Scaling results CCRL list

[Rebel]

A comparison between the 40/4 list and 40/40 list and how well (or bad) engines scale. Minimum games 300.

http://rebel13.nl/scaling_unsorted.txt [sequential]
http://rebel13.nl/scaling_sorted.txt [sorted on scaling]

Not sure if it has any value because of the (alleged) bad scaling of the top programs especially on 4CPU. Big exception:

Code: Select all

Engine (Minimum games 300)               40/4   40/40  Scaling
Bouquet 1.8 64-bit 4CPU                  3075    3155   80

[Uri Blass]

A comparison between the 40/4 list and 40/40 list and how well (or bad) engines scale. Minimum games 300.

http://rebel13.nl/scaling_unsorted.txt [sequential]
http://rebel13.nl/scaling_sorted.txt [sorted on scaling]

Not sure if it has any value because of the (alleged) bad scaling of the top programs especially on 4CPU. Big exception:

Code: Select all

Engine (Minimum games 300)               40/4   40/40  Scaling
Bouquet 1.8 64-bit 4CPU                  3075    3155   80

something seem strange with this engine
Bouguet 1.8 64 bit 4 CPU seem to be weaker at 40/4 relative to other engines in the same family including Bouquet 1.6 64-bit


http://www.computerchess.org.uk/ccrl/40 ... +opponents

Edit:I think that it may be interesting to see if Bouquet1.8 does not have a bug in time management that cause it to play too fast in time trouble or simply not to use 3 minutes of the time because it can explain the good scaling.

[Rebel]

A comparison between the 40/4 list and 40/40 list and how well (or bad) engines scale. Minimum games 300.

http://rebel13.nl/scaling_unsorted.txt [sequential]
http://rebel13.nl/scaling_sorted.txt [sorted on scaling]

Not sure if it has any value because of the (alleged) bad scaling of the top programs especially on 4CPU. Big exception:

Code: Select all

Engine (Minimum games 300)               40/4   40/40  Scaling
Bouquet 1.8 64-bit 4CPU                  3075    3155   80

Doing the same for the CEGT the scaling numbers makes much more sense.

http://rebel13.nl/cegt_scaling_unsorted.txt [sequential]
http://rebel13.nl/cegt_scaling_sorted.txt [sorted on scaling]

Probably the difference between Bayeselo (CCRL) and Ordo (CEGT)

[Uri Blass]

A comparison between the 40/4 list and 40/40 list and how well (or bad) engines scale. Minimum games 300.

http://rebel13.nl/scaling_unsorted.txt [sequential]
http://rebel13.nl/scaling_sorted.txt [sorted on scaling]

Not sure if it has any value because of the (alleged) bad scaling of the top programs especially on 4CPU. Big exception:

Code: Select all

Engine (Minimum games 300)               40/4   40/40  Scaling
Bouquet 1.8 64-bit 4CPU                  3075    3155   80

Doing the same for the CEGT the scaling numbers makes much more sense.

http://rebel13.nl/cegt_scaling_unsorted.txt [sequential]
http://rebel13.nl/cegt_scaling_sorted.txt [sorted on scaling]

Probably the difference between Bayeselo (CCRL) and Ordo (CEGT)

For some reason I miss Movei there.

CEGT 40/20 all versions
1090 Movei 0.08.438 P10 2462 13 13 1419 54.4% 2430 34.7%
1137 Movei 0.08.438 2440 13 13 1678 49.3% 2446 32.1%

CEGT 40/4 all versions

1406 Movei 00.8.438 2415 10 10 3300 53.3% 2390 28.8%

2 things
1)I find it strange that I do not find P10 at 40/4 and I suspect that somebody practically unified the 2 personalities because if I remember correctly CEGT tested both also in the blitz list.

2)I find it strange that I do not find Movei0.08.438 in the list that you give because I could expect to find it at least with the number 25(I believe the real number was bigger).

Edit:I see that the problem is that
00.8!=0.08 and 00.8 is for 40/4 and 0.08 is for 40/20