Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Deep Fritz Speedup

Author: Slater Wold

Date: 22:46:28 11/20/01

Go up one level in this thread


On November 20, 2001 at 23:13:02, Uri Blass wrote:

>On November 20, 2001 at 22:10:09, Slater Wold wrote:
>
>>1 CPU:
>>
>>Deep Fritz - W,S
>>r2qrb1k/1p1b2p1/p2ppn1p/8/3NP3/1BN5/PPP3QP/1K3RR1 w - - 0 1
>>
>>Analysis by Deep Fritz:
>>
>>1.Qg6 Rc8 2.h4
>>  =  (0.06)   Depth: 3/11   00:00:00
>>1.Qg6 Rc8 2.h4
>>  =  (0.06)   Depth: 3/11   00:00:00
>>1.Qg6 Rc8 2.h4
>>  =  (0.06)   Depth: 3/11   00:00:00
>>1.Qg6 Rc8 2.h4
>>  =  (0.06)   Depth: 3/11   00:00:00
>>1.Qg6 Rc8 2.h4 Qc7
>>  =  (-0.06)   Depth: 4/15   00:00:00  1kN
>>1.Qg6 Rc8 2.h4 Qc7
>>  =  (-0.06)   Depth: 4/15   00:00:00  1kN
>>1.Qg6 Rc8 2.h4 Qc7
>>  =  (-0.06)   Depth: 4/15   00:00:00  1kN
>>1.Nde2!
>>  =  (-0.03)   Depth: 4/15   00:00:00  4kN
>>1.Nde2 Rc8 2.Nf4 g5 3.Ng6+ Kh7
>>  =  (0.03)   Depth: 5/17   00:00:00  8kN
>>1.Nce2!
>>  =  (0.06)   Depth: 5/18   00:00:00  16kN
>>1.Nce2 g5 2.h4 g4 3.Nf4 Qc8
>>  =  (0.16)   Depth: 6/20   00:00:00  34kN
>>1.Nce2 e5 2.Nf5 Bxf5 3.Rxf5 Rc8 4.Nc3 Qc7
>>  =  (0.09)   Depth: 7/25   00:00:00  102kN
>>1.Nce2 e5 2.Nf5 Bxf5 3.Rxf5 Rc8 4.Nc3 Qc7 5.Bd5
>>  ²  (0.38)   Depth: 8/26   00:00:00  314kN
>>1.Nce2 e5 2.Nf5 Bxf5 3.Rxf5 Rc8 4.Bd5 Qc7 5.Nc3 b5
>>  ²  (0.38)   Depth: 9/24   00:00:00  668kN
>>1.Nf3!
>>  ²  (0.41)   Depth: 9/29   00:00:01  1487kN
>>1.Nf3! Qa5 2.Nh4 Qg5 3.Qxg5 hxg5 4.Ng6+ Kh7 5.Rxg5 Be7
>>  ²  (0.44)   Depth: 9/29   00:00:02  2617kN
>>1.Nf3 Qa5 2.Nh4 Qg5 3.Qxg5 hxg5 4.Ng6+ Kh7 5.Rxg5 Kh6 6.h4 Be7
>>  ²  (0.38)   Depth: 10/28   00:00:04  4195kN
>>1.Nce2!
>>  ²  (0.41)   Depth: 10/28   00:00:04  4662kN
>>1.Nce2!
>>  ±  (0.72)   Depth: 11/30   00:00:08  8592kN
>>1.Nce2 e5 2.Nf5 Bxf5 3.Rxf5 Qe7 4.Rgf1 d5 5.Bxd5 Nxd5 6.exd5 Qd6
>>  ²  (0.66)   Depth: 12/35   00:00:16  17113kN
>>1.Nf3!
>>  ²  (0.69)   Depth: 12/35   00:00:25  27163kN
>>1.Nf3! Bc6 2.Nh4 g5 3.Ng6+ Kg7 4.Nxf8 Rxf8 5.h4 Kf7 6.hxg5 hxg5
>>  ±  (0.97)   Depth: 12/35   00:00:43  45849kN
>>1.Nf3 Qa5 2.Nh4 Qg5 3.Ng6+ Kh7 4.Qf3 Qa5 5.Nxf8+ Rxf8 6.Qg3 Nh5
>>  ±  (0.97)   Depth: 13/35   00:01:18  83931kN
>>
>>(W,  20.11.2001)
>>
>>2 CPU's:
>>
>>Deep Fritz - W,S
>>r2qrb1k/1p1b2p1/p2ppn1p/8/3NP3/1BN5/PPP3QP/1K3RR1 w - - 0 1
>>
>>Analysis by Deep Fritz:
>>
>>1.Qg6 Rc8
>>  =  (0.06)   Depth: 2/8   00:00:00
>>1.Qg6 Rc8 2.h4
>>  =  (0.06)   Depth: 3/11   00:00:00
>>1.Qg6 Rc8 2.h4
>>  =  (0.06)   Depth: 3/11   00:00:00
>>1.Qg6 Rc8 2.h4
>>  =  (0.06)   Depth: 3/11   00:00:00
>>1.Qg6 Rc8 2.h4
>>  =  (0.06)   Depth: 3/11   00:00:00
>>1.Qg6 Rc8 2.h4 Qc7
>>  =  (-0.06)   Depth: 4/15   00:00:00  1kN
>>1.Qg6 Rc8 2.h4 Qc7
>>  =  (-0.06)   Depth: 4/15   00:00:00  1kN
>>1.Nde2!
>>  =  (-0.03)   Depth: 4/15   00:00:00  4kN
>>1.Nde2 Rc8 2.Nf4 g5 3.Ng6+ Kh7
>>  =  (0.03)   Depth: 5/14   00:00:00  8kN
>>1.Nce2!
>>  =  (0.06)   Depth: 5/18   00:00:00  17kN
>>1.Nce2 g5 2.h4 g4 3.Nf4 Qc8
>>  =  (0.16)   Depth: 6/20   00:00:00  34kN
>>1.Nce2 e5 2.Nf5 Bxf5 3.Rxf5 Rc8 4.Nc3 Qc7
>>  =  (0.09)   Depth: 7/25   00:00:00  103kN
>>1.Nce2 e5 2.Nf5 Bxf5 3.Rxf5 Rc8 4.Nc3 Qc7 5.Bd5
>>  ²  (0.38)   Depth: 8/25   00:00:00  317kN
>>1.Nce2 e5 2.Nf5 Bxf5 3.Rxf5 Rc8 4.Bd5 Qc7 5.Nc3 b5
>>  ²  (0.38)   Depth: 9/24   00:00:00  661kN
>>1.Nf3!
>>  ²  (0.41)   Depth: 9/27   00:00:01  1599kN
>>1.Nf3! Qa5 2.Nh4 Qg5 3.Qxg5 hxg5 4.Ng6+ Kh7
>>  ²  (0.44)   Depth: 9/27   00:00:02  3413kN
>>1.Nf3 Qa5 2.Nh4 Qg5 3.Qxg5 hxg5 4.Ng6+ Kh7 5.Nxf8+ Rxf8 6.Rd1 Ne8
>>  ²  (0.38)   Depth: 10/30   00:00:03  4601kN
>>1.Nce2!
>>  ²  (0.41)   Depth: 10/30   00:00:04  5076kN
>>1.Nce2!
>>  ±  (0.72)   Depth: 11/29   00:00:06  8587kN
>>1.Nce2 e5 2.Nf5 Bxf5 3.Rxf5 Qe7 4.Rgf1 d5 5.Bxd5 Nxd5 6.exd5 Qd6
>>  ²  (0.66)   Depth: 12/31   00:00:12  16798kN
>>1.Nf3!
>>  ²  (0.69)   Depth: 12/32   00:00:20  26626kN
>>1.Nf3! Bc6 2.Nh4 g5 3.Ng6+ Kg7 4.Nxf8 Rxf8 5.h4 Kf7 6.hxg5 hxg5
>>  ±  (0.97)   Depth: 12/33   00:00:33  45510kN
>>1.Nf3 Qa5 2.Nh4 Qg5 3.Ng6+ Kh7 4.Qf3 Qa5 5.Nxf8+ Rxf8 6.Qg3 Nh5
>>  ±  (1.00)   Depth: 13/35   00:00:59  81622kN
>>
>>(W,  20.11.2001)
>>
>>
>>What's the speedup on that?
>
>25/20=1.25
>
>Deep Fritz one processor needed 25 seconds to find Nf3 when Deep Fritz one
>processor needed 20 seconds to find Nf3.
>
>The speed up seems to be slightly better at longer time control(Deep Fritz
>needed 59 seconds to get a score at depth 13 with 2 proccesors against 78
>seconds with one processor and almost the same pv
>
>78/59>1.25
>
>Uri

So to answer some people's questions out there, yes, Chessbase lied.

I have had 3 SMP systems, and all have run DF.  None of them have _EVER_ gotten
an 80% speedup.

The 2xP31.0Ghz got the best speedup; 1.4 on most positions.

The 2xXP1.53Ghz gets the worst speedup; 1.25 (1.3 in my longer tests) on most
positions.

I don't know many people who use DF, but the ones I have talked to, get lousy
results too.



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.