Actually even with real clean code, it is not as strong as Rybka 3 if you change a few parameters that are default on R3. I have been trying for a few days to compile 100 games against R3 at 10 0 time controls. I gave up on the task as the engine crashes every 2 or 3 games at 10 0. In the 14 games I managed to collect R3 was clearly ahead by 3 wins, but I finally gave up because of too many crashes. It will probably score pretty well against R3 but not come out ahead if left to continue ... and that is normal since it is basically a messy buggy R3.
M ANSARI wrote:Actually even with real clean code, it is not as strong as Rybka 3 if you change a few parameters that are default on R3. I have been trying for a few days to compile 100 games against R3 at 10 0 time controls. I gave up on the task as the engine crashes every 2 or 3 games at 10 0. In the 14 games I managed to collect R3 was clearly ahead by 3 wins, but I finally gave up because of too many crashes. It will probably score pretty well against R3 but not come out ahead if left to continue ... and that is normal since it is basically a messy buggy R3.
That is the kind of stuff I saw from the original IPPOLIT when I tried it. _VERY_ strong. But if you win 3 games by outplaying your opponent, and lose 2 due to crashes, things are not quite as rosy then.
Actually even with real clean code, it is not as strong as Rybka 3 if you change a few parameters that are default on R3. I have been trying for a few days to compile 100 games against R3 at 10 0 time controls. I gave up on the task as the engine crashes every 2 or 3 games at 10 0. In the 14 games I managed to collect R3 was clearly ahead by 3 wins, but I finally gave up because of too many crashes. It will probably score pretty well against R3 but not come out ahead if left to continue ... and that is normal since it is basically a messy buggy R3.
Very strange. I have been running RobboLito for 8 days now (1627 2 12 games) without restarting it and have not observed a single crash. It appears very stable. This is on a Core Duo (1 CPU).
Actually I did not try Robolito, but I assume it is based on Ippolit. I wanted to test for myself to see if this was a Rybka clone. Most likely it is a more stable Ippolit and will thus play pretty much the same but not crash as much. I already have a registered legal Rybka 3 that has additional evaluation tables and MP support among other things, so this engine is not interesting to me even if it is stable and it was legal. I do feel bad for Vas because that takes some of his competitive advantage away by theft, but he seems pretty upbeat about R4 and what it has to offer. So I really look forward to R4 and what it will bring.