Nirvanachess 2.1 Release

[gerold]

Very good. Thanks Thomas.

Best,
Gerold.

[carldaman]

Hi Thomas,

There's some strange behavior, where Nirvanachess 2.1 actually considers moving its King into check in the visible PV, before switching to a different move. I don't recall ever seeing something like this before. Hard to say if this actually weakens the engine, but cpu time repeatedly appears to be spent analyzing illegally moving into check. I've only tested with ponder off and Arena 3.5.

Just a heads-up.

Thanks,
CL

@Gordon
Hmm. Well in any case the issue is fixed. Here is a direct link.

http://www.mediafire.com/download/1ied4 ... ss+2.1.rar


@Tim
Take a look again. In the main post the ratings are 39 and -39 respectively so combining the results gets 78. 80 is a nice round number and is within the error bars so I chose to use that.

Also something else I forgot to mention that I added into Nirvana in this version was a change to move ordering. Previously any searched quite move that did not produce a cut off had its history score reduced if the node turned out to be a cut node. However now if the move failed high during a reduced search and then failed low during the research its score is left alone with the idea being that if a similar position is found in the future this move will need to be researched again if its score is reduced.

[Zenmastur]

Steve I sent you a PM in response to one you sent me about a month ago that I missed.

Regards,

Zen

[cetormenter]

Strange. However it might be something with arena? I don't trust Arena 3.5 as a gui. There are far too many bugs. I would "downgrade" to version 3.0 and see if the problems persist.

One easy way to see how broken Arena 3.5 is, is to attempt to run a tournament using pondering and see how almost every engine crashes before it can complete its first game.

[carldaman]

Strange. However it might be something with arena? I don't trust Arena 3.5 as a gui. There are far too many bugs. I would "downgrade" to version 3.0 and see if the problems persist.

One easy way to see how broken Arena 3.5 is, is to attempt to run a tournament using pondering and see how almost every engine crashes before it can complete its first game.

I can reproduce the behavior under Arena 3.0, but everything looks okay under Chessbase.

Here is the Arena output:

[d]2r4k/p1r2p1p/4pNpP/1p1p2P1/qPnP1P2/1RPR4/P7/KQ6 w - - 1 52

Code: Select all

Nirvanachess 2.1 x64_pop MP:
  1/4	00:00	          29	25,000	-3.92	Nf6-d7 Rc7xd7
  1/4	00:00	      10,985	844,307	-2.02	Nf6-g4
  1/7	00:00	      20,660	794,346	+0.47	Rd3-d1
  2/4	00:00	      29,976	788,657	+0.59	Rd3-d1 e6-e5 Qb1-d3
  3/8	00:00	      30,347	778,051	+0.61	Rd3-d1 Nc4-e3 Rd1-c1
  4/11	00:00	      30,605	784,615	+0.61	Rd3-d1 Nc4-e3 Rd1-c1 Ne3-c4
  5/13-	00:00	      32,243	806,000	 0.00	Ka1-b2
  5/5	00:00	      32,251	806,250	 0.00	Qb1-e1
  6/6	00:00	      32,291	807,225	 0.00	Qb1-e1
  7/7	00:00	      32,329	808,200	 0.00	Qb1-e1
  8/8	00:00	      32,606	815,075	 0.00	Qb1-e1
  8/11+	00:00	      35,539	826,279	+0.08	Qb1-d1
  8/11+	00:00	      40,487	861,297	+0.28	Qb1-d1
  8/13	00:00	      44,510	872,607	+0.01	Qb1-d1 a7-a5 Rb3-b1 Qa4xd1 Rb1xd1 a5xb4 c3xb4 Rc7-a7
  9/17	00:00	      67,796	941,486	 0.00	Qb1-c2 Rc7-c6 Rd3-h3
 10/12	00:00	      72,136	949,105	 0.00	Qb1-c2 Rc7-c6 Rd3-h3
 11/20-	00:00	      95,121	1,001,126	-0.07	Ka1-b2
 11/21+	00:00	     147,024	1,122,206	+0.07	Nf6-g4
 11/20	00:00	     194,854	1,202,709	 0.00	Nf6-g4 Rc8-a8 Ng4-e5 Kh8-g8 Ne5-g4 Kg8-h8
 12/25	00:00	     233,879	1,277,978	 0.00	Nf6-g4 Rc8-a8 Ng4-e5 Kh8-g8 Ne5-g4 Kg8-h8
 13/21	00:00	     261,126	1,325,446	 0.00	Nf6-g4 Rc8-a8 Ng4-e5 Kh8-g8
 14/23	00:00	     384,361	1,455,878	 0.00	Nf6-g4 Rc8-a8 Ng4-e5 Kh8-g8 Ne5-g4 Kg8-h8
 15/20	00:00	     424,527	1,500,060	 0.00	Nf6-g4 Rc8-a8
 16/21	00:00	     594,804	1,638,542	 0.00	Nf6-g4 Rc8-a8 Ng4-e5 Kh8-g8
 17/30	00:00	     856,920	1,792,686	 0.00	Nf6-g4 Rc8-a8 Ng4-e5 Kh8-g8
 18/29	00:00	   1,326,004	1,886,186	 0.00	Nf6-g4 Rc8-a8 Ng4-e5 Kh8-g8 Ne5-g4 Kg8-h8
 19/32-	00:01	   2,137,086	1,978,765	-0.07	Ka1-b2
 19/29-	00:01	   2,511,444	2,012,355	-0.15	Ka1-b2
 19/38	00:02	   4,746,237	2,059,098	 0.00	Qb1-c1 a7-a5 Rd3-h3 Rc7-a7 Rh3-h2 Rc8-a8 Qc1-c2 a5xb4 Rb3xb4 Qa4-a5 Rh2-f2 Ra7-a6 Rf2-h2 Ra6-a7
 20/33	00:02	   5,411,924	2,083,107	 0.00	Qb1-c1 a7-a5 Rd3-h3 Rc7-a7 Rh3-h2 Rc8-a8 Qc1-c2 a5xb4 Rb3xb4 Qa4-a3 Rb4-b3 Nc4-e3 Qc2-b1 Qa3-a5 Rh2-f2 Ne3-c4 Rf2-h2 Qa5-c7 Rh2-f2 Qc7-a5
 21/33	00:03	   6,508,203	2,126,859	 0.00	Qb1-c1 a7-a5 Rd3-h3 Rc7-a7 Rh3-h2 Rc8-a8 Qc1-c2 a5xb4 Rb3xb4 Qa4-a3 Rb4-b3 Nc4-e3 Qc2-b1
 22/41-	00:03	   8,170,663	2,165,557	-0.07	Ka1-b2
 22/33-	00:04	   9,562,743	2,177,803	-0.15	Ka1-b2
 22/35	00:04	  10,157,716	2,180,700	 0.00	Qb1-c1 a7-a5 Rd3-h3 Rc7-a7 Rh3-h2 Rc8-a8 Qc1-b1 a5xb4 Rb3xb4 Qa4-a3 Rb4-b3 Qa3-d6 Rh2-f2 Ra7-a3 Rb3xb5 Ra3xc3 Rb5-b7 Qd6-f8 Rb7-b3 Rc3xb3 Qb1xb3 Qf8-b8 Qb3xb8+ Ra8xb8 a2-a4 Rb8-a8
 23/34-	00:05	  11,055,918	2,182,804	-0.07	Ka1-b2
 23/37-	00:05	  12,132,014	2,191,078	-0.15	Ka1-b2
 23/40	00:06	  13,592,327	2,210,132	 0.00	Qb1-c1 a7-a5 Rd3-h3 Rc7-a7 Rh3-h2 Rc8-a8 Qc1-b1 a5xb4 Rb3xb4 Qa4-a3 Rb4-b3 Qa3-d6 Rh2-f2 Ra7-a3 Nf6-g4 Ra3xb3 Qb1xb3 Qd6-a3 Ka1-b1 Qa3-a5 Ng4-f6 Qa5-a4
 24/44-	00:07	  16,242,998	2,217,775	-0.07	Ka1-b2
 24/37-	00:08	  18,683,150	2,236,698	-0.15	Ka1-b2
 24/40	00:10	  23,260,089	2,243,232	-0.21	Qb1-c1 a7-a5 Rd3-h3 Rc7-a7 Rh3-h2 Rc8-a8 Qc1-b1 a5xb4 Rb3xb4 Qa4-a3 Rb4-b3 Qa3-d6 Rh2-f2 Ra7-a3 Nf6-g4 Nc4-a5 Rb3xb5 Qd6-c7 Rb5-c5 Na5-b3+ Qb1xb3 Qc7xc5 d4xc5 Ra3xb3
 25/40+	00:12	  27,648,760	2,234,964	-0.13	Ka1-b2
 25/41	00:13	  29,007,635	2,235,999	-0.06	Qb1-c1 a7-a5 Rd3-h3 Rc7-a7 Rh3-h2 Rc8-a8 Qc1-b1 a5xb4 Rb3xb4 Qa4-a3 Rb4-b3 Qa3-d6 Rh2-f2 Ra7-a3 Nf6-g4 Nc4-a5 Rb3xa3 Qd6xa3 Qb1-b2 Na5-c4 Qb2xa3 Ra8xa3 Rf2-c2 Ra3-a5 Ng4-e5 Kh8-g8
 26/39-	00:14	  31,469,890	2,200,229	-0.14	Ka1-b2
 26/39-	00:15	  34,395,836	2,167,349	-0.21	Ka1-b2

No illegal moves were made, regardless of the strange PV output.

[cetormenter]

Small bugfix to the incorrect mate scores. If you did not experience them then there is no need to update.

http://www.mediafire.com/download/5hw5b ... s_2.1b.rar

[cetormenter]

And one final change to SMP there was a bug that would cause Nirvana to play much weaker under SMP conditions. Just like the previous version no need to update if you are not running Nirvana under these conditions.

http://www.mediafire.com/download/8hslh ... s_2.1c.rar

[Dariusz]

Hello Thomas,

Is any chance to compile Nirvana for Mac (OSX) ?

[cdani]

Is doing really well in TCEC!! Congratulations!!