Windows thread question

[bob]

looking at the .s output, there is no difference I can see in the two. Not sure where the (dummy) came from, but not from me I don't think...

[Tord Romstad]

Hi Jim!

Latest Glaurung development version works fine compiled with Cygwin.

For all you testers who are having trouble compiling a working Win32 binary.

http://www.verzend.be/v/1098236/glaurun ... 2.zip.html

Does it work even when compiled with MaxNumOfThreads = 4? If it does, this is great news! I would greatly appreciate if someone with a quad-CPU machine running Windows could verify that Jim's binary works.

Tord

[Jim Ablett]

Hi Tord,

Yes, it was compiled with MaxNumOfThreads = 4.
Works fine on my single cpu.

Regards,
Jim.

[Jim Ablett]

The old download link may be unavailable.

I've uploaded a fresh new Win32 cygwin binary here (compiled with MaxNumOfThreads = 4 again - working fine 1cpu).

http://www.orbitfiles.com/download/id271127127

Jim.

[Tord Romstad]

Hi Tord,

Yes, it was compiled with MaxNumOfThreads = 4.
Works fine on my single cpu.

Thanks, sounds good! I assume that the cygwin binary uses POSIX threads rather than Windows threads, and that this may have some impact on performance. Can someone with a dual please test how Jim's binary compares to Fritz Reul's binary when using one and two threads? Glaurung 060612 contains a new benchmark feature, which is used as follows (this is the Unix command line, I don't know exactly how it would look in Windows):

Code: Select all

./glaurung bench 30 2 128

The first number, 30, is the number of seconds Glaurung should spend on each position in the benchmark (there are 10 positions). The second number, 2, is the number of threads Glaurung should use. The third number, 128, is the hash table size.

It would be very nice if someone could do "./glaurung bench 30 1 128" and "./glaurung bench 30 2 128" with Jim's and Fritz's binaries and compare the results. If somebody happens to own a quad, it would also be interesting to see the results of "./glaurung bench 30 4 128" with Jim's binary (Fritz's binary is limited to a maximum of two threads).

Tord

[Jim Ablett]

Can someone with a dual please test how Jim's binary compares to Fritz Reul's binary

On a single cpu my build is a bit quicker than Fritz's. >

Fritz Ruel's:

fen = 3rrk2/1bnqppb1/p2p2p1/1p1P2P1/1p2PP1R/2NQBN2/1PP2K2/7R w - - bm Rh8+
2 1.41 0:00 529 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6
3 0.19 0:01 1311 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6 gxf6+ Kxh8 fxe7+ Kh7
exd8=Q Qxd8 Qxb4
3 0.83 0:03 1898 Bd4 f6 gxf6 exf6
4 0.52 0:04 2703 Bd4 e5 Rh8+ Bxh8 Rxh8+ Ke7
5 0.48 0:06 7563 Bd4 f6 gxf6 exf6 Ne2
6 0.42 0:08 16942 Bd4 f6 Ne2 a5 Rh7 Qg4
7 0.53 0:10 71531 Bd4 f6 Ne2 a5 Ng3 e5 gxf6 Bxf6
7 3.95 0:14 94720 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2
8 3.80 0:18 116506 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5
9 3.72 0:24 185032 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5 Ke3
328941 nodes total, 10964 nodes/second

fen = 2rq1rk1/pp1bnpbp/4p1p1/3pP1N1/3P2Q1/2PB4/P4PPP/R1B1R1K1 w - - bm Nxh7
2 0.70 0:00 329 Qh4 h5
3 0.38 0:01 565 Qh4 h5 Bd2
3 0.69 0:02 668 Qh3 h5 Rb1
4 -0.52 0:03 2510 Qh3 h6 Nf3 Rxc3
4 0.48 0:05 3151 Bd2 h6 Nf3 Nf5 Bxf5 gxf5
5 0.48 0:06 7469 Bd2 h6 Nf3 Nf5 Bxf5 gxf5
6 0.53 0:08 19058 Bd2 h6 Nh3 Nf5 Rab1 h5
7 0.27 0:10 38570 Bd2 h6 Nh3 Nf5 Nf4 Ba4 Bxf5 gxf5
8 0.66 0:22 117031 Bd2 h6 Nh3 Ba4 Rab1 Qd7 Qh4 h5
9 0.44 0:28 170987 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qd7 Qh4
180000 nodes total, 6000 nodes/second

Mine:

fen = 3rrk2/1bnqppb1/p2p2p1/1p1P2P1/1p2PP1R/2NQBN2/1PP2K2/7R w - - bm Rh8+
2 1.34 0:00 538 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6
3 0.08 0:00 1322 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6 gxf6+ Kxh8 fxe7+ Kh7
exd8=Q Qxd8 Qxb4
3 0.77 0:00 1924 Bd4 f6 Ne2
4 0.39 0:00 2939 Bd4 e5 Rh8+ Bxh8 Rxh8+ Ke7
5 0.73 0:00 8450 Bd4 f6 Ne2 a5 Rh7
6 0.50 0:01 15852 Bd4 f6 Ne2 a5 Ng3 e5
7 0.47 0:02 56713 Bd4 f6 Ne2 a5 Ng3 e5 gxf6 Bxf6
7 3.84 0:05 80037 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2
8 3.75 0:07 100903 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5
9 3.67 0:11 166508 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5 Ke3
10 4.23 0:17 362474 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ng5 bxc3 Qh3 Rg7 Bxg7 cxb2 Bxb2
442103 nodes total, 14736 nodes/second

fen = 2rq1rk1/pp1bnpbp/4p1p1/3pP1N1/3P2Q1/2PB4/P4PPP/R1B1R1K1 w - - bm Nxh7
2 0.70 0:00 324 Qh4 h5
3 0.38 0:00 564 Qh4 h5 Bd2
3 0.64 0:00 666 Qh3 h5 Rb1 Rxc3 Rxb7
4 -0.45 0:00 2458 Qh3 h6 Nf3 Rxc3
4 0.53 0:00 2915 Bd2 Nf5 Rab1 Qc7
5 0.42 0:00 6527 Bd2 h6 Nf3 h5 Qh4
6 0.53 0:01 14301 Bd2 h6 Nh3 Nf5 Rab1 h5
7 0.34 0:02 32598 Bd2 h6 Nf3 h5 Qg3 Nf5 Bxf5 gxf5
7 0.36 0:08 74735 Qd1 Nf5 g4 Ne7 Rb1 Rxc3 Rxb7
8 -0.25 0:12 115309 Qd1 Rxc3 Rb1 Rc7 Ba3 h6 Nf3 h5
8 0.53 0:13 127813 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qc7
9 0.44 0:17 166466 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qc7 Bg5
10 0.38 0:26 275808 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Qg3 Nf5 Bxf5 gxf5
360000 nodes total, 12000 nodes/second

Jim.

[Jim Ablett]

On a single cpu 32 bit, 2 threads (bench) by build is faster than Bryan's too on my computer >

Bryan Hoffman:

fen = 3rrk2/1bnqppb1/p2p2p1/1p1P2P1/1p2PP1R/2NQBN2/1PP2K2/7R w - - bm Rh8+
2 1.41 0:00 530 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6
3 0.19 0:01 1354 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6 gxf6+ Kxh8 fxe7+ Kh7
exd8=Q Qxd8 Qxb4
3 0.88 0:03 1940 Bd4 f6 gxf6 exf6
4 0.52 0:04 2738 Bd4 e5 Rh8+ Bxh8 Rxh8+ Ke7
5 0.53 0:06 7884 Bd4 f6 gxf6 exf6 Ne2
6 0.42 0:08 17362 Bd4 f6 Ne2 a5 Rh7 Qg4
7 0.53 0:10 88308 Bd4 f6 Ne2 a5 Ng3 e5 gxf6 Bxf6
7 1.67 0:14 117097 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ Kxh8 f7+ Kh7 Qd1
Qg4 fxe8=Q Rxe8 Ng5+ Qxg5 fxg5 bxc3 Bxc3
8 3.83 0:19 161885 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5
9 3.77 0:25 243123 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5 Ned4
369418 nodes total, 12313 nodes/second

fen = 2rq1rk1/pp1bnpbp/4p1p1/3pP1N1/3P2Q1/2PB4/P4PPP/R1B1R1K1 w - - bm Nxh7
2 0.70 0:00 351 Qh4 h5
3 0.38 0:01 589 Qh4 h5 Bd2
3 0.69 0:02 689 Qh3 h5 Rb1
4 -0.53 0:03 2555 Qh3 h6 Nf3 Rxc3
4 0.48 0:04 3168 Bd2 h6 Nf3 Nf5 Bxf5 gxf5
5 0.48 0:06 6992 Bd2 h6 Nf3 Nf5 Bxf5 gxf5
6 0.52 0:08 18146 Bd2 h6 Nh3 h5 Qf3 Ba4
7 0.33 0:10 38484 Bd2 h6 Nf3 h5 Qg3 Ba4 Rab1
8 0.48 0:21 89128 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qc7
9 0.39 0:28 144144 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qc7 Bg5
180000 nodes total, 6000 nodes/second

Mine:

fen = 3rrk2/1bnqppb1/p2p2p1/1p1P2P1/1p2PP1R/2NQBN2/1PP2K2/7R w - - bm Rh8+
2 1.34 0:00 538 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6
3 0.08 0:00 1322 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6 gxf6+ Kxh8 fxe7+ Kh7
exd8=Q Qxd8 Qxb4
3 0.77 0:00 1924 Bd4 f6 Ne2
4 0.39 0:00 2939 Bd4 e5 Rh8+ Bxh8 Rxh8+ Ke7
5 0.73 0:00 8450 Bd4 f6 Ne2 a5 Rh7
6 0.50 0:01 15852 Bd4 f6 Ne2 a5 Ng3 e5
7 0.47 0:02 56713 Bd4 f6 Ne2 a5 Ng3 e5 gxf6 Bxf6
7 3.84 0:05 80037 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2
8 3.75 0:07 100903 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5
9 3.67 0:11 166508 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5 Ke3
10 4.23 0:17 362474 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ng5 bxc3 Qh3 Rg7 Bxg7 cxb2 Bxb2
442103 nodes total, 14736 nodes/second

fen = 2rq1rk1/pp1bnpbp/4p1p1/3pP1N1/3P2Q1/2PB4/P4PPP/R1B1R1K1 w - - bm Nxh7
2 0.70 0:00 324 Qh4 h5
3 0.38 0:00 564 Qh4 h5 Bd2
3 0.64 0:00 666 Qh3 h5 Rb1 Rxc3 Rxb7
4 -0.45 0:00 2458 Qh3 h6 Nf3 Rxc3
4 0.53 0:00 2915 Bd2 Nf5 Rab1 Qc7
5 0.42 0:00 6527 Bd2 h6 Nf3 h5 Qh4
6 0.53 0:01 14301 Bd2 h6 Nh3 Nf5 Rab1 h5
7 0.34 0:02 32598 Bd2 h6 Nf3 h5 Qg3 Nf5 Bxf5 gxf5
7 0.36 0:08 74735 Qd1 Nf5 g4 Ne7 Rb1 Rxc3 Rxb7
8 -0.25 0:12 115309 Qd1 Rxc3 Rb1 Rc7 Ba3 h6 Nf3 h5
8 0.53 0:13 127813 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qc7
9 0.44 0:17 166466 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qc7 Bg5
10 0.38 0:26 275808 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Qg3 Nf5 Bxf5 gxf5
360000 nodes total, 12000 nodes/second

Jim.

[Bryan Hofmann]

Jim

You can't use these benchmarks as they are starting 2 threads on a single CPU. On a Single CPU you need to use "bench 30 1 128". The above results show nothing more then the MS builds are more accuate in the test you ran in you environment over the GCC builds.


Bryan

[Jim Ablett]

You can't use these benchmarks as they are starting 2 threads on a single CPU. On a Single CPU you need to use "bench 30 1 128". The above results show nothing more then the MS builds are more accuate in the test you ran in you environment over the GCC builds.

Here's my revised benchmarks using 1 thread.

Fritz Ruel:

fen = 3rrk2/1bnqppb1/p2p2p1/1p1P2P1/1p2PP1R/2NQBN2/1PP2K2/7R w - - bm Rh8+
2 1.41 0:00 529 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6
3 0.19 0:00 1311 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6 gxf6+ Kxh8 fxe7+ Kh7
exd8=Q Qxd8 Qxb4
3 0.83 0:00 1898 Bd4 f6 gxf6 exf6
4 0.52 0:00 2703 Bd4 e5 Rh8+ Bxh8 Rxh8+ Ke7
5 0.48 0:00 7563 Bd4 f6 gxf6 exf6 Ne2
6 0.42 0:00 16942 Bd4 f6 Ne2 a5 Rh7 Qg4
7 0.53 0:00 68172 Bd4 f6 Ne2 a5 Ng3 e5 gxf6 Bxf6
7 3.95 0:00 91330 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2
8 3.80 0:00 113120 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5
9 3.72 0:00 179605 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5 Ke3
10 3.75 0:01 379110 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 Rf7 Qd4 a5
11 5.30 0:02 937229 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 e5 Kf8 Qxg6 Bxd5 Qh6+ Kg8 Qh8+
Kf7 Qh7+ Ke6
12 5.81 0:07 2739k Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kf8 Rxd7 Rxd7 Ng5 Rf7 Qh3 Rxf6 Qh8+ Ke7 Qg7+
Kd8 Qxf6+ Kc8 Ne2
13 6.66 0:16 6102k Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Ng5+ Kxf6 e5+ dxe5 Nce4+ Kg7 Rh7+ Kf8 Qh3 Qxh3 Rf7+

Kg8 Nf6+ Kh8 Nxh3 Rd7 Rxd7 Re7 Rxe7 Bxd5 Rxc7
14 #16 0:23 9255k Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Ng5+ Kxf6 e5+ dxe5 Nce4+ Kg7 Rh7+ Kf8 Qh3 Qxh3 Rf7+

Kg8 Nf6+ Kh8 Nxh3 Rd7 Rxd7 Re7 Rxe7 Bxd5 Rh7#
12000000 nodes total, 400000 nodes/second

fen = 2rq1rk1/pp1bnpbp/4p1p1/3pP1N1/3P2Q1/2PB4/P4PPP/R1B1R1K1 w - - bm Nxh7
2 0.70 0:00 329 Qh4 h5
3 0.38 0:00 565 Qh4 h5 Bd2
3 0.69 0:00 668 Qh3 h5 Rb1
4 -0.52 0:00 2510 Qh3 h6 Nf3 Rxc3
4 0.48 0:00 3151 Bd2 h6 Nf3 Nf5 Bxf5 gxf5
5 0.48 0:00 7469 Bd2 h6 Nf3 Nf5 Bxf5 gxf5
6 0.53 0:00 19059 Bd2 h6 Nh3 Nf5 Rab1 h5
7 0.27 0:00 38592 Bd2 h6 Nh3 Nf5 Nf4 Ba4 Bxf5 gxf5
8 0.66 0:00 117480 Bd2 h6 Nh3 Ba4 Rab1 Qd7 Qh4 h5
9 0.41 0:00 168917 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Qh4
10 0.55 0:00 325348 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qd7 Rb4 Rfe8
11 0.56 0:01 618986 Bd2 h6 Nf3 h5 Qh3 Re8 Rab1 Rc7 Ng5 Nf5 Bxf5 exf5
12 0.64 0:03 1136k Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Ng5 Nf5 Rb4 Qd7 Bxf5

exf5
13 0.64 0:07 2649k Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Ng5 Nf5 Rb4 Qd7 Bxf5

exf5
14 0.64 0:16 5871k Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Ng5 Nf5 Rb4 Qd7 Bxf5

exf5
10480000 nodes total, 349333 nodes/second



Bryan Hoffman:

fen = 3rrk2/1bnqppb1/p2p2p1/1p1P2P1/1p2PP1R/2NQBN2/1PP2K2/7R w - - bm Rh8+
2 1.41 0:00 530 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6
3 0.19 0:00 1354 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6 gxf6+ Kxh8 fxe7+ Kh7
exd8=Q Qxd8 Qxb4
3 0.88 0:00 1940 Bd4 f6 gxf6 exf6
4 0.52 0:00 2738 Bd4 e5 Rh8+ Bxh8 Rxh8+ Ke7
5 0.53 0:00 7884 Bd4 f6 gxf6 exf6 Ne2
6 0.42 0:00 17362 Bd4 f6 Ne2 a5 Rh7 Qg4
7 0.53 0:00 87966 Bd4 f6 Ne2 a5 Ng3 e5 gxf6 Bxf6
7 1.67 0:00 116757 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ Kxh8 f7+ Kh7 Qd1
Qg4 fxe8=Q Rxe8 Ng5+ Qxg5 fxg5 bxc3 Bxc3
8 3.83 0:00 160236 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5
9 3.77 0:00 241017 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5 Ned4
10 3.86 0:01 475371 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5 Ng5 Kf8
11 5.30 0:03 1214k Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kf8 Ng5 Qxh7 Nxh7+ Kf7 Bxd8 bxc3 Qxc3 Nxd5
Ng5+ Kg8 exd5 Bxd5
12 5.89 0:07 3032k Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kf8 Ng5 Qxh7 Nxh7+ Kf7 Bxd8 Rxd8 Ng5+ Ke8 Ne2
a5 Nd4
13 6.73 0:17 7103k Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Ng5+ Kxf6 e5+ dxe5 Nce4+ Kg7 Rh7+ Kf8 Qh3 Qxh3 Rf7+

Kg8 Nf6+ Kh8 Nxh3 Rd7 Rxd7 Re7 Rxe7 Bxd5 Rxc7
14 #16 0:26 10777k Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Ng5+ Kxf6 e5+ dxe5 Nce4+ Kg7 Rh7+ Kf8 Qh3 Qxh3 Rf7+

Kg8 Nf6+ Kh8 Nxh3 Rd7 Rxd7 Re7 Rxe7 Bxd5 Rh7#
12480000 nodes total, 416000 nodes/second

fen = 2rq1rk1/pp1bnpbp/4p1p1/3pP1N1/3P2Q1/2PB4/P4PPP/R1B1R1K1 w - - bm Nxh7
2 0.70 0:00 351 Qh4 h5
3 0.38 0:00 589 Qh4 h5 Bd2
3 0.69 0:00 689 Qh3 h5 Rb1
4 -0.53 0:00 2555 Qh3 h6 Nf3 Rxc3
4 0.48 0:00 3168 Bd2 h6 Nf3 Nf5 Bxf5 gxf5
5 0.48 0:00 6992 Bd2 h6 Nf3 Nf5 Bxf5 gxf5
6 0.52 0:00 18150 Bd2 h6 Nh3 h5 Qf3 Ba4
7 0.33 0:00 38114 Bd2 h6 Nf3 h5 Qg3 Ba4 Rab1
8 0.48 0:00 87095 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qc7
9 0.39 0:00 144363 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qc7 Bg5
10 0.53 0:00 355288 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Rb4 Qd7
11 0.56 0:01 596793 Bd2 h6 Nf3 h5 Qh3 Re8 Rab1 Rc7 Ng5 Nf5 Bxf5 exf5
12 0.52 0:02 1078k Bd2 h6 Nf3 h5 Qg3 Qa5 Reb1 Rc7 Nh4 Ba4 Qf3 Kh8
13 0.61 0:06 2587k Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Ng5 Nf5 Rb4 Qd7 Bxf5

exf5
14 0.61 0:11 4198k Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Ng5 Nf5 Rb4 Qd7 Bxf5

exf5
10760000 nodes total, 358666 nodes/second



Jim Ablett:

fen = 3rrk2/1bnqppb1/p2p2p1/1p1P2P1/1p2PP1R/2NQBN2/1PP2K2/7R w - - bm Rh8+
2 1.34 0:00 538 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6
3 0.08 0:00 1322 Rh8+ Bxh8 Rxh8+ Kg7 Qd4+ f6 gxf6+ Kxh8 fxe7+ Kh7
exd8=Q Qxd8 Qxb4
3 0.77 0:00 1924 Bd4 f6 Ne2
4 0.39 0:00 2939 Bd4 e5 Rh8+ Bxh8 Rxh8+ Ke7
5 0.73 0:00 8450 Bd4 f6 Ne2 a5 Rh7
6 0.50 0:00 15875 Bd4 f6 Ne2 a5 Ng3 e5
7 0.47 0:00 56779 Bd4 f6 Ne2 a5 Ng3 e5 gxf6 Bxf6
7 3.84 0:00 80104 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2
8 3.75 0:00 100968 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5
9 3.67 0:00 165907 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 a5 Ke3
10 3.67 0:01 365600 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kg8 Rxd7 Rxd7 Ne2 Rh7 Ng5 Rh2+ Ke3
11 5.14 0:03 978716 Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kf8 Ng5 Qxh7 Nxh7+ Kf7 Bxd8 bxc3 Qxc3 Nxd5
Ng5+ Kg8 exd5 Rxd8
12 5.72 0:08 2603k Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Rh7+ Kf8 Ng5 Qxh7 Nxh7+ Kf7 Bxd8 Rxd8 Ng5+ Ke8 Ne2
a5 Ke3
13 6.61 0:18 6019k Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Ng5+ Kxf6 e5+ dxe5 Nce4+ Kg7 Rh7+ Kf8 Qh3 Qxh3 Rf7+

Kg8 Nf6+ Kh8 Nxh3 Rd7 Rxd7 Re7 Rxe7 Bxd5 Rxc7
14 #16 0:27 9335k Rh8+ Bxh8 Rxh8+ Kg7 Bd4+ f6 gxf6+ exf6 Bxf6+ Kf7
Ng5+ Kxf6 e5+ dxe5 Nce4+ Kg7 Rh7+ Kf8 Qh3 Qxh3 Rf7+

Kg8 Nf6+ Kh8 Nxh3 Rd7 Rxd7 Re7 Rxe7 Bxd5 Rh7#
10360000 nodes total, 345333 nodes/second

fen = 2rq1rk1/pp1bnpbp/4p1p1/3pP1N1/3P2Q1/2PB4/P4PPP/R1B1R1K1 w - - bm Nxh7
2 0.70 0:00 324 Qh4 h5
3 0.38 0:00 564 Qh4 h5 Bd2
3 0.64 0:00 666 Qh3 h5 Rb1 Rxc3 Rxb7
4 -0.45 0:00 2458 Qh3 h6 Nf3 Rxc3
4 0.53 0:00 2915 Bd2 Nf5 Rab1 Qc7
5 0.42 0:00 6527 Bd2 h6 Nf3 h5 Qh4
6 0.53 0:00 14301 Bd2 h6 Nh3 Nf5 Rab1 h5
7 0.34 0:00 32619 Bd2 h6 Nf3 h5 Qg3 Nf5 Bxf5 gxf5
8 0.53 0:00 101616 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qc7
9 0.44 0:00 140841 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Qc7 Bg5
10 0.38 0:00 255242 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Qg3 Nf5 Bxf5 gxf5
11 0.56 0:01 596878 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Rb4 Qe8 c4 dxc4 Bxc4

12 0.58 0:03 977802 Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Rb4 Qd7 Reb1 Rfc8
13 0.58 0:07 2244k Bd2 h6 Nf3 h5 Qh3 Ba4 Rab1 Rc7 Rb4 Qd7 Ng5 Nf5 Bxf5

exf5
14 0.56 0:14 4372k Bd2 h6 Nf3 h5 Qh3 Qa5 Rab1 Rc7 Rb2 Nf5 Reb1 Rb8
Bxf5 exf5
9260000 nodes total, 308666 nodes/second

Don't you think I've had enough disappointment heaped
on my this week with England getting knocked out the world cup!

Jim

[Bryan Hofmann]

Don't you think I've had enough disappointment heaped
on my this week with England getting knocked out the world cup!

Jim

Not much of a sports fan myself, and have not watch a game since I left England back in 89 (was stationed at Bentwaters). Also you should note that the benmark test runs on 10 positions and the stats are shown for each position run and are not combined, so you are only displaying the last 2 of the 10 postions. Sorry don't mean to heap more on you.


Bryan

[Tord Romstad]

Don't you think I've had enough disappointment heaped
on my this week with England getting knocked out the world cup!

Jim

Not much of a sports fan myself, and have not watch a game since I left England back in 89 (was stationed at Bentwaters). Also you should note that the benmark test runs on 10 positions and the stats are shown for each position run and are not combined, so you are only displaying the last 2 of the 10 postions. Sorry don't mean to heap more on you.

And I don't feel any need to heap any more on either of you. You have both (along with lots of other helpful people) given me plenty of help, and thanks to all you guys I now feel reasonably confident that Glaurung finally does work correctly with 4 threads in Windows.

Thanks, everyone!

Tord