Leto wrote:Looking forward to Houdini 4. The big question is will an engine be able to overcome Houdini 3 before Houdini 4 arrives?
Hmm..actually the latest versions of Stockfish are already able to beat H3 in long-time control games quite frequently...not sure though....needs more Testing by the professional Testers here.
Not likely. It has been demonstrated many times over the years that engines don't benefit much from the longer time controls. If engine X performs 50 elo higher than engine Z at blitz, chances are very good that engine X will also perform around 50 elo higher than engine Z at longer time controls.
You are wrong about that. Even if both engines scale equally well, the difference in ELO between them will shrink with increasing depth.
I don't believe Stockfish is close to Houdini 3 however. What I have seen is a nice improvement but not the 100 ELO or so required to catch Houdini 3. If they don't run out of steam they may yet make up the ground.
I do believe that Stockfish is one of the most scalable program out there but I have not seen any program have such great scalability it can make up for 100 ELO going from 5 minutes to a couple of hours. It's possible that it will be very close to Houdini 3 given a day per move perhaps but we can only speculate and I have no knowledge of how Stockfish 3 in particular scales.
Stockfish may however have a superior MP implementation. I'm not saying it does, but that is another variable.
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.
Leto wrote:Looking forward to Houdini 4. The big question is will an engine be able to overcome Houdini 3 before Houdini 4 arrives?
Hmm..actually the latest versions of Stockfish are already able to beat H3 in long-time control games quite frequently...not sure though....needs more Testing by the professional Testers here.
Not likely. It has been demonstrated many times over the years that engines don't benefit much from the longer time controls. If engine X performs 50 elo higher than engine Z at blitz, chances are very good that engine X will also perform around 50 elo higher than engine Z at longer time controls.
You are wrong about that. Even if both engines scale equally well, the difference in ELO between them will shrink with increasing depth.
I don't believe Stockfish is close to Houdini 3 however. What I have seen is a nice improvement but not the 100 ELO or so required to catch Houdini 3. If they don't run out of steam they may yet make up the ground.
I do believe that Stockfish is one of the most scalable program out there but I have not seen any program have such great scalability it can make up for 100 ELO going from 5 minutes to a couple of hours. It's possible that it will be very close to Houdini 3 given a day per move perhaps but we can only speculate and I have no knowledge of how Stockfish 3 in particular scales.
Stockfish may however have a superior MP implementation. I'm not saying it does, but that is another variable.
Absolutely ! Its certainly very scalable and has a superior MP implementation.
Latest versions of Stockfish and my 6-core certainly seem to suit each other... have myself beaten other strong players who were using Houdini 3, in long-time control games but not always, some draws, mostly against the guys using 12-16 cores.
