Author: Robert Hyatt
Date: 06:18:26 03/19/98
Go up one level in this thread
On March 19, 1998 at 06:23:45, Roland Pfister wrote: >On March 19, 1998 at 02:24:55, Christophe Theron wrote: > >>On March 18, 1998 at 22:25:24, Robert Hyatt wrote: >> >>>here's crafty running on my notebook, default (small) hash and all: >>> >>> depth time score variation (1) >>> 1 0.01 -4.98 1. Ke3 >>> 1-> 0.01 -4.98 1. Ke3 >>> 2 0.02 -5.14 1. Ke3 Re6+ 2. Kf4 >>> 2 0.03 -5.09 1. Rf4 Re6 >>> 2-> 0.03 -5.09 1. Rf4 Re6 >>> 3 0.04 -4.95 1. Rf4 Re6 2. Kg3 >>> 3-> 0.05 -4.95 1. Rf4 Re6 2. Kg3 >>> 4 0.06 -- 1. Rf4 >>> 4 0.06 -5.37 1. Rf4 Rg2+ 2. Kf1 Rd2 3. Rf2 Rxf2+ >>> 4. Kxf2 >>> 4 0.08 ++ 1. Rd1+!! >>> 4 0.08 -5.06 1. Rd1+ Rg1 2. Rd8 Rg2+ 3. Ke3 Kg1 >>> 4-> 0.08 -5.06 1. Rd1+ Rg1 2. Rd8 Rg2+ 3. Ke3 Kg1 >>> 5 0.10 ++ 1. Rd1+!! >>> 5 0.11 -4.48 1. Rd1+ Rg1 2. Rd6 Ra1 3. Rg6 Ra3 4. >>> Rg8 >>> 5-> 0.12 -4.48 1. Rd1+ Rg1 2. Rd6 Ra1 3. Rg6 Ra3 4. >>> Rg8 >>> 6 0.14 -4.64 1. Rd1+ Rg1 2. Rd6 Ra1 3. Rg6 a5 4. >>> bxa5 Rxa5 5. Ke3 >>> 6-> 0.15 -4.64 1. Rd1+ Rg1 2. Rd6 Ra1 3. Rg6 a5 4. >>> bxa5 Rxa5 5. Ke3 >>> 7 0.22 -4.61 1. Rd1+ Rg1 2. Rd6 Ra1 3. Re6 Ra2+ >>> 4. Kf3 Ra3+ 5. Ke4 Ra2 >>> 7-> 0.23 -4.61 1. Rd1+ Rg1 2. Rd6 Ra1 3. Re6 Ra2+ >>> 4. Kf3 Ra3+ 5. Ke4 Ra2 >>> 8 0.30 -4.48 1. Rd1+ Rg1 2. Rd6 Ra1 3. Re6 Ra2+ >>> 4. Kf3 Ra3+ 5. Ke4 Ra1 6. Rf6 >>> 8-> 0.43 -4.48 1. Rd1+ Rg1 2. Rd6 Ra1 3. Re6 Ra2+ >>> 4. Kf3 Ra3+ 5. Ke4 Ra1 6. Rf6 >>> 9 0.66 ++ 1. Rd1+!! >>> 9 0.68 Mat07 1. Rd1+ Rg1 2. Rf1 a5 3. bxa5 b4 4. >>> a6 Rxf1+ 5. Kxf1 b3 6. a7 b2 7. >>>a8=Q# >>> 9-> 0.73 Mat07 1. Rd1+ Rg1 2. Rf1 a5 3. bxa5 b4 4. >>> a6 Rxf1+ 5. Kxf1 b3 6. a7 b2 7. >>>a8=Q# >>> time: 0.74 cpu:83% mat:-2 n:51447 nps:10000 >> > >something is wrong here: >51447 nodes in 0.74 seconds is never 10000 nps rather 70000 nps. >or should it be 5.14 seconds ;-) > >>If Crafty does only 10000 nps on your notebook, this computer is quite >>slow. So Crafty is probably the best program so far on this position, > ^^^^ >see above it's an old bug. I used to time in units of 1 second, and obviously if I use less than a second, computing NPS will get me an arithmetic overflow when I divide by zero. I've never fixed it. So if Crafty computes for less than a second, it leaves NPS alone. Somewhere it is apparently initialized to 10K, probably a carryover from when it was searching 8-10K early in the development... > >>doing better than Genius. >> >> >> Christophe
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.