stockfish go wtime 8000 btime 8000 winc 500 binc 500'
This is one (of the many) quirks of C++. There is a subtle difference between:
new Foo
new Foo()
The first statement calls the default constructor (that in case of a POD leaves data members
uninitialized), the second one performs a value-initialization (that in case of POD is
equivalent to a zero-initialization)
Aram Tumanian [Wed, 21 Dec 2016 11:40:55 +0000 (13:40 +0200)]
Don't clear EasyMove in search()
EasyMove is cleared after every iteration of the
search if the 3rd move in the PV of the main thread
changes from the previous iteration. Therefore,
clearing EasyMove during a search iteration may be
excessive. The tests show that this is indeed unnecessary.
In the new version the EasyMove variable is used only in
the Thread::search function.
Sergei Antonov [Mon, 12 Dec 2016 15:04:16 +0000 (16:04 +0100)]
Threefold repetition detection
Implement a threefold repetition detection. Below are the examples of
problems fixed by this change.
Loosing move in a drawn position.
position fen 8/k7/3p4/p2P1p2/P2P1P2/8/8/K7 w - - 0 1 moves a1a2 a7a8 a2a1
The old code suggested a loosing move "bestmove a8a7", the new code suggests "bestmove a8b7" leading to a draw.
Incorrect evaluation (happened in a real game in TCEC Season 9).
position fen 4rbkr/1q3pp1/b3pn2/7p/1pN5/1P1BBP1P/P1R2QP1/3R2K1 w - - 5 31 moves e3d4 h8h6 d4e3
The old code evaluated it as "cp 0", the new code evaluation is around "cp -50" which is adequate.
Patch has been rewritten into current form for simplification and
logic slightly changed so that return a draw score if the position
repeats once earlier but after or at the root, or repeats twice
strictly before the root. In its original form, repetition at root
was not returned as an immediate draw.
After retestimng testing both version with SPRT[-3, 1], both passed
succesfully, but this version was chosen becuase more natural. There is
an argument about MultiPV in which an extended draw at root may be sensible.
See discussion here:
Alain SAVARD [Wed, 28 Dec 2016 22:14:09 +0000 (17:14 -0500)]
Small eval cleanup and renaming
Non-functional changes
a) splitting the threat array to avoid using an enum
b) reorder the scores according to functions where they are used.
c) declarations in evaluate_pieces after the const(s) like elsewhere
d) more compact definitions of KingFlank,
now that we need it also for the PanwLessFlank penalty.
e) reuse CenterFiles in evaluate_space
f) move one line inside next popcount
Explicitly use alpha+1 for beta in NonPV search (#939)
Fixes the only exception, in razoring.
The code already does assert(PvNode || (alpha == beta - 1)), and it can be verified by studying the program flow that this is indeed the case, also for the modified line.
This patch tweaks some pawn values to favor flank attacks.
The first part of the patch increases the midgame psqt values of external pawns to launch more attacks (credits to user GuardianRM for this idea), while the second part increases the endgame connection values for pawns on upper ranks.
Andrey Neporada [Sat, 3 Dec 2016 08:37:07 +0000 (12:37 +0400)]
Help GCC to optimize msb() to single instruction
GCC compiles builtin_clzll to “63 ^ BSR”. BSR is processor instruction "Bit Scan Reverse".
So old msb() function is basically 63 - 63 ^ BSR.
Unfortunately, GCC fails to simplify this expression.
Marco Costalba [Fri, 25 Nov 2016 15:51:24 +0000 (16:51 +0100)]
Fix compile under Windows XP
The needed Windows API for processor groups could be missed from old Windows
versions, so instead of calling them directly (forcing the linker to resolve
the calls at compile time), try to load them at runtime. To do this we need
first to define the corresponding function pointers.
Also don't interfere with running fishtest on numa hardware with Windows.
Avoid all stockfish one-threaded processes will run on the same node
Aram Tumanian [Mon, 21 Nov 2016 15:17:47 +0000 (17:17 +0200)]
Fix the pawn hash failure when the pawn key is 0
This patch fixed bugs #859 and #882.
At initialization we generate a new random key (Zobrist::noPawns).
It's added to the pawn key of all positions, so that the pawn key
of a pawnless position is no longer 0.
Marco Costalba [Tue, 22 Nov 2016 06:41:46 +0000 (07:41 +0100)]
Handle Windows Processors Groups
Under Windows it is not possible for a process to run on more than one
logical processor group. This usually means to be limited to use max 64
cores. To overcome this, some special platform specific API should be
called to set group affinity for each thread. Original code from Texel by
Peter Österlund.
Tested by Jean-Paul Vael on a Xeon E7-8890 v4 with 88 threads and confimed
speed up between 44 and 88 threads is about 30%, as expected.
projects / stockfish / log