🚀 go-pugleaf

Thanks Kong Sian, for the info on the DEEPNULL defines for pl68!
The changes were at line 1144, 1304 and 1314 in the copy of search.c I got
from prep.ai.mit.edu.

I implemented the changes to the Makefile you suggest.  Here are the vitals:
OPT=  -DUSEINT -DCACHE -DGDX -DHASGETTIMEOFDAY -DKILLT -DDEEPNULL
CC=	gcc -W  $(OPT)
CFLAGS=  -O2 -funroll-loops -traditional-cpp  # gnu cc  2.00 on SunOS
(Should NULLMOVE be defined also??)

This configuration gave the about 5.5 minutes per position with depth
set to ply 6, as compared with about 4.5 minutes with earlier patch levels,
and with 16 min for pl68 before making these changes.

I am posting a position that causes problems for gnuchess 4.0 pl68.
It shows there is some bug remaining in the code I have.
The position plays havoc with the alpha/beta window, but that's not all...
It evaluates white's drawing move Qf7+ as zero at each ply. (OK)
It gives it up at each ply in favor of the move Nxb6, which produces a
score of about 2 pawns. (OK -- except I don't see how it really merits
the evaluation, but that's another issue :=)
Finally, gnuchess lists Nxb6 as its final choice, skips writing any
stats lines, and claims the draw it could only get if it played Qf7+ 
instead! (NOT OK!)

Here's the line leading up to the position:
   1. e4 c6 2. d4 d5 3. exd5 cxd5 4. Bd3 e6
{I can't find any theory on black's fourth move.}
   5. Nf3 Nf6 6. c4 Be7 7. Nc3
{This sequence reaches a line of the Panov 4.c4 Nf6 5.Nc3 e6 6.Nf3 Be7 7.Bd3
where white loses a tempo to 7...dxc4 8.Bxc4 O-O 9.O-O, according to Varnusz.
In this line, white has several improvements to move 7, so something is
suspect in the continuation}
   7...O-O 8. c5
{Replacing 7.Bd3 in the Panov line above with 7.c5 O-O 8.Bd3 reproduces the
current position, as in ECO B14/13}
   8...b6 9. b4 a5 10. Na4 Nfd7 11. h4
{Now, follow B14/13 in footnote #72}
   11...h6 12. Rh3 e5 13. Bxh6 Bf6 14. Be3 exd4 15. Bh7+
{The suggested line 15...Kh8 16.Ng5 g6 17.Bxg6 fxg6 18.h5 is given +- in ECO,
and if black takes the bishop on move 15, the attack on the king side is
quite powerful.  Is it enough to win or is it only a draw? (I don't know!)}
   15...Kxh7 16. Ng5+ Kg8 17. Qh5 Re8 18. Qxf7+
{First instance of this position}
   18...Kh8 19. Qh5+ Kg8 20. Qf7+
{Second instance}
   20...Kh8 21. Qh5+ Kg8

8 |r|n|b|q|r|/|k|/|
7 |/| |/|n|/| |p| |
6 | |p| |/| |b| |/|
5 |p| |P|p|/| |N|Q|
4 |N|P| |p| |/| |P|
3 |/| |/| |B| |/|R|
2 |P|/| |/| |P|P|/|
1 |R| |/| |K| |/| |
   a b c d e f g h

{and} 22.Qf7+ {draws on the third instance of position}

Here's what to tell gnuchess to duplicate my result:
----------------------------------
% gnuchessr
1 1
post
force
e2e4 c7c6 d2d4 d7d5 e4d5 c6d5 f1d3 e7e6 g1f3 g8f6 c2c4 f8e7 b1c3 e8g8 c4c5
b7b6 b2b4 a7a5 c3a4 f6d7 h2h4 h7h6 h1h3 e6e5 c1h6 e7f6 h6e3 e5d4 d3h7 g8h7
f3g5 h7g8 d1h5 f8e8 h5f7 g8h8 f7h5 h8g8 h5f7 g8h8 f7h5 h8g8
depth
6
board
go
-----------------------------------
Gnuchess output:

Enter #moves #minutes: 
White 10:00  Black 10:00
<empty board deleted>
Book used 0 of 250000 entries.
Warning can't find book.
Playing without hashfile

Your move is? 
<rest deleted>
Your move is? depth= 
Your move is? 

White 209:59  Black 209:59

 R N B Q R - K -
 - - - N - - P -
 - P - - - B - -
 P - p P - - n q
 n p - P - - - p
 - - - - b - - r
 p - - - - p p -
 r - - - k - - -


Your move is? 
Move# 22    Target= 1319839    Clock: 1319939
 2.    135    2    11558   b4b5 
 3-   -144    2    11652   b4b5  d4e3  h3e3 
 3-      0    3    13317   h5f7  g8h8  f7h5  h8g8  h5f7  g8h8  a4b6 
 3+    278    3    13318   a4b6 
 3&   -113   14    71498   a4b6  d7b6  h5f7  g8h8  f7h5  h8g8 
 3&      0   14    74365   h5f7  g8h8  f7h5  h8g8  h5f7  g8h8  a4b6 
 3&    126   14    74367   b4b5 
 3.    126   17    92663   b4b5 
 4-     -8   26   140849   b4b5  f6g5  a4b6 
 4-      0   27   151506   h5f7  g8h8  f7h5  h8g8  h5f7  g8h8  f7d5  e8e3 
                           f2e3 
 4+    279   27   151507   a4b6 
 4&   -113  108   605834   a4b6  d7b6  h5f7  g8h8  f7h5  h8g8  h5f7 
 4&      0  113   639982   h5f7  g8h8  f7h5  h8g8  h5f7  g8h8  a1d1 
 4&    127  113   639984   b4b5 
 4.    127  127   729889   b4b5 
 5-     -9  129   738816   b4b5  f6g5  a4b6 
 5-      0  135   777495   h5f7  g8h8  g5e4 
 5+    280  135   777496   a4b6 
 5&   -113  227  1283300   a4b6  d7b6  h5f7  g8h8  f7h5  h8g8  h5f7 
 5&      0  280  1394253   h5f7  g8h8  f7h5  h8g8  h5f7  g8h8  f7h5  h8g8 
                           h5f7  g8h8  f7h5  h8g8 
 5&    138  280  1394255   b4b5 
 5.    138  303  1538933   b4b5 
 6-      1  307  1563486   b4b5  f6g5  h5g4 
 6-      3  532  2634551   b4a5  f6g5  h3f3 
 6-      9  559  2775511   a1d1 
 6-      9  709  3221743   h5e2 
 6-      9  796  3431541   g5f3 
 6&   -621 2464 12674691   g5f3  d7e5  f3g5  d4e3  h5h7  g8f8  h3e3  e5d3 
                           h7d3  e8e3  f2e3  f6a1  c5b6 
 6&   -222 6557 36214591   a1d1  d7e5  a4b6  f6g5  b6a8  g5e3  f2e3  c8h3 
                           g2h3  d4e3  d1d5 
 6&      0 7669 42945940   h5f7  g8h8  f7h5  h8g8  g5f7  e8e3  f2e3  b6b5 
 6&    281 7669 42945942   a4b6 
 6.    281 7682 43021602   a4b6 
1. ... a4b6
Drawn game!
----------------------------------------
The "Drawn game!" came from gnuchess.  Where's the stats line?

Mike