Vinvin wrote:Very interresting tournament !
May be more rounds are required to get consistent results ...
Please, re-run the same tournament 3 more times ...
Well, thanks, possibly in January I could do that.
Best,
PS: Please take a look at todays writings in my group.
Bright 0.4a w32 won an event. I thought I should not
add it to the tournament because it would claim last place.
With Naum you have to edit a file to make it use more than 1 CPU. Check in task manager and make sure all CPU's are being used. If not then you have to edit the file that the engine runs with to make sure it uses all available CPU's.
Do I understand correctly, that in this case, Rybka 3 outscored Robbolito very convicingly?
If so, then what is there for Rybka to worry about Robbolito? Just make a few improvements with Rybka 4, and it will again be, safely above robbolito!
S.Taylor wrote:Do I understand correctly, that in this case, Rybka 3 outscored Robbolito very convicingly?
If so, then what is there for Rybka to worry about Robbolito? Just make a few improvements with Rybka 4, and it will again be, safely above robbolito!
7 games per engine is a far cry from a statistically significant sample.
S.Taylor wrote:Do I understand correctly, that in this case, Rybka 3 outscored Robbolito very convicingly?
If so, then what is there for Rybka to worry about Robbolito? Just make a few improvements with Rybka 4, and it will again be, safely above robbolito!
2 points to note:
1) If same start positions are played with reversed colours, results may change significantly.
2) This particular RobboLito is not optimized for fischer time controls and sometimes even loses on time. There is a RobboLito 0.085et4 circulating out there with better time management.
S.Taylor wrote:Do I understand correctly, that in this case, Rybka 3 outscored Robbolito very convicingly?
If so, then what is there for Rybka to worry about Robbolito? Just make a few improvements with Rybka 4, and it will again be, safely above robbolito!
1) If same start positions are played with reversed colours, results may change significantly.
Agreed.
2) This particular RobboLito is not optimized for fischer time controls and sometimes even loses on time. There is a RobboLito 0.085et4 circulating out there with better time management.
I think Harun is using the last executable modified by Kranium with a better time management (0.085e4)
I would add:
3. Keep in mind that the winner of this tournament is using 4 cores. Robbolito got the same score of Naum 4 with 4 threads. That is pretty impressive.
4. More games are needed to determine the real strenght of Robbolito.