Author: Paul Richards
Date: 09:49:00 05/07/99
I setup LGGold 2.0 as an engine under Fritz 5.32 on a PII 400 machine with 128MB to test the position below given earlier by Mr. Monkman: FEN "r4rk1/pp2qpp1/nnp1p2p/2b1P1N1/2P2P2/6N1/PP2Q1PP/R1B2R1K w - - 0 1" The correct solution to this position is 1. f5 hxg5 2. f6 Qc7 3. fxg7 Kxg7 4. Bxg5, etc., at which point any program will see the win quickly. Fritz is a fast searcher, but for some reason it has to go well into a 15 ply search to find the winning line, which takes many hours even on a PII 400 for this position. However once supplied with 1. f5 it finds the rest of the line before 12 ply. LGGold 2.0 smokes this position faster than any other engine I've seen so far. It finds the solution at 12 ply, which is actually where it SHOULD be found, but for some reason Fritz has to search to almost 16 ply to see the solution. LGGold 2.0 finds the solution in 8 minutes, 36 seconds. The LGGold log file gives 12 72 516 223292230 f4f5, which agrees with the Fritz interface that the nps was in the neighborhood of 428,000! (BTW, first parameter is depth, third is time, fourth is nodes, and fifth is pv, what does the second parameter represent? Selective depth?). While I have seen Fritz do 550,000 on this machine, it varies according to the position, and for this one Fritz gets 349,000 nps compared to 428,000 for LGGold. Any program that searches faster than Fritz is damn fast, that's all I can say. It also handles this position much better than Fritz. I look forward to testing LGGold on other positions Fritz has a hard time with.
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.