07th, 2017 (start point): In the past I had
different contacts to Fabien. I learned a lot
from him and participate from his comments to
computer chess. A chess
programmer with a clear vision of his own way of
thinking and intentions, creativity. In my
humble opinion one of the programming geniuses
in computer chess history with an impressive
reputation. Sure, that many programmers learned
from Fabien's open sources. So wonderful and
easy can be chess programming ... such comments
we read in the past from amateur & professional
programmers again and again. I must say that I
dealt with a very pleasant, polite person, who
was very friendly, more human and very
professional. To read from him is an event for
me, to add his
work on my page a honour.
Fabien in green ...
Fabien wrote me im
that I should added the flag from Japan to
Senpai on my link selection. So, I am thinking at
this time that a pride programmer honors a pride
country and Japanese people and his mentality. Wilhelm had an issue / theme for
wonderful new logo!
Information to Senpai
will be collected here!
"earlier colleague") and kōhai (後輩,
"later colleague") ...
2018 / March 20th, 2018 / March 22nd, 2018 /
April 03rd, 2018:
After the Senpai FEOBOS analysis I have to test
the new FEOBOS v20.1 Contempt 3
I will test the book with 40
moves in 40 minutes repeatedly tourneys with 22
For sure with Senpai 2.0 BMI2 x64
as one of the 22 participants!
FEOBOS v20.1 with Contempt = 3,
in 40 minutes repeatedly, ponder = off, resign / offer draw = on,
512Mb for hash,
4-pieces endgame bases,
1 core on Intel Core i7-5930k / i7-4770k, 4.3Ghz,
Windows 10 / 7 Prof., 4 rounds = 924 games!
2926 Elo after 84 games.
2916 Elo after 84 games.
2952 Elo after 84 games.
Performance for 252 games = ~ 2928 Elo (Elostat
Statistics and games:
Significanthy lesser fast lost games for
Senpai 2.0 if we compare with
Fabien solved the
biggest criticism with his last changes.
Great are the stats to the perfect running
FEOBOS v20.1 opening book with "Contempt = 3".
Senpai 2.0 BMI2 x64 will be
analyzed now th 41.614 FEOBOS positions with 2
minutes per move and 4 cores. This need two
120 sec. per
position,, 4.096 for hash tables, 4 cores / 4
threads (no HT), Intell Core i7-4770k, 4.0Ghz,
Windows 7 Prof.
A) ~ +85 for Senpai 2.0
C) Senpai a B01 expert?
Since days I try to win with my favorite
Scandinavian line vs. Senpai. It's not possible
for me. Vs. other engines I made some points
here because I know the line and all the
important variants. Playing the line since many
years. Senpai is playing here very positional
1. e4 d5, 2. exd5 Qxd5, 3. Nc3 Qa5, 4. d4 Nf6,
5.Nf3 c6, 6. Bd2 Qc7
In background isn't the eval to see ... much
more easy a "clock" only :-)
possibilites for me as black player with moves
like g6 or later b6 and strong bishops.
Castling to the other site with attacking moves
on white's king size, many chess programs have
Black have many time to make
the position from move to move better.
out vs. Senpai. Senpai is playing allways the
Good news to Senpai from Wolfgang Batting (CEGT
Elo to the predecessor Senpai 1.0!!
A 32bit version can be found in TalkChess under:
by Fabien Letouzey is available!
senpai_20.zip = 1.304Mb (sources, mac &
windows compiles, logo / graphics)
a few years in computer draughts, I am back to
chess for a little while. I see too many draws
at top level, and am afraid that it will be too
late in a couple of years. Big hardware and
large opening books are also not helping IMO.
So I'm here to
announce Senpai 2.0. The code looks all
different because I started from my draughts
program, but it's actually still Senpai inside.
I prefer having a consistent codebase for
multiple games (also Shogi and Othello).
As a rewrite it's
hard to list changes, but here is an attempt:
is baked in, not an add-on; Senpai 2 plays chess
only as a special case!
restricted singular extensions
additional reduction/pruning of "losing" moves
(SEE < 0)
tempo bonus (it seems I forgot in 1.0 ...)
"space" (glorified pawn-chain PST); this changes
playing style a lot
scoring by logistic regression; this gives me
more freedom for eval features
copy/make (customary in games with fewer piece
types than chess)
optional PEXT bitboards (BMI2)
As I recall, no
single change brought more than 10 Elo. So it's
more a sum of small (IMO) things.
This release has
been rushed a little, so it's still lacking UCI
options and multi-PV; sorry about that.
everybody who helped with this release:
- XXX (<-
this line for people I forgot, sorry in
That's all for
It should be clear that Senpai 2.0 will be
added as one of the 10 standard analyze engine for FEOBOS
project. It's with Wasp an engine, available on
my site and the FEOBOS book files should be
optimised for Senpai. The Engine with the hightest draw score will
be killed from the group of 10 FEOBOS standard
engines (at the moment the famos Stockfish).
Stockfish with 15.062% draws ...
I discussed with my FEOBOS partner / Excel
Klaus Wlotzka about it.
Klaus have the same
Senpai is strong in positional and strategic
game. A playing style = must have for FEOBOS.
We will do it end of January 2018.
in beta test phase of Senpai 2.0 the
compatiblity for Shredder Classic GUI.
2.0 BMI2 x64 works fine (*.epd analysis) and is
on start position on FEOBOS runway 21 and clear
Helping persons for Senpai 2.0 compiles:
Listed below persons gave great helps for
computer chess community and are computer chess
Daniel José Queraltó, chess program
Andscacs, windows compiles!
Michael (Mike) Byrne, variety of activities, mac compiles!
attached - macOS_popcnt and macOS_bmi2 .
The bmi2 version was about 5% faster than the
popcnt version on a bmi2 mac ( mostly recent
macs). A pseudo bench on the iMac ( go
depth 20) was about 1.430 Mnps, the popcnt
version on a 2010 Mac Pro 0.990 Mnps.
For SMP :
2010 Mac Pro - 12
real cores w/popcnt, go depth 26
info depth 26
seldepth 39 nodes 375724239 time 43386 nps
2015 iMac I7, 4 real
cores w/ bmi2, go depth 26
info depth 26 seldepth 38 nodes 173250055
time 34896 nps 4964778
Thanks Daniel / Mike!
And thanks again to
Gerd Isenberg and the other authors for
CPW (Chess Programming Wiki)
... my symbolic long time Award for Gerd!
November 07th, 2017 / November 08th,
Wilhelm Hudetz for a new Senpai logo.
started directly the work and sent wonderful
Fabien's art of chess programming should have
a comparable logo with an
Fabien gave directly green
light for the new logo versions!
November 07th, 2017:
If you like you can download here the older
(sources, mac, windows, android & linux