Author: Dann Corbit
Date: 16:07:51 01/11/00
Go up one level in this thread
On January 11, 2000 at 18:41:35, Andrew Walker wrote: > >For fun I picked one of the positions from your list at random and >analysed it overnight on a 450Mhz machine using Crafty 17.5 >The nps count seems a bit low, it may have been slowing down at the >larger depths. Anyway, it shows that in this position crafty does eventually >get to a more equal score. Is anyone willing to try this or others >with crafty or other progs such as rebel? [snip] [D]r1bq1r1k/p4Ppp/1p3n2/2p1n3/2B5/2P1PN2/P3QPPP/R1B2RK1 b [snip] > 17 228:06 -0.46 1. ... Bg4 2. e4 Nh5 3. Bd5 Bxf3 4. > gxf3 Ng6 5. Qd1 Nhf4 6. Bxf4 Nxf4 7. > Kh1 Nxd5 8. exd5 Qf6 9. c4 Rxf7 <HT> > 17-> 261:55 -0.46 1. ... Bg4 2. e4 Nh5 3. Bd5 Bxf3 4. > gxf3 Ng6 5. Qd1 Nhf4 6. Bxf4 Nxf4 7. > Kh1 Nxd5 8. exd5 Qf6 9. c4 Rxf7 <HT> > 18 507:18 -0.48 1. ... Bg4 2. e4 Nh5 3. Bd5 Bxf3 4. > gxf3 Ng6 5. Qd1 Nhf4 6. Bxf4 Nxf4 7. > Kh1 Nxd5 8. exd5 Qf6 9. c4 Rxf7 10. > Re1 Qc3 11. Rc1 Qxf3+ 12. Qxf3 Rxf3 > 18-> 613:30 -0.48 1. ... Bg4 2. e4 Nh5 3. Bd5 Bxf3 4. > gxf3 Ng6 5. Qd1 Nhf4 6. Bxf4 Nxf4 7. > Kh1 Nxd5 8. exd5 Qf6 9. c4 Rxf7 10. > Re1 Qc3 11. Rc1 Qxf3+ 12. Qxf3 Rxf3 [snip] Yes, exactly. Notice how the score is steadily dropping. The supercomputer came up with about half that number: r1bq1r1k/p4Ppp/1p3n2/2p1n3/2B5/2P1PN2/P3QPPP/R1B2RK1 b - - bm Bg4; ce -28; Which is much more in line with what would be expected from theory (the position is about even for a frequently played opening position). We will see this behavior all the time for gambit type openings. It takes the computer a long time to see the benefits (if it ever does). Thank you very much for looking into this for me. It is yet another confirmation of my suspicions about the superiority of the supercomputer solution. Interestingly, we are using only a tiny fraction of the supercomputer's true power and the positions are only being run for a few minutes. Pretty awesome, isn't 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.