I have noticed that increasing the bench depth produces
progressively smaller and slightly faster executables at
the cost of longer compile times. Also using bench "time"
instead of "depth" seems to produce slightly smaller/faster
executables given comparable compile times.
I have made a new Makefile that generates smaller and
about 1% to 2% faster profile executables at only a
little extra compile time. On my mobile 2GHz i7 a
full profile build time goes from 3'48" to 4'13" and
the exe goes down by 5% from 416,310 bytes to 395,567
bytes.
The rationale behind this is that I've never managed to add a
Queen on 7th rank bonus in DiscoCheck, because it never showed
to be positive (evne slightly) in testing. The only thing that
worked is Rook on 7th rank.
In terms of SF code, it seemed natural to group it with QueenOnPawn
as well as those are done together. I know you're against groupping
in general, but when it comes to non regression test, you are being
more conservative by groupping. If the group passes SPRT(-3,1) it's
safer to commit, than test every component in SPRT(-3,1) and end up
with the risk of commiting several -1 elo regression instead of just
one -1 elo regression.
In chess terms, perhaps it's just easier to manouver a Queen (which
can more also diagonaly) than a Rook. Therefore you can let the search
do its job without needing eval ad-hoc terms to guide it. For the Rook
which takes more moves to manouver such eval terms can be (marginally)
useful.
Marco Costalba [Sun, 30 Mar 2014 08:11:32 +0000 (10:11 +0200)]
Show evaluation from white POV in trace
We chose this instead of negamax sign convention
(ie. from the point of view of the side to move)
because it is more in line to how the eval
table is presented.
Marco Costalba [Sat, 29 Mar 2014 10:12:37 +0000 (11:12 +0100)]
Raise VALUE_INFINITE
In some legal positions like this one:
R6R/3Q4/1Q4Q1/4Q3/2Q4Q/Q4Q2/Np1Q4/kB1N1KB1 b -- 0 1
We can have a very high score, in this case 30177 and 29267
for midgame and endgame respectively, and because
VALUE_INFINITE = 30001 we have an assert in interpolate()
Midgame and endgame scores are stored in 16 bit signed integers
so we can rise VALUE_INFINITE a little bit. This does not fix
the possibility of overflow in general case, just makes the
condition more difficult to trigger and anyhow better uses all
the score width.
Raising VALUE_INFINITE to 32000 seems to fix the problem for this
particular case.
Marco Costalba [Sat, 22 Mar 2014 10:15:28 +0000 (11:15 +0100)]
Fix a subtle bug in UCI options printing
We want all the UCI options are printed in the order in which are
assigned, so we use an index that, depending on Options.size(),
increases after each option is added to the map. The problem is
that, for instance, in the first assignment:
o["Write Debug Log"] = Option(false, on_logger);
Options.size() can value 0 or 1 according if the l-value (that
increments the size) has been evaluated after or before the
r-value (that uses the size value).
The culprit is that assignment operator in C++ is not a
sequence point:
http://en.wikipedia.org/wiki/Sequence_point
(Note: to be nitpick here we actually use std::map::operator=()
that being a function can evaluate its arguments in any order)
So there is no guarantee on what term is evaluated first and
behavior is undefined by standard in this case. The net result
is that in case r-value is evaluated after l-value the last
idx is not size() - 1, but size() and in the printing loop
we miss the last option!
Bug was there since ages but only recently has been exposed by
the removal of UCI_Analyze option so that the last one becomes
UCI_Chess960 and when it is missing engine cannot play anymore
Chess960.
The fix is trivial (although a bit hacky): just increase the
last loop index.
Reported by Eric Mullins that found it on an ARM and MIPS
platforms with gcc 4.7
Marco Costalba [Sat, 15 Mar 2014 22:43:35 +0000 (23:43 +0100)]
Split also if no slaves are found
Because we test for available slaves before
entering split(), we almost always allocate a
slave, only in the rare case of a race (less
then 2% of cases) this is not true, but to
special case this occurrence is not worth
the added complexity.
Mysseno [Fri, 14 Mar 2014 10:00:50 +0000 (21:00 +1100)]
Depth dependant aspiration window delta
Split delta value in aspiration window so that when
search depth is less than 24 a smaller delta value
is used. The idea is that the search is likely to
be more accurate at lower depths and so we can exclude
more possibilities, 25% to be exact.
Marco Costalba [Fri, 14 Mar 2014 08:57:34 +0000 (09:57 +0100)]
Use legal positions for endgame initialization
During endgame initialization we get the material
hash key of each endgame forging and ad-hoc position
that in same cases is illegal (leaves teh king under
capture). This is not a problem for the material key,
but rises an assert when SF is run in debug mode with
'testKingCapture' set in pos_is_ok().
So rewrite the code to always produce legal positions.
Marco Costalba [Mon, 10 Mar 2014 07:38:23 +0000 (08:38 +0100)]
Simplify pseudo_legal()
Big simplification of pawn move check.
Code has been tested with a brute force approach: for
every position reached during a bench search, the function
has been called for each combinations of Move(from, to)
and verified the result is the same of old code.
Actually this function is very critical becuase is the
one that ensures corrupted TT moves are discarded, so
to properly test it a simple bench is not enough.
Marco Costalba [Sun, 16 Feb 2014 11:20:37 +0000 (12:20 +0100)]
Increase MAX_PLY from 100 to 120
Under some very rare case 100 plies of search
could be not enough. Increasing more could lead
to crashes due to reached stack size limit on
some platforms.
Marco Costalba [Sun, 16 Feb 2014 10:37:29 +0000 (11:37 +0100)]
Fix material key for King
Currently king has no material key associated because
it can never happen to find a legal position without
both kings, so there is no need to keep track of it.
The consequence is that a position with only the two
kings has material key set at zero and if the material
hash table is empty any entry will match and this is
wrong.
Normally bug is hidden becuase the checking for a draw
with pos.is_draw() is done earlier than evaluate() call,
so that we never check in gameplay the material key of a
position with two kings.
Nevertheless the bug is there and can be reproduced setting
at startup a position with only two kings and typing
'eval' from prompt.
The fix is very simple: add a random key also for the king.
Also fixed the condition in material.cpp to avoid asserting
when a 'just 2 kings' postion is evaluated.
Marco Costalba [Sat, 15 Feb 2014 21:17:58 +0000 (22:17 +0100)]
Restore PorbCut name
Actually MultiCut is too different from current scheme.
Note that neither ProbCut is exactly what we do because
we try just a handful of captures instead of all moves,
nevertheless it seems more in line with what we do.
Joerg Oster [Thu, 13 Feb 2014 22:44:12 +0000 (23:44 +0100)]
Return static eval when reaching MAX_PLY
Makes more sense than returning a draw score. Tested
with reduced MAX_PLY = 30 and passed both short TC
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 17434 W: 3345 L: 3194 D: 10895
And long TC
LLR: 2.97 (-2.94,2.94) [0.00,6.00]
Total: 2610 W: 488 L: 373 D: 1749
With current limit of MAX_PLY = 100 the patch should not
introduce any measurable change, nevertheless is the correct
approach.
Idea of returning eval is from Michel Van den Bergh.
Marco Costalba [Fri, 14 Feb 2014 09:43:37 +0000 (10:43 +0100)]
Fix magic boosters conversion
Fix small overflow error while converting
magic boosters from right rotate to left rotate,
in particular booster 38 was converted to 4122
instead of the corrcet value 26.
Marco Costalba [Wed, 12 Feb 2014 13:47:36 +0000 (14:47 +0100)]
Move magic random to RKISS
When initializing the magic numbers used to
compute sliding attacks, we endless generate a
random and test it as a possible magic.
In the general case this takes a lot of iterations,
but here, insteaad of picking a casual random, we
rotate it a couple of times and generate a number that
we know has a good probability to be a magic candidate.
This is becuase the quantities by which we rotate the
number are known in advance to produce quickly a good
canidate.
The patch, inspired by DON, just moves the shuffle to RKISS
changing the boosters to take in account a left rotation
instead of a right rotation as in the original.
Marco Costalba [Sat, 8 Feb 2014 12:07:57 +0000 (13:07 +0100)]
Don't fear races when are harmless
Actually race conditions do exist in an engine, just
think for a moment to TT concurrent access. Racy code
is not a problem per se, if the consequences are well
known and correctly handled.
In case of TT access we ensure that the TT move is validated
before to be tried, here we just retry the same move in less
that 1 case out of a million: this is totally harmless considering
that very probably the second time the move is tried we get
immediately a TT hit and search quickly returns.
Lucas Braesch [Tue, 4 Feb 2014 07:18:19 +0000 (08:18 +0100)]
Better document null search window
Hopefully this patch makes the code more:
* Self-documenting: Null search is always a zero window search,
because it is testing for a fail high. It should never be done
on a full window! The current code only works because we don't
do it at PV nodes, and therefore (alpha, beta) = (beta-1, beta):
that's the kind of "clever" trick we should avoid.
* Idiot-proof: If we want to enable null search at PV nodes, all we
need to do now is comment out the !PvNode condition. It's that simple!
In theory, null search should not be done at PV nodes, because PV nodes
should never fail high. But in practice, they DO fail high, because of
aspiration windows, and search inconsistencies, for example. So it makes
sense to keep that flexibility in the code.
Lucas Braesch [Mon, 3 Feb 2014 01:41:32 +0000 (09:41 +0800)]
Better document razoring
Use ralpha instead of rbeta
* rbeta is confusing people. It took THREE attempts to code razoring
at PV nodes correctly in a recent test, because of the rbeta trick.
Unnecessary tricks should be avoided.
* The more correct and self-documenting way of doing this, is to say
that we use a zero window around alpha-margin, not beta-margin.
The fact that, because we only do it at PV nodes, alpha happens to be
beta-1 and that the current stuff with rbeta works, may be correct,
but is confusing.
Remove the misleading and partially erroneous comment about returning
v + margin:
* comments should explain what the code does, not what it could have done.
* this comment is partially wrong in saying that v+margin is "logical",
and that it is "surprising" that is doesn't work.
From a theoretical perspective, at least 3 ways of doing this are equally
defendable:
1/ fail hard: return alpha: The most conservative. We bet that the search
will fail low, but we don't know by how much and don't want to take risks.
2/ aggressive fail soft: return v (what the current code does). This
corresponds to normal fail soft, with the added assumption that we don't
care about the reduction effect (see below point 3/)
3/ conservative fail soft: return v + margin. If the reduced search (qsearch)
gives us a score <= v, we bet that the non reduced search will give us a
score <= v + margin.
* Saying that 2/ is "logical" implies that 1/ and 3/ are not, which is
arguably wrong. Besides, experimental results tell us that 2/ beats 3/,
and that's not something we can argue against: experimental results are
the only trusted metric.
* Also, with the benefit of hindsight, I don't think the fact that 2/ is
better than 3/ is surprising at all. The point is that it is YOUR turn to
move, and you are assuming that by NOT playing (and letting the opponent
capture your hanging pieces in QS) you cannot generally GAIN razor_margin(depth).
Uri Blass [Mon, 27 Jan 2014 19:08:31 +0000 (20:08 +0100)]
Reduce VALUE_KNOWN_WIN to 10000
With some positions like
8/8/8/2p2K2/1pp5/br1p1b2/2p2r2/qqkqq3 w - -
The eval score is higher than VALUE_INFINITE because
is the sum of VALUE_KNOWN_WIN plus a big material
advantage. This leads to an assert. Here are the
steps to reproduce:
Compile SF with debug=yes then do
./stockfish
position fen 8/8/8/2p2K2/1pp5/br1p1b2/2p2r2/qqkqq3 w - -
go depth 1
This patch fixes the issue in this case, but do exsist
other positions for which the patch is not enough and
we will need to limit the eval score to be sure not
overflow the limit.
Note that is not possible to increase the value of
VALUE_INFINITE because should remain within int16_t
type to be stored in a TT entry.
Instead of a fixed reduction of ONE_PLY, now
Null move dynamic reduction based on value can
grow larger in case we are above beta of a value
much higher then PawnValueMg.
Note that now an eval returning VALUE_KNOWN_WIN
makes null search to drop in qsearch.
Passed both short TC:
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 26141 W: 4871 L: 4699 D: 16571
And long TC:
LLR: 2.97 (-2.94,2.94) [0.00,6.00]
Total: 33695 W: 5309 L: 5056 D: 23330
Marco Costalba [Sat, 18 Jan 2014 17:19:16 +0000 (18:19 +0100)]
Increase max hash size to 16GB
TCEC season 3, which is due to start in a few weeks, just
had its server upgraded to 64GB RAM and will therefore allow
16GB hash to be used per engine.
This is almost the upper limit without changing the
type of size and hashMask. After this we need to
move to uint64_t instead of uint32_t.
Chris Cain [Thu, 16 Jan 2014 21:50:08 +0000 (21:50 +0000)]
Simplify pawnless endgame evaluation
Retire KmmKm evaluation function. Instead give a very drawish
scale factor when the material advantage is small and not much
material remains.
Retire NoPawnsSF array. Pawnless endgames without a bishop will
now be scored higher. Pawnless endgames with a bishop pair will
be scored lower. The effect of this is hopefully small.
Consistent results both at short TC (fixed games):
ELO: -0.00 +-2.1 (95%) LOS: 50.0%
Total: 40000 W: 7405 L: 7405 D: 25190
And long TC (fixed games):
ELO: 0.77 +-1.9 (95%) LOS: 78.7%
Total: 39690 W: 6179 L: 6091 D: 27420
When we have a fail-high of a quiet move, store it in
a Followupmoves table indexed by the previous move of
the same color (instead of immediate previous move as
is in countermoves case).
Then use this table for quiet moves ordering in the same
way we are already doing with countermoves.
These followup moves will be tried just after countermoves
and before remaining quiet moves.
Passed both short TC
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 10350 W: 1998 L: 1866 D: 6486
And long TC
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 14066 W: 2303 L: 2137 D: 9626
Marco Costalba [Sun, 12 Jan 2014 21:32:41 +0000 (22:32 +0100)]
Retire KBBKN endgame
As pointed out by Joona, Lucas and otehr people in
the forum, this endgame is not a known, there are many
positions where it takes more than 50 moves to claim the
win and becasue exact rules is not possible better to
retire and allow the search to workout the endgame for us.