And We Have the First Freeware Results!

[mloftus955]

Hi George,

Very interesting tournaments, I like to see how the lesser programs and programs from the past, do against each other. These play at levels I can better understand since I am a club player rated 1430.


Thanks,

Mark

[JBNielsen]

As for Dabbaba, I am mildly surprised with its success. Usually at 40/40 a max hash of 24 is a hindrance. But it hasn't bothered your engine. (I remember a particular Hiarcs version that maxed at 32MB hash at long controls). But I hope you will fix it so I can set it myself.

Best and keep up the play.

george


PS: In the tourn. you mentioned where MatMoi finished 4th, it did finish ahead of a few here. But exactly 50% of the engines in that tourn. are not in this one. One might argue some are stronger- but I am not quite so sure. And it would be 7 different out of 12 if I include your new Dabbaba version I am running that was not out then.

For some reason I cannot make hashtables bigger than 20MB in dabbaba. As far as I remember, only 12MB (512K elements) are used.

Ver 6.50 and ver 6.52 has the same playing strength - the difference is only a fix to make it run properly under Linux.

Dabbaba has just lost two games - now the score is more as expected...

[JBNielsen]

Hi George,

Very interesting tournaments, I like to see how the lesser programs and programs from the past, do against each other. These play at levels I can better understand since I am a club player rated 1430.


Thanks,

Mark

Where can we see the games?

[JVMerlino]

Strange things happen in these RRs all the time. Generally, what should happen likely won't. How many of you know going-in, which engine was the highest rated in Group 13 above? Lime 66, who is holding onto 14th place with a score of 2.5 out of 8.
george

Having played against many of these engines in my personal testing of Myrddin, I am definitely surprised that Lime is doing so poorly. But it is, of course, only 8 games. I would have picked Kurt to win Group 13.

As for Group 14, the only result that surprises me is MatMoi. Although I wasn't expecting it to win the group, I think the setup might be broken. It's definitely not that bad.

jm

[geots]

As for Dabbaba, I am mildly surprised with its success. Usually at 40/40 a max hash of 24 is a hindrance. But it hasn't bothered your engine. (I remember a particular Hiarcs version that maxed at 32MB hash at long controls). But I hope you will fix it so I can set it myself.

Best and keep up the play.

george


PS: In the tourn. you mentioned where MatMoi finished 4th, it did finish ahead of a few here. But exactly 50% of the engines in that tourn. are not in this one. One might argue some are stronger- but I am not quite so sure. And it would be 7 different out of 12 if I include your new Dabbaba version I am running that was not out then.

For some reason I cannot make hashtables bigger than 20MB in dabbaba. As far as I remember, only 12MB (512K elements) are used.

Ver 6.50 and ver 6.52 has the same playing strength - the difference is only a fix to make it run properly under Linux.

Dabbaba has just lost two games - now the score is more as expected...

Sorry I didn't get to you earlier. Had some power problems in our area and I missed a days/nights play. About the games, as soon as a RR is completed, a link will be available to you for the complete set of PGNs. Look, I know nothing about programming, but if I were you I would get Hyatt or someone to help me figure the hash problem out.


Best,

george

[geots]

Strange things happen in these RRs all the time. Generally, what should happen likely won't. How many of you know going-in, which engine was the highest rated in Group 13 above? Lime 66, who is holding onto 14th place with a score of 2.5 out of 8.
george

Having played against many of these engines in my personal testing of Myrddin, I am definitely surprised that Lime is doing so poorly. But it is, of course, only 8 games. I would have picked Kurt to win Group 13.

As for Group 14, the only result that surprises me is MatMoi. Although I wasn't expecting it to win the group, I think the setup might be broken. It's definitely not that bad.

jm

I agree with you- what surprises you is also surprising me. I suppose you could be right about something wrong with MatMoi. Problem is I have no idea what it could be. The set up is so simple. Just an exe and ini file is all he sent in the rar. All I had to do was set the hash on 128 in the ini file and that is it. There werent but one or 2 other commands in there. There was a "bin" file but of course I had to remove it. Maybe I should go back and make a copy of it but leave it empty and put it in the folder.

But I have something that will solve it- I believe. I have a "position-set-up" that is about as foolproof as you can get- just set the position up, and let MatMoi think and decide this move. If he shows +.20 or +.30 ther is a problem with the engine. But if he shows +1.00 or above- the engine is working correctly. I have tried in the past to trick it and make it show an error and it wont. It is about the best position test I have ever seen to know if an engine is working correctly. I will let you know the results. If MatMoi has been unfairly treated, it will be made up to him one way or the other. I am willing to put the folder in rar and send it to anyone who would be interested in checking my set-up.

But again, you are "dead-on" with all your observations. And thanks.


Best,

george

(If I do have something wrong with MatMoi I will tell you a secret- it would be far from the first time, and not nearly the last time. It just happens.)

[geots]

Hi George,

Very interesting tournaments, I like to see how the lesser programs and programs from the past, do against each other. These play at levels I can better understand since I am a club player rated 1430.


Thanks,

Mark

I try to answer everyone if at all possible. You don't know how good it makes me feel when people like you show an interest. I used to be a "fair" player, but at 63 with no games played in 4 or 5 years,I imagine your 1450 would beat me to death. So don't feel bad.

I have another quad that I am now going to set up to run 3 diff. RRs itself, and possibly add another to this quad. That would be 5, maybe 6 at once. So it won't be bad as time goes. But if anyone thinks getting together, arranging, and posting 6 RR updates at once is not a killer- discuss that aspect with Graham banks. Besides me, he is the only one I know who ever takes that much on at once- tho there may be a few more. Believeme- it grinds you down.

Best and appreciate your interest.

george

[geots]

Strange things happen in these RRs all the time. Generally, what should happen likely won't. How many of you know going-in, which engine was the highest rated in Group 13 above? Lime 66, who is holding onto 14th place with a score of 2.5 out of 8.
george

Having played against many of these engines in my personal testing of Myrddin, I am definitely surprised that Lime is doing so poorly. But it is, of course, only 8 games. I would have picked Kurt to win Group 13.

As for Group 14, the only result that surprises me is MatMoi. Although I wasn't expecting it to win the group, I think the setup might be broken. It's definitely not that bad.

jm

While I was loading some groups onto my other quad, I was running MatMoi ag. Warrior and Predateur in some 40/3 repeating blitz just for the hell of it- and tried to keep one eye on what was going on. Look, he was kicking the living shit out of both of them. It looked like men ag. boys. I keep trying to come up with a theory- but I just don't know. There is only 1 way to load an engine. It's over my head.

george

[JVMerlino]

Strange things happen in these RRs all the time. Generally, what should happen likely won't. How many of you know going-in, which engine was the highest rated in Group 13 above? Lime 66, who is holding onto 14th place with a score of 2.5 out of 8.
george

Having played against many of these engines in my personal testing of Myrddin, I am definitely surprised that Lime is doing so poorly. But it is, of course, only 8 games. I would have picked Kurt to win Group 13.

As for Group 14, the only result that surprises me is MatMoi. Although I wasn't expecting it to win the group, I think the setup might be broken. It's definitely not that bad.

jm

I agree with you- what surprises you is also surprising me. I suppose you could be right about something wrong with MatMoi. Problem is I have no idea what it could be. The set up is so simple. Just an exe and ini file is all he sent in the rar. All I had to do was set the hash on 128 in the ini file and that is it. There werent but one or 2 other commands in there. There was a "bin" file but of course I had to remove it. Maybe I should go back and make a copy of it but leave it empty and put it in the folder.

But I have something that will solve it- I believe. I have a "position-set-up" that is about as foolproof as you can get- just set the position up, and let MatMoi think and decide this move. If he shows +.20 or +.30 ther is a problem with the engine. But if he shows +1.00 or above- the engine is working correctly. I have tried in the past to trick it and make it show an error and it wont. It is about the best position test I have ever seen to know if an engine is working correctly. I will let you know the results. If MatMoi has been unfairly treated, it will be made up to him one way or the other. I am willing to put the folder in rar and send it to anyone who would be interested in checking my set-up.

But again, you are "dead-on" with all your observations. And thanks.

Best,

george

(If I do have something wrong with MatMoi I will tell you a secret- it would be far from the first time, and not nearly the last time. It just happens.)

Although I ever only used one version of MatMoi, I had heard that relative strength of released versions could sometimes go down quite noticeably from the previous version. Either way, it could just be bad luck.

Also, I just noticed that you mentioned how well Myrddin works in various GUIs. Thanks for the compliment, but I really didn't work on GUI support much at all.

I only use three GUIs for my testing, Winboard for FICS, Wildcat for gauntlets and Chessmaster (to be able to play against specific personalities I was curious about). So I did the minimum work necessary to make Myrddin play a game in those GUIs. But it doesn't support the "draw" command, so it still isn't very good for human play on FICS, for example. I just expected people to send me bugs if Myrddin didn't work in their favorite GUI, and I think I've only gotten two GUI bug reports in nearly four years.

Maybe less is more in this case?

jm

[geots]

Strange things happen in these RRs all the time. Generally, what should happen likely won't. How many of you know going-in, which engine was the highest rated in Group 13 above? Lime 66, who is holding onto 14th place with a score of 2.5 out of 8.
george

Having played against many of these engines in my personal testing of Myrddin, I am definitely surprised that Lime is doing so poorly. But it is, of course, only 8 games. I would have picked Kurt to win Group 13.

As for Group 14, the only result that surprises me is MatMoi. Although I wasn't expecting it to win the group, I think the setup might be broken. It's definitely not that bad.

jm

I agree with you- what surprises you is also surprising me. I suppose you could be right about something wrong with MatMoi. Problem is I have no idea what it could be. The set up is so simple. Just an exe and ini file is all he sent in the rar. All I had to do was set the hash on 128 in the ini file and that is it. There werent but one or 2 other commands in there. There was a "bin" file but of course I had to remove it. Maybe I should go back and make a copy of it but leave it empty and put it in the folder.

But I have something that will solve it- I believe. I have a "position-set-up" that is about as foolproof as you can get- just set the position up, and let MatMoi think and decide this move. If he shows +.20 or +.30 ther is a problem with the engine. But if he shows +1.00 or above- the engine is working correctly. I have tried in the past to trick it and make it show an error and it wont. It is about the best position test I have ever seen to know if an engine is working correctly. I will let you know the results. If MatMoi has been unfairly treated, it will be made up to him one way or the other. I am willing to put the folder in rar and send it to anyone who would be interested in checking my set-up.

But again, you are "dead-on" with all your observations. And thanks.

Best,

george

(If I do have something wrong with MatMoi I will tell you a secret- it would be far from the first time, and not nearly the last time. It just happens.)

Although I ever only used one version of MatMoi, I had heard that relative strength of released versions could sometimes go down quite noticeably from the previous version. Either way, it could just be bad luck.

Also, I just noticed that you mentioned how well Myrddin works in various GUIs. Thanks for the compliment, but I really didn't work on GUI support much at all.

I only use three GUIs for my testing, Winboard for FICS, Wildcat for gauntlets and Chessmaster (to be able to play against specific personalities I was curious about). So I did the minimum work necessary to make Myrddin play a game in those GUIs. But it doesn't support the "draw" command, so it still isn't very good for human play on FICS, for example. I just expected people to send me bugs if Myrddin didn't work in their favorite GUI, and I think I've only gotten two GUI bug reports in nearly four years.

Maybe less is more in this case?

jm

I read somewhere an article where this guy said luck is a big part of RRs. More so in a 2 cycle like this.


Best,

gts