TalkChess.com

Marc, Michael, George, Vinvin,Jouni

do you have Opening Theme games using these "bandits".

thanks.

Eizenhammer wrote:

Dr.Wael Deeb wrote:
No comment....
This must be the Rybka 5 source code stolen....otherwise I have no explanation for the result

Your never ending flooding is like an indoctrination, you have stated your opinion about 2000 times now, can't you leave some room for others??
Everybody knows now, that you have been disappointed by Vas, nobody knows why, maybe because of some book stuff, who cares, but please end this stupid campaign.

No, please don't.

S.Taylor wrote:

Eizenhammer wrote:

Dr.Wael Deeb wrote:
No comment....
This must be the Rybka 5 source code stolen....otherwise I have no explanation for the result

Your never ending flooding is like an indoctrination, you have stated your opinion about 2000 times now, can't you leave some room for others??
Everybody knows now, that you have been disappointed by Vas, nobody knows why, maybe because of some book stuff, who cares, but please end this stupid campaign.

No, please don't.

Let him bark like an old dog Shimon,he doesn't bother me at all....after all,he's an ordinary Rybka fanboy....

He's annoyed because I've stated my opinion over 2000 times now....Bob for example is stating his opinion about certain issues over 10 000 time but he doesn't attack him....
Dr.D

After testing this for some time I have no doubt Robbolito is a clone of Rybka 3. It gets better results than Rybka 3 at short time controls and I would put the reason as different time management than R3, and due to R3 having a 15 contempt at default. I have tried to figure out why would an engine be stronger than its original, and from the last couple of days of testing I have seen that it is due to a different time management. At first I couldn't figure it out, but then when I look at the games I can see that usually R3 loses due to having almost no time in the end. In some cases R3 is down to seconds while the Robbo still has minutes. Somehow that seems to favor Robbo and playing fast in the beginning does not affect its results. So far in my testing that has held true with Robbolito advantage decreasing linearily as time controls go up. The latest testing I have is at 16_1 ponder on using 4Ghz cores for each 64bit engine (e4), and after 30 games out of a 100 R3 is actually ahead. I guess Wael or someone else that is good at tuning engine parameters can probably figure out a better setting for default R3 and you will find that R3 is actually equal if not slightly stronger at all time controls ... only need to change the time management of R3. R3 time management was I guess designed against non Rybka engines. I will try to see how changing the time management of R3 will affect results against Robbolito.

One more thing, it could be that Robbo has a faster execution of the engine code since it is unencrypted and unobfuscated. I know that R3 code is obfuscated but not sure about encryption in order to put some sort of protection on the source. Vas has mentioned that he left his "door open" in R3 code so he might has skipped the encryption. These things can hurt performance which is more apparent at fast time controls. But at longer time controls the difference should be less important since I would guess the performance hit of obfuscating and encrypting code to be around 5% to 10%. This seems like not much but could be important at fast time control games.

M ANSARI wrote:After testing this for some time I have no doubt Robbolito is a clone of Rybka 3. It gets better results than Rybka 3 at short time controls and I would put the reason as different time management than R3, and due to R3 having a 15 contempt at default. I have tried to figure out why would an engine be stronger than its original, and from the last couple of days of testing I have seen that it is due to a different time management. At first I couldn't figure it out, but then when I look at the games I can see that usually R3 loses due to having almost no time in the end. In some cases R3 is down to seconds while the Robbo still has minutes. Somehow that seems to favor Robbo and playing fast in the beginning does not affect its results. So far in my testing that has held true with Robbolito advantage decreasing linearily as time controls go up. The latest testing I have is at 16_1 ponder on using 4Ghz cores for each 64bit engine (e4), and after 30 games out of a 100 R3 is actually ahead. I guess Wael or someone else that is good at tuning engine parameters can probably figure out a better setting for default R3 and you will find that R3 is actually equal if not slightly stronger at all time controls ... only need to change the time management of R3. R3 time management was I guess designed against non Rybka engines. I will try to see how changing the time management of R3 will affect results against Robbolito.

One more thing, it could be that Robbo has a faster execution of the engine code since it is unencrypted and unobfuscated. I know that R3 code is obfuscated but not sure about encryption in order to put some sort of protection on the source. Vas has mentioned that he left his "door open" in R3 code so he might has skipped the encryption. These things can hurt performance which is more apparent at fast time controls. But at longer time controls the difference should be less important since I would guess the performance hit of obfuscating and encrypting code to be around 5% to 10%. This seems like not much but could be important at fast time control games.

I have used time controls like 4 seconds for every move, to preclude time management, and tested Rybka with contempt=0, and RobboLito still outperforms. The difference at longer time controls is not as pronounced, but it is still there. How did you determine that the Rybka 3 executable was encrypted or obfuscated?

I have played some 5 ply games in Arena to see how these top engines play under this setting. Mystery engine A and Mystery engine B are two of the engines currently disputed. Here are the current results after 4 Rounds.

Engine Score Ry Ry My St Fr My S-B
1: Rybka 2.2 32 bit 16.0/20 ·········· 1=1= 0=0= 1111 1111 1111 128.00
2: Rybka v1.0 Beta.w32 15.0/20 0=0= ·········· =111 11==? 1111 =111 116.50
3: Mystery A 11.0/20 1=1= =000 ·········· 0==1 =110 ==11 98.50
4: Stockfish_151_ja 8.0/20 0000 00==? 1==0 ·········· 01=1 =1== 62.00
5: Fruit-2-3-1 6.0/20 0000 0000 =001 10=0 ·········· =1=1 40.50
6: Mystery B 4.0/20 0000 =000 ==00 =0== =0=0 ·········· 36.50

60 of 150 games played
Name of the tournament: Big Boys Test - 5 Ply
Site/ Country: Hoover, Alabama, USA, United States
Level: 5 Half moves
Hardware: AMD Turion(tm) 64 Mobile Technology ML-32 1794 MHz with 1,024 MB Memory
Operating system: Microsoft Windows XP Home Edition Service Pack 3 (Build 2600)
PGN-File: C:\Program Files\Arena\BigBoy5Ply.pgn
Website:
E-Mail Address:

It seems already clear to me that under these settings that there are major differences between Rybka and the other engines.

regards

Nick

Spacious_Mind wrote:I have played some 5 ply games

If you mean depth=5, no wonder Rybka is ahead,
since the equivalent depth for the Mystery engines would be 8 or 9, and yet they are constrained to 5 !!!!

ernest wrote:

Spacious_Mind wrote:I have played some 5 ply games

If you mean depth=5, no wonder Rybka is ahead,
since the equivalent depth for the Mystery engines would be 8 or 9, and yet they are constrained to 5 !!!!

Its not quite clear with the way Nick has the scoring above.
Maybe Engine A has won the most games.

ernest wrote:

Spacious_Mind wrote:I have played some 5 ply games

If you mean depth=5, no wonder Rybka is ahead,
since the equivalent depth for the Mystery engines would be 8 or 9, and yet they are constrained to 5 !!!!

Cool are you saying that Rybka minimum is 8 ply ? Because that would explain the big difference.

btw Mystery A is not the one being discussed in todays topics, that only leaves..........

ernest wrote:

Spacious_Mind wrote:I have played some 5 ply games

If you mean depth=5, no wonder Rybka is ahead,
since the equivalent depth for the Mystery engines would be 8 or 9, and yet they are constrained to 5 !!!!

No Gerold, the results are pasted from Arena and the score is after the name. Mystery A has 11 from 20 points.

rgds

Nick