Here my first findings: with DeepSaros 2.3f i didn't have any error,crash or loose on time yet!It's playing with the old Robbobases.This on a core i7 970 6core (6c) system @4.2Ghz with HT OFF.Engines without 6c has played with 4cores!
I see people directly complain that this is not a good engine..well did you test it out..when you play with default settings Deepsaros loose badly from Houdini..but i see a few games where he play well with long time control..so Roberto told me ,when you use fast games it's better to use a higher TimeTank value 40-50. So i tried till i find better results for my system!
I have go from his 40-50 to 2000 even 5000.But finally i come back to 120 as best from these 2 days testing.
Here with TimeTank(TT)=120 i see that Deepsaros begin to win from Houdini 4cores!
Therefore i thought i have to go higher if he plays against Houdini 6cores.
TT=120
Core i7 970 @4.2Ghz , Blitz 5m 0
DeepSaros ver.2.3f 6c 2800 - Houdini 1.5 x64 2800 5.5 - 4.5 +2/-1/=7 55.00%
DeepSaros ver.2.3f 6c 2800 - Houdini 1.5a x64SD12 2800 6.0 - 4.0 +3/-1/=6 60.00%
DeepSaros ver.2.3f 6c 2800 - Houdini 1.5 x64SD16 2800 5.5 - 4.5 +2/-1/=7 55.00%
Core i7 970 @4.2Ghz , Blitz 5m 0
DeepSaros ver.2.3f 6c 2800 - Ivanhoe 52hUSo1-x64 6c 2800 5.0 - 5.0 +1/-1/=8 50.00%
But as you can see at TT=300 he still loose badly against 6core Houdini
TT= 300
Core i7 970 @4.2Ghz , Blitz 5m 0
1 Houdini 1.5a x64SD13 6c 2800 ½½1½10½111 7.0/10
2 DeepSaros ver.2.3f 6c 2800 ½½0½01½000 3.0/10
Core i7 970 @4.2Ghz , Blitz 5m 0
1 Houdini 1.5a x64 6c 2800 ½10½1½½111 7.0/10
2 DeepSaros ver.2.3f 6c 2800 ½01½0½½000 3.0/10
And at TT=1000 i get this..already much closer! (Have to test more out)
TT=1000
Core i7 970 @4.2Ghz , Blitz 5m 0
1 Houdini 1.5a x64 6c 2800 ½½½½½½½½1½ 5.5/10
2 DeepSaros ver.2.3f 6c 2800 ½½½½½½½½0½ 4.5/10
But at TT=2000 was to high and the score get worse,so i stop it.
TT=2000
Core i7 970 @4.2Ghz , Blitz 5m 0
1 Houdini 1.5a x64 6c 2800 ½½01½111 5.5/8
2 DeepSaros ver.2.3f 6c 2800 ½½10½000 2.5/8
Then i tried closer to the TT=1000 again ,again to see if i get better result.
But it was not the case.
TT=1250
Core i7 970 @4.2Ghz , Blitz 5m 0
1 Houdini 1.5a x64 6c 2800 1101½½ 4.0/6
2 DeepSaros ver.2.3f 6c 2800 0010½½ 2.0/6
TT=750
Core i7 970 @4.2Ghz , Blitz 5m 0
1 Houdini 1.5a x64SD13 6c 2800 1½½111 5.0/6
2 DeepSaros ver.2.3f 6c 2800 0½½000 1.0/6
Finally i go back to TT=120 ,because i have most results with it.
TT=120
Core i7 970 @4.2Ghz , Blitz 5m 0
1 Houdini 1.5a x64SD13 6c 2800 01½½½1½½½1 6.0/10
2 DeepSaros ver.2.3f 6c 2800 10½½½0½½½0 4.0/10
As you can see,this is just a first step,i need many time to find out what will be the best TimeTank value for this system.So for every system you will have a different TT.And can only be find out by testing it!
I have to find a averange,because TT=120 can be good for playing against Houdini but not against other engines?!
DeepSaros has one big problem,his bad time control that we also see in the other IvanHoe's..but the most i like is,it's stable,it running here without any problem as i see many times problem with other Ivanhoe's.
Roberto gives the sourse..so it must be easy for a programmer to fix this TimeControl ,that it use same time as Houdini!
Ipman.
_________________
nothing is ever truly lost , just misplaced and awaiting us