Nid Hogge wrote:Well I for one really hope that anthony will be back and give us stronger Zappa.
BUT, If I'd get a penny every time I heard about the "New Rybka Killer", from seferal sources, I would be a very rich man right now.
People don't understand how really strong is Rybka. Zap would need almost +-100 ELO gain to start beating it head to head AND its still going to be tight.
Having said that, I really wish Zappa all the best of luck, and I hope we finally see something that can compete with Rybka.
Competition is alwats better and more enjoying.
The difference on 8 cpu is smaller than 100 elo(zappa performs better with more processors)
Nid Hogge wrote:Well I for one really hope that anthony will be back and give us stronger Zappa.
BUT, If I'd get a penny every time I heard about the "New Rybka Killer", from seferal sources, I would be a very rich man right now.
People don't understand how really strong is Rybka. Zap would need almost +-100 ELO gain to start beating it head to head AND its still going to be tight.
Having said that, I really wish Zappa all the best of luck, and I hope we finally see something that can compete with Rybka.
Competition is alwats better and more enjoying.
The difference on 8 cpu is smaller than 100 elo(zappa performs better with more processors)
Yes, I stand corrected. I had mistakenly looked at Rybka 2.3.2a x64 2CPU vs. Zap!Chess Zanzibar x64 2CPU where the difference is 117 elo according to CEGT.
It's still a hefty 67 points tho.
However we know Zappa scales great with more CPU's, So with 8 cores it should narrow the gap even further.
In conclusion, we must assume both programs gain from 4 to 8 CPU's, and even if Zappa scales better, It's all open, as we don't know the changes made in BOTH programs. Larry said on the Rybka forums that unreleased version is already quite stronger in comparion to Rybka 2.3.2A, and ofcourse we know Zappa has improved as well.
So all in all .. this should a hell of a fight! I expect a very interesting close match.
According to CCRL Rybka 2.3.2a 3b has 3019 I guess 64 bit version would has 3030 (I just didn't see Rybka 2.3.2a 64b 1CPU tested in CCRL) and Rybka 2.3.2a 64b 4CPU has 3118 it mean a improve of 24 elo for each processor, Zanzibar 64 bit 1 CPU has 2903 elo vs 3051 in 4 CPU it gain 148 elo!! 37 elo per processor. But as I said Rybka 2.3.2a 64 Bit must have 3030 or so, instead it have less than 24 elo per processor ..maybe 20 elo , if you scale this pattern and give 20 elo for Rybka and 37 for Zanzibar we have Rybka 2.3.2a 8CPU would have 3190 Zanzibar 8CPU wolud have 3199!! but I guess the new version of zappa is stronger than zanzibar so a Blood bath is waiting for this poor little and claver fish ..... come on Antony Cozzie the force is with you!!!
ArmyBridge wrote:According to CCRL Rybka 2.3.2a 3b has 3019 I guess 64 bit version would has 3030 (I just didn't see Rybka 2.3.2a 64b 1CPU tested in CCRL) and Rybka 2.3.2a 64b 4CPU has 3118 it mean a improve of 24 elo for each processor, Zanzibar 64 bit 1 CPU has 2903 elo vs 3051 in 4 CPU it gain 148 elo!! 37 elo per processor. But as I said Rybka 2.3.2a 64 Bit must have 3030 or so, instead it have less than 24 elo per processor ..maybe 20 elo , if you scale this pattern and give 20 elo for Rybka and 37 for Zanzibar we have Rybka 2.3.2a 8CPU would have 3190 Zanzibar 8CPU wolud have 3199!! but I guess the new version of zappa is stronger than zanzibar so a Blood bath is waiting for this poor little and claver fish ..... come on Antony Cozzie the force is with you!!!
The improvement is not linear in the number of processors.
You calculation is also wrong because even if we assume that it is linear in the number of processors then the advantage of zap from adding 3 processors is 148 elo so the advantage from adding one processor is 148/3 and not 148/4
rrraatttss Ok you are right, I remember that Zap in monster CPU didn´t won the WCC 2006 (Junior did) but it seem that Zappa performence better than Rybka in mp, so AC has really chance of win and I want to see what happen if Zappa is tuned very agressive, it + deep search would destroy Rybka, and dont forget that Zapp team seams is his helped by Hiarcs team
ArmyBridge wrote:According to CCRL Rybka 2.3.2a 3b has 3019 I guess 64 bit version would has 3030 (I just didn't see Rybka 2.3.2a 64b 1CPU tested in CCRL) and Rybka 2.3.2a 64b 4CPU has 3118 it mean a improve of 24 elo for each processor, Zanzibar 64 bit 1 CPU has 2903 elo vs 3051 in 4 CPU it gain 148 elo!! 37 elo per processor. But as I said Rybka 2.3.2a 64 Bit must have 3030 or so, instead it have less than 24 elo per processor ..maybe 20 elo , if you scale this pattern and give 20 elo for Rybka and 37 for Zanzibar we have Rybka 2.3.2a 8CPU would have 3190 Zanzibar 8CPU wolud have 3199!! but I guess the new version of zappa is stronger than zanzibar so a Blood bath is waiting for this poor little and claver fish ..... come on Antony Cozzie the force is with you!!!
Your math is _severely_ flawed. I don't think Rybka 64-bit 0 CPU is stronger than 32-bit 1 CPU, for example...
..... 
come on Antony Cozzie the force is with you!!! 
Ok you are right, I remember that Zap in monster CPU didn´t won the WCC 2006 (Junior did) but it seem that Zappa performence better than Rybka in mp, so AC has really chance of win and I want to see what happen if Zappa is tuned very agressive, it + deep search would destroy Rybka, and dont forget that Zapp team seams is his helped by Hiarcs team