[CCB] Danasah 3.97b - 2559!

[swami]

I've added the Danasah 3.97b to the same GBT 1 tournament gauntlet against the 100 distinct opponents to see the progress, here are the results:

Danasah 3.97b - Score: 54.5/100

Rank: 42

Previous Version Stats:

Danasah 3.70- Score: 42.5/100

Rank: 62

Notes: This is also a good improvement, I'd guess that the latest version main focus was on improvement in blitz. And It succeeded convincingly!

[pedrox]

I've added the Danasah 3.97b to the same GBT 1 tournament gauntlet against the 100 distinct opponents to see the progress, here are the results:

Danasah 3.97b - Score: 54.5/100

Rank: 42

Previous Version Stats:

Danasah 3.70- Score: 42.5/100

Rank: 62

Notes: This is also a good improvement, I'd guess that the latest version main focus was on improvement in blitz. And It succeeded convincingly!

Thanks for the test Swami, with the results I could think of an increase of approximately 100 points of ELO in 1+1, which would be according to my test. In times slower perhash the half, the CCRL has a gain of 55 points in 40/40 but still not enough games.

The gain, perhash by 30% faster of the program and better transition between middlegame and endgame.

Pedro

[swami]

I've added the Danasah 3.97b to the same GBT 1 tournament gauntlet against the 100 distinct opponents to see the progress, here are the results:

Danasah 3.97b - Score: 54.5/100

Rank: 42

Previous Version Stats:

Danasah 3.70- Score: 42.5/100

Rank: 62

Notes: This is also a good improvement, I'd guess that the latest version main focus was on improvement in blitz. And It succeeded convincingly!

Thanks for the test Swami, with the results I could think of an increase of approximately 100 points of ELO in 1+1, which would be according to my test. In times slower perhash the half, the CCRL has a gain of 55 points in 40/40 but still not enough games.

The gain, perhash by 30% faster of the program and better transition between middlegame and endgame.

Pedro

Improvement of anywhere between 50-80 is ofcourse expected, Pedro. Well done! Even fewer implementation do make big difference sometimes.