IvanHoe T52E- Back Up and Back Off!

[geots]

After testing yesterday the IvanHoe 9.47c compile that was taken and "tinkered with" by one of the best, I had planned to run the original compile now to see if there was a modest or more strength increase. But I began to like the idea less and less- it seemed redundant and time wasting when I already knew it was stronger- just a matter of how much. Testing another version rather than a "I told you so" situation seemed to make more sense. When you are 1 person with 1CPU- you try to use your time wisely.

The IvanHoe version tonight was introduced earlier in a 100 game match with Fritz 13. And an extremely painful experience for Fritz 13. I was shocked NOT that this IvanHoe version beat Fritz 13, but rather the way he toyed with it like a cat would a mouse. But against Rybka 4.1.........??


1CPU/32bit
128MB hash
Bases=NONE (Rybka 4.1 clings to the Nalimovs)
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games


IvanHoe T52E [AHMED]......+42----+14/-8/=28-----56.00%-----28.0/50

Rybka 4.1...........................-42------+8/-14/=28----44.00%-----22.0/50



Ok, Rybka 4.1 was stiffer competition than Fritz 13, so no surprise there. Again, IvanHoe is what it is. That was evident after a match or 2. These matches are an effort to give one at least some kind of idea which compile he might want to search for- as some of the best have played here. There is certainly no solid guarantee that you would see the same results. But I don't think they would disappoint.

It is about time to consider switching horses. Tho there are more than enough IvanHoe compiles out there to test, I am beginning to come face to face with the law of diminishing returns. The benefit just gets smaller with each match, and I find myself thinking, "Ok, how many Elo points will Rybka lose this match by?"

The answer would be fresh competition facing fresh competition. There one must be careful. Nothing to gain by testing 2 engines already solidly listed in their places in our CCRL rating list.

We do have Fire and Firebird versions, along with the MishasMaulers and the D. Rybka Killers (they kill more than just Rybkas). Maybe just 1 more IvanHoe version while awaiting further releases............? Or maybe a 64bit match or 2- assuming I ever have that system up and running. Choices, choices----- so many engines, so little time.


And To All A Good Night,

g
e
o
r
g
e

[Terry McCracken]

After testing yesterday the IvanHoe 9.47c compile that was taken and "tinkered with" by one of the best, I had planned to run the original compile now to see if there was a modest or more strength increase. But I began to like the idea less and less- it seemed redundant and time wasting when I already knew it was stronger- just a matter of how much. Testing another version rather than a "I told you so" situation seemed to make more sense. When you are 1 person with 1CPU- you try to use your time wisely.

The IvanHoe version tonight was introduced earlier in a 100 game match with Fritz 13. And an extremely painful experience for Fritz 13. I was shocked NOT that this IvanHoe version beat Fritz 13, but rather the way he toyed with it like a cat would a mouse. But against Rybka 4.1.........??


1CPU/32bit
128MB hash
Bases=NONE (Rybka 4.1 clings to the Nalimovs)
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games


IvanHoe T52E [AHMED]......+42----+14/-8/=28-----56.00%-----28.0/50

Rybka 4.1...........................-42------+8/-14/=28----44.00%-----22.0/50



Ok, Rybka 4.1 was stiffer competition than Fritz 13, so no surprise there. Again, IvanHoe is what it is. That was evident after a match or 2. These matches are an effort to give one at least some kind of idea which compile he might want to search for- as some of the best have played here. There is certainly no solid guarantee that you would see the same results. But I don't think they would disappoint.

It is about time to consider switching horses. Tho there are more than enough IvanHoe compiles out there to test, I am beginning to come face to face with the law of diminishing returns. The benefit just gets smaller with each match, and I find myself thinking, "Ok, how many Elo points will Rybka lose this match by?"

The answer would be fresh competition facing fresh competition. There one must be careful. Nothing to gain by testing 2 engines already solidly listed in their places in our CCRL rating list.

We do have Fire and Firebird versions, along with the MishasMaulers and the D. Rybka Killers (they kill more than just Rybkas). Maybe just 1 more IvanHoe version while awaiting further releases............? Or maybe a 64bit match or 2- assuming I ever have that system up and running. Choices, choices----- so many engines, so little time.


And To All A Good Night,

g
e
o
r
g
e

Did someone take it down?

[geots]

After testing yesterday the IvanHoe 9.47c compile that was taken and "tinkered with" by one of the best, I had planned to run the original compile now to see if there was a modest or more strength increase. But I began to like the idea less and less- it seemed redundant and time wasting when I already knew it was stronger- just a matter of how much. Testing another version rather than a "I told you so" situation seemed to make more sense. When you are 1 person with 1CPU- you try to use your time wisely.

The IvanHoe version tonight was introduced earlier in a 100 game match with Fritz 13. And an extremely painful experience for Fritz 13. I was shocked NOT that this IvanHoe version beat Fritz 13, but rather the way he toyed with it like a cat would a mouse. But against Rybka 4.1.........??


1CPU/32bit
128MB hash
Bases=NONE (Rybka 4.1 clings to the Nalimovs)
Ponder_Learning=OFF
Perfect 12.32 book w/12-move limit
40/3 Repeating
Match=50 games


IvanHoe T52E [AHMED]......+42----+14/-8/=28-----56.00%-----28.0/50

Rybka 4.1...........................-42------+8/-14/=28----44.00%-----22.0/50



Ok, Rybka 4.1 was stiffer competition than Fritz 13, so no surprise there. Again, IvanHoe is what it is. That was evident after a match or 2. These matches are an effort to give one at least some kind of idea which compile he might want to search for- as some of the best have played here. There is certainly no solid guarantee that you would see the same results. But I don't think they would disappoint.

It is about time to consider switching horses. Tho there are more than enough IvanHoe compiles out there to test, I am beginning to come face to face with the law of diminishing returns. The benefit just gets smaller with each match, and I find myself thinking, "Ok, how many Elo points will Rybka lose this match by?"

The answer would be fresh competition facing fresh competition. There one must be careful. Nothing to gain by testing 2 engines already solidly listed in their places in our CCRL rating list.

We do have Fire and Firebird versions, along with the MishasMaulers and the D. Rybka Killers (they kill more than just Rybkas). Maybe just 1 more IvanHoe version while awaiting further releases............? Or maybe a 64bit match or 2- assuming I ever have that system up and running. Choices, choices----- so many engines, so little time.


And To All A Good Night,

g
e
o
r
g
e

Did someone take it down?

Terry< in this tournament and matches section I try to read and respond to all questions concerning the match in question. So I don't mind reading your worthy replies. I did not want you to think I was refusing to reply to you. But the problem is- the question you asked, I am not sure what it is in relation to. I don't understand "what might have been taken down".


Best regards

george