Author: Robert Hyatt
Date: 13:54:32 05/19/98
Go up one level in this thread
On May 19, 1998 at 10:18:07, Bernhard Bauer wrote:
>>
>>Here's the position in EPD format:
>>
>>r1bq1rk1/8/6p1/1PN2n2/p2pB2p/3Nb3/P5PP/1R1Q1R1K w - -
>>
>>Solution times so far (ranking according to times to play a different
>>move than Bxa8, details below, all times on a P233MMX, 128MB RAM):
>>
if we are going to make a table, here is the 4 ppro/200 result from my
ALR box:
10 23.05 5.54 26. Bxa8 Ng3+ 27. hxg3 hxg3 28.
Rxf8+
Qxf8 29. Bd5+ Kh8 30. Bf3 Qd8 31.
Nf4
Qh4+ 32. Nh3 Bxh3 <HT>
10-> 32.09 5.54 26. Bxa8 Ng3+ 27. hxg3 hxg3 28.
Rxf8+
Qxf8 29. Bd5+ Kh8 30. Bf3 Qd8 31.
Nf4
Qh4+ 32. Nh3 Bxh3 <HT>
11 36.55 -- 26. Bxa8
11 53.51 0.56 26. Bxa8 Ng3+ 27. hxg3 hxg3 28. Bd5+
Kh7 29. Nf4 Bxf4 30. Rf3 Qxd5 31.
Rc3
Qh5+ 32. Qxh5+ gxh5 33. Rd3 Bf5
11 1:25 ++ 26. b6!!
11 2:15 3.16 26. b6 Rb8 27. b7 Ng3+ 28. hxg3
Rxf1+
29. Qxf1 hxg3 30. Nf4 Be6 31. Ncxe6
Qh4+ 32. Nh3 Qxe4 33. Qf7+ Kxf7 34.
Neg5+ Ke7 35. Nxe4
so at 1:25 it changes to b6.
>
>No Program / Hardware / Ne5 / b6 / g4
> -------------------------------------------------------------
> 1. DarkThought / 21164a600MHz / - / - / 01:45
> 2. Crafty 15.8 / 2xPPro233 / - / - / 3:11 (6M/2.5M)
> 3. MCP 7.1 / P233MMX / - / 12:27 / 03:29
> 4. Crafty 15.8 / 2xPPro233 / 4:01 / 8:42 / 14:46 (24M/10M)
> 5. Fritz 5 / P233MMX / - / 05:39 / 24:54
> 6. Hiarcs 6 / P233MMX / - / 10:57 / 15:44
> 7. Shredder 2 / P233MMX / 21:47 / 01:09:48 / 01:53:44
> 8. Rebel 9 / P233MMX / 35:54 / 51:44 / not tested
> 9. CSTal Paris / P233MMX / - / 01:20:46 / 02:08:42
>
>
>Crafty 15.8 ran on a dual PPro200 overclocked to 233 MHz 64Mb memory,
>hash=24M, hashp=10M
> 3. Crafty 15.8 / 2xPPro233 / 4:01 / 8:42 / 14:46
> everything found on ply 11
>
>hash=6M, hashp=2.5M
> Crafty 15.8 / 2xPPro233 / - / - / 3:11
> 10-> 3:31 3.84 1. g4 Ra5 2. gxf5 gxf5 3. Nb7 Bxb7
> 4. Bxb7 a3 5. Qf3 f4 6. b6
>
>
>The above list is updated (DarkThought and Crafty added)
>The curious thing is that Crafty with (6M/2.5M) hashsize does
>better than Crafty with (24M/10M) hashsize.
>This looks somewhat odd. T saw this behavior several times when
>analyzing for a longer time (10 min).
>In principal a bigger hashsize should result in a shorter
>solution time or in a better evaluation. But I see again and again
>the contrary.
>
>By the way Crafty 15.8 with 2 threads does *not* solve fine70.
>Crafty 15.8 with 1 thread gives the solution Kb1.
>
>Kind regards
>Bernhard
yes... I am looking at this one. fine 70 is a *horrible* hashing
position... the position takes a 26 ply search + captures to see
winning the first pawn. Crafty generally solves it at depth=18
because of hashing, but changing the hash table size changes this is
dramatic ways. But with SMP, it doesn't solve it any more, most likely
due to hashing... I'm looking at it, but the tree is too big to make
much sense of it...
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.