Today we have the first running of an IvanHoe 9.47c compile. Have no idea exactly whose compile was used here, but it was "tinkered with" by one who Jim Ablett says is the person to look for if you want an increase in engine strength.
Tomorrow we will feature KLO's compile of this version. Then we can get some kind of idea how much tonight's results increased the strength of the compile.
Probably run in reverse order would have been better: the compile first- followed by the attempt to strengthen it. What can I say- life aint always fair.
As always- results based on 50 games- not half a million. So it is what it is:
1CPU/32bit
128MB hash
Bases=NONE (Rybka still clinging to the Nalimovs)
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games
IvanHoe 9.47cWHB.......+63-------+18/-9/=23------59.00%------29.5/50
Rybka 4.1 ......................-63-------+9/-18/=23------41.00%------20.5/50
A couple of things of importance. My new dual core had the problems in it fixed today- so I am anxious to see how the 64bit versions compare. It may be a bit of difference in 64bit vs 64bit, meaning more than I expect. Because in the list that follows, you will see that the IvanHoe version in 5th place in Frank's tournament here- if I am wrong correct me- but I believe it is KLO's 9.46f version he is running. The 64bit shows up as I said in 5th place for him, and is actually the worst of all the 32bit versions I have run ag. Rybka 4.1- as follows:
1. Strelka 5.1........................ +70
2. IvanHoe 9.47cWHB.............+63* (*doctored* compile featured above)
3/4. Ivanhoe 46f (PP).............+42
3/4. Ivanhoe 46h (PP)............+42
5. IvanHoe 9.46h (KLO)..........+28
6. Ivanhoe B50kBf (PP)..........+21
7. Ivanhoe B47e (N/A)...........+14
8. IvanHoe 9.46f (KLO)...........+0
I can promise you it would be no surprise to Jim Ablett who is leading the pack of IvanHoes here!
Lastly- if you had told me 2 years ago I would be using Rybka 4.1 for a punching bag/measuring stick- I would have called you nuts.
g
e
o
r
g
e
Away We Go- Special Edition IvanHoe
Moderator: Ras
-
geots
- Posts: 4790
- Joined: Sat Mar 11, 2006 12:42 am
Re: Away We Go- Special Edition IvanHoe
Today we have the first running of an IvanHoe 9.47c compile. Have no idea exactly whose compile was used here, but it was "tinkered with" by one who Jim Ablett says is the person to look for if you want an increase in engine strength.
Tomorrow we will feature KLO's compile of this version. Then we can get some kind of idea how much tonight's results increased the strength of the compile.
Probably run in reverse order would have been better: the compile first- followed by the attempt to strengthen it. What can I say- life aint always fair.
As always- results based on 50 games- not half a million. So it is what it is:
1CPU/32bit
128MB hash
Bases=NONE (Rybka still clinging to the Nalimovs)
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games
IvanHoe 9.47cWHB.......+63-------+18/-9/=23------59.00%------29.5/50
Rybka 4.1 ......................-63-------+9/-18/=23------41.00%------20.5/50
A couple of things of importance. My new dual core had the problems in it fixed today- so I am anxious to see how the 64bit versions compare. It may be a bit of difference in 64bit vs 64bit, meaning more than I expect. Because in the list that follows, you will see that the IvanHoe version in 5th place in Frank's tournament here- if I am wrong correct me- but I believe it is KLO's 9.46f version he is running. The 64bit shows up as I said in 5th place for him, and is actually the worst of all the 32bit versions I have run ag. Rybka 4.1- as follows:
1. Strelka 5.1........................ +70
2. IvanHoe 9.47cWHB.............+63* (*doctored* compile featured above)
3/4. Ivanhoe 46f (PP).............+42
3/4. Ivanhoe 46h (PP)............+42
5. IvanHoe 9.46h (KLO)..........+28
6. Ivanhoe B50kBf (PP)..........+21
7. Ivanhoe B47e (N/A)...........+14
8. IvanHoe 9.46f (KLO)...........+0
I can promise you it would be no surprise to Jim Ablett who is leading the pack of IvanHoes here!
Lastly- if you had told me 2 years ago I would be using Rybka 4.1 for a punching bag/measuring stick- I would have called you nuts.
g
e
o
r
g
e
PS: Any programmers who can toy with Rybka like this were not home sick the day God passed out the brains.
Tomorrow we will feature KLO's compile of this version. Then we can get some kind of idea how much tonight's results increased the strength of the compile.
Probably run in reverse order would have been better: the compile first- followed by the attempt to strengthen it. What can I say- life aint always fair.
As always- results based on 50 games- not half a million. So it is what it is:
1CPU/32bit
128MB hash
Bases=NONE (Rybka still clinging to the Nalimovs)
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games
IvanHoe 9.47cWHB.......+63-------+18/-9/=23------59.00%------29.5/50
Rybka 4.1 ......................-63-------+9/-18/=23------41.00%------20.5/50
A couple of things of importance. My new dual core had the problems in it fixed today- so I am anxious to see how the 64bit versions compare. It may be a bit of difference in 64bit vs 64bit, meaning more than I expect. Because in the list that follows, you will see that the IvanHoe version in 5th place in Frank's tournament here- if I am wrong correct me- but I believe it is KLO's 9.46f version he is running. The 64bit shows up as I said in 5th place for him, and is actually the worst of all the 32bit versions I have run ag. Rybka 4.1- as follows:
1. Strelka 5.1........................ +70
2. IvanHoe 9.47cWHB.............+63* (*doctored* compile featured above)
3/4. Ivanhoe 46f (PP).............+42
3/4. Ivanhoe 46h (PP)............+42
5. IvanHoe 9.46h (KLO)..........+28
6. Ivanhoe B50kBf (PP)..........+21
7. Ivanhoe B47e (N/A)...........+14
8. IvanHoe 9.46f (KLO)...........+0
I can promise you it would be no surprise to Jim Ablett who is leading the pack of IvanHoes here!
Lastly- if you had told me 2 years ago I would be using Rybka 4.1 for a punching bag/measuring stick- I would have called you nuts.
g
e
o
r
g
e
PS: Any programmers who can toy with Rybka like this were not home sick the day God passed out the brains.