Author: Heiner Marxen
Date: 12:55:59 03/17/02
Go up one level in this thread
On March 17, 2002 at 08:28:37, leonid wrote:
>On March 17, 2002 at 05:54:26, Tim Foden wrote:
>
>>On March 16, 2002 at 07:17:13, leonid wrote:
>>
>>>[D]3r4/1nkrnqqp/3b4/3qQQQQ/2bN2BB/1qN3QK/pR3QQ1/R3QQ2 w - -
>>>
>>>Please indicate your result.
>>
>>Hi Leonid,
>
>Hi, Tim,
Hi Tim, hi Leonid!
>I actually don't know if your solution in 7 moves is real. It could be.
Chest actually confirms this to be a mate in 7.
> My
>problem is that on this position I had pretty bad branching factor, around 30.
My branching factor also is alarming, here:
# 3 0.02s 2kN [ 23.42] 0.97 95- 0
# 4 0.69s [ 34.50] 44kN [ 28.75] 1.05 1867- 0
# 5 20.57s [ 29.81] 1157kN [ 26.45] 1.42 69216- 0
# 6 404.29s [ 19.65] 22294kN [ 19.27] 1.82 1530318- 8
# 7 7170.33s [ 17.74] 396740kN [ 17.80] 2.06 27904701- 19156805
It also started around 30, but fortunately dropped below 20, probably due
to better hashing effectivity at larger depthes. Still, 20 is laaarge.
>So, when I look into this position (that was created somewhere in last Summer) I
>see only that mate was searched 6 moves deep by brute force. With hash and on
>more quick computer this position, for sure, could be searched easily 7 moves by
>brute force.
>
>9 moves selective took 23 seconds. Celeron 600Mhz. No hash.
>
>>
>>Green Light finds that Ndb5+ is a mate in 7 after 40.06 seconds.
>>
>>[Note: I worry that this is still showing a bug, as you don't seem to be posting
>>such sort mates recently. :)]
Chest on K7/600 with 350 MB hash needs 2 hours to find 3 key moves for
mate in 7:
PV: Ndb5+ Bxb5 Nxb5+ Kc8 Nxd6+ Qxd6 Qec5+ Qxc5 Q5xc5+ Nxc5 Qgxc5+ Nc6 Qcxc6#
PV: Nxd5+ Qxd5 Qxb3 Bxb3 Nb5+ Qxb5 Qfxb5 Nc6 Qgxc6+ Kb8 Qcxd6+ Rc7 Qgxd8#
PV: Qxd6+ Rxd6 Ndb5+ Qbxb5 Nxb5+ Qxb5 Qgxe7+ Qxe7 Qxc4+ Qc6 Qfa5+ Nxa5 Qhxa5#
Hence, your solution appears to be correct, Tim!
>It could be that you have as little time to look into your code as I have.
>Already for good two months I wrote some new code that, presumably, should speed
>my mate search. Last writing contain visible bugs. Now I am just unable to find
>free week to go into full debugging.
A problem which many of us appear to share with you.
My last EGTB statistics addition produce apparently nonsense numbers, yet I
have not debugged that since weeks. :-(
Well, it's just a hobby, and only so important.
Chest knocked down another one... isn't that great? :-O
Cheers,
Heiner
>Cheers,
>Leonid.
>
>>Cheers, Tim.
>>
>>Analysis (GLC2.18, Athlon XP1700+, 48MB hash):
>>
>>>anal
>> Game stage: Opening
>> Current eval: 49.043
>> Ply Time Score Nodes Principal Variation
>> 2 0.07 +58.862 6221 Qexg7 Qdxf5 {--}
>> 2 0.07 +49.425 9215 Qexg7 Qxg2+ 2. Qgxg2 Qxg7
>> 2 0.07 +58.863 9552 Nxd5+ Bxd5 2. Qexg7 {++}
>> 2 0.08 +59.862 10299 Nxd5+ Bxd5 2. Qexg7 Bxg2+ 3. Q1xg2 Nxf5 4. Nxb3
>> 2 0.10 +59.862 10507 Nxd5+ Bxd5 2. Qexg7 Bxg2+ 3. Q1xg2 Nxf5 4. Nxb3
>> 3 0.20 +59.462 93701 Nxd5+ Bxd5 2. Qexg7 Bxg2+ 3. Q1xg2 Nxf5 {--}
>> 3 0.23 +58.925 111225 Nxd5+ Bxd5 2. Rc1+ Kb8 3. Qexg7 Bxg2+ 4. Q1xg2 Qxg7
>> 5. Nxb3
>> 3 0.31 +58.925 176264 Nxd5+ Bxd5 2. Rc1+ Kb8 3. Qexg7 Bxg2+ 4. Q1xg2 Qxg7
>> 5. Nxb3
>> 4 0.94 +58.525 751317 Nxd5+ Bxd5 2. Rc1+ Kb8 3. Qexg7 Bxg2+ 4. Q1xg2 Nxf5
>> {--}
>> 4 1.20 +58.359 966077 Nxd5+ Bxd5 2. Qexd5 Qfxd5 3. Qfxd5 Qxb2 4. Qxb2
>> Bxg3 5. Qxg7 Rxd5 6. Bxg3+
>> 4 1.54 +58.359 1298367 Nxd5+ Bxd5 2. Qexd5 Qfxd5 3. Qfxd5 Qxb2 4. Qxb2
>> Bxg3 5. Qxg7 Rxd5 6. Bxg3+
>> 5 1.79 +58.759 1500348 Nxd5+ Bxd5 2. Qexd5 Qfxd5 3. Qxb3 Nxf5 4. Qgxd5
>> {++}
>> 5 3.15 +60.390 2732941 Nxd5+ Qxd5 2. Qexd5 Qxg3+ 3. Bxg3 Bxd5 4. Bxd6+
>> Rxd6 5. Qexe7+ Qxe7 6. Qxe7+ R6d7
>> 5 7.11 +60.390 6418746 Nxd5+ Qxd5 2. Qexd5 Qxg3+ 3. Bxg3 Bxd5 4. Bxd6+
>> Rxd6 5. Qexe7+ Qxe7 6. Qxe7+ R6d7
>> 6 7.29 +60.790 6518798 Nxd5+ Qxd5 2. Qexd5 Qxg3+ 3. Bxg3 Bxd5 4. Bxd6+
>> Rxd6 5. Nb5+ Kc6 6. Rc1+ Nc5 7. Rxc5+ Kb6 8.
>> Nxd6+ {++}
>> 6 14.09 +67.813 12926k Nxd5+ Qxd5 2. Qxd6+ Rxd6 3. Qexe7+ Qxe7 4. Qxe7+
>> Kb6 5. Nxb3+ Nc5 6. Nxc5+ Kc6 7. Qgxd5+ Rxd5
>> 6 26.79 +73.116 24279k Ndb5+ Bxb5 2. Nxb5+ Kc8 3. Rc1+ Qc3 4. Qe1xc3+ Nc6
>> 5. Nxd6+ Kb8 6. Qexd5 Qxh5 7. Bxh5 Qxc3 8. Rxc3
>> Rxd6 9. Rxc6
>> 6 36.38 +73.116 34092k Ndb5+ Bxb5 2. Nxb5+ Kc8 3. Rc1+ Qc3 4. Qe1xc3+ Nc6
>> 5. Nxd6+ Kb8 6. Qexd5 Qxh5 7. Bxh5 Qxc3 8. Rxc3
>> Rxd6 9. Rxc6
>> 7 36.54 +73.516 34222k Ndb5+ Bxb5 2. Nxb5+ Kc8 3. Rc1+ Qc3 4. Qe1xc3+ Nc6
>> 5. Nxd6+ Kb8 6. Rxb7+ Rxb7 7. Qxd8+ Nxd8 {ht}
>> {++}
>> 7 40.09 +Mate07 40249k Ndb5+ Bxb5 2. Nxb5+ Kc8 3. Rc1+ Qc3 4. Qe1xc3+ Nc6
>> 5. Qxc6+ Qxc6 6. Qxc6+ Kb8 7. Qa7#
>> 7 49.34 +Mate07 56317k Ndb5+ Bxb5 2. Nxb5+ Kc8 3. Rc1+ Qc3 4. Qe1xc3+ Nc6
>> 5. Qxc6+ Qxc6 6. Qxc6+ Kb8 7. Qa7#
>> 8 1:01 +Mate07 75847k Ndb5+ Bxb5 2. Nxb5+ Kc8 3. Rc1+ Qc3 4. Qe1xc3+ Nc6
>> 5. Qxc6+ Qxc6 6. Qxc6+ Kb8 7. Qa7#
>> 8 1:28 +Mate07 121227k Ndb5+ Bxb5 2. Nxb5+ Kc8 3. Rc1+ Qc3 4. Qe1xc3+ Nc6
>> 5. Qxc6+ Qxc6 6. Qxc6+ Kb8 7. Qa7#
>> 9 1:60 +Mate07 171972k Ndb5+ Bxb5 2. Nxb5+ Kc8 3. Rc1+ Qc3 4. Qe1xc3+ Nc6
>> 5. Qxc6+ Qxc6 6. Qxc6+ Kb8 7. Qa7#
>> 9 4:34 +Mate07 437006k Ndb5+ Bxb5 2. Nxb5+ Kc8 3. Rc1+ Qc3 4. Qe1xc3+ Nc6
>> 5. Qxc6+ Qxc6 6. Qxc6+ Kb8 7. Qa7#
>> 10 6:08 +Mate07 600308k Ndb5+ Bxb5 2. Nxb5+ Kc8 3. Rc1+ Qc3 4. Qe1xc3+ Nc6
>> 5. Qxc6+ Qxc6 6. Qxc6+ Kb8 7. Qa7#
>> 10 9:23 4/94 Qxd6+ >exit
>> local: t=9:29 nps=1675156.8 n=952775554 (22.5% / 77.5%) fh=95.8%
>> total: t=9:29 nps=1675156.8 n=952775554 draws=20789
>> trans: probes=65302496 hits=6139216 (9.40%) draft=4043726 (6.19%)
>> tcuts: exact=91 (0.00%) upper=2018837 (3.09%) lower=1854448 (2.84%)
>> tstor: exact=662 (0.00%) upper=23273107 (69.37%) lower=10273484 (30.62%)
>> ext: check=99211808 recap=697508 ppush=0 1rep=1528110 thrt=0
>> q-moves: gen=58508721 tested=7789580 made/un=3903284 max-dep=10
>> max eval diff: part-1=2.284 part-2=2.982
>>>
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.