Author: Richard A. Fowell (fowell@netcom.com)
Date: 22:11:03 01/24/98
Go up one level in this thread
On January 24, 1998 at 22:26:15, Carsten Kossendey wrote:
>On January 24, 1998 at 20:11:39, Jay Scott wrote:
>
>>Then I tried MacChess 4.0. Searching to depth 7 took it less
>>than one second and didn't give an accurate speed, so I went
>>to depth 9. It reported ~250K nps. From my experience, that's
>>at the lower end of its normal range--it's what you'd expect
>>in this position.
>>
>>Considering chip performance alone, I'd expect crafty to run
>>rather faster on a G3 Mac than on a PPro 200.
The classic tradeoff in chess is - look at lots of positions
superficially?
Or look at fewer positions carefully?
For instance, on my 7300/180:
- HIARCS 6.0 is in 22,000 nps range,
- Crafty 14.5 is in the 40,000 nps range,
- MacChess is in the 150,000 nps range.
However, in terms of strength,
HIARCS > Crafty > MacChess (IMHO).
As I understand it, HIARCS and Crafty probably just have much more
complex evaluation functions than MacChess.
It seems there is a large difference even between
Crafty 12.9 and Crafty 14.5 here - not so much in nps,
but in time to search to 9 ply - I'm getting a
factor of three in time, with 14.5 taking longer.
<snip>
Here's my canonical "a3" at depth 9 test output for Crafty 14.5:
White(1): book off
book file disabled.
White(1): hash 6M
hash table memory = 6M bytes.
White(1): hashp 2M
pawn hash table memory = 1.3M bytes.
White(1): ponder off
pondering disabled.
White(1): sd 9
search depth set to 9.
White(1): a3
time used: 49.35
time surplus 0.00 time limit 30.00 (3:00)
depth time score variation (1)
6 0.37 -0.02 d5 d4 Qd6 Nf3 Bf5 Bg5
6-> 0.82 -0.02 d5 d4 Qd6 Nf3 Bf5 Bg5
7 1.42 0.06 d5 d4 Qd6 Nc3 Bf5 Bg5 Nc6
7-> 2.49 0.06 d5 d4 Qd6 Nc3 Bf5 Bg5 Nc6
8 4.85 -0.01 d5 d4 Qd6 Nc3 Bf5 Bg5 Nf6 Nf3
8 17.29 0.00 e5 Nc3 Nf6 e4 d5 Nf3 Nxe4 Nxe4 <HT>
8-> 20.85 0.00 e5 Nc3 Nf6 e4 d5 Nf3 Nxe4 Nxe4 <HT>
9 30.47 0.01 e5 Nc3 Nf6 e4 d5 exd5 Nxd5 Qh5 Nc6
Bb5
9-> 51.72 0.01 e5 Nc3 Nf6 e4 d5 exd5 Nxd5 Qh5 Nc6
Bb5
time: 51.80 cpu:100% mat:0 n:2155330 nps:41608
ext-> checks:32991 recaps:13074 pawns:5 1rep:31843
predicted:0 nodes:2155330 evals:908180
endgame tablebase-> probes done: 0 successful: 0
hashing-> trans/ref:26% pawn:92% used:w91% b83%
Black(1): e5
time used: 51.92
Crafty 12.9, (which Carsten might have juiced up a little) under
the same conditions, takes much less time to search to depth 9, but
with fewer positions, and only slightly more nodes/second:
Crafty v12.9
White(1): book off
book file disabled.
White(1): hash 6M
hash table memory = 6M bytes.
White(1): hashp 2M
pawn hash table memory = 1M bytes.
White(1): sd 9
search depth set to 9.
White(1): ponder off
pondering disabled.
White(1): a3
time used: 45.65
clearing transposition table
clearing pawn hash tables
time surplus 0.00 time limit 30.00
depth time score variation (1)
6-> 0.45 -0.338 d5 d4 Bf5 Bf4 Nf6 Nf3
7 0.58 ++ d5!!
7 0.98 -0.090 d5 Nf3 Nc6 d4 Bf5 Bf4 Nf6
7-> 1.25 -0.090 d5 Nf3 Nc6 d4 Bf5 Bf4 Nf6
8 1.35 -- d5
8 2.81 -0.378 d5 Nf3 Nc6 d4 Bg4 Bf4 Bxf3 exf3
Nf6
8 5.50 -0.254 e5 Nc3 Nc6 Nf3 d5 d4 exd4 Nxd4 Bc5
<HT>
8-> 5.83 -0.254 e5 Nc3 Nc6 Nf3 d5 d4 exd4 Nxd4 Bc5
<HT>
9 7.88 ++ e5!!
9 13.88 -0.099 e5 Nc3 Nf6 Nf3 Nc6 d4 exd4 Nxd4
Nxd4
Qxd4 Be7 <HT>
9-> 16.96 -0.099 e5 Nc3 Nf6 Nf3 Nc6 d4 exd4 Nxd4
Nxd4
Qxd4 Be7 <HT>
time: 17.13 cpu:100% mat:0 n:760250 nps:44381
ext-> checks:9234 recaps:3852 pawns:210 1rep:11425
predicted:0 nodes:760250 evals:340558
endgame tablebase-> probes done: 0 successful: 0
hashing-> trans/ref:38% pawn:92% used:w53% b49%
Black(1): e5
time used: 17.25
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.