2009 ACCA World Computer Rapid Chess Championships

[Zach Wegner]

But surely if Rybka wins on similar hardware to the rest there is no argument - if it wins on 52 cores who cares?! if the engine that came 2nd had <=8?

http://www.csvn.nl/index.php?option=com ... &Itemid=28

[Harvey Williamson]

But surely if Rybka wins on similar hardware to the rest there is no argument - if it wins on 52 cores who cares?! if the engine that came 2nd had <=8?

http://www.csvn.nl/index.php?option=com ... &Itemid=28

Not sure what your point is? yes we played on hardware you can walk into any reputable dealer and order/buy, hardware that any top engine will run on - we beat the opposition that was there if some stayed at home that is their choice - Hans was kept busy helping me and operating Joker. If uniform hardware is desired then that is fine with me.

I saw a member of the Sjeng team post yesterday that he was not impressed with our win over Ktulu - I am sure he would be very impressed if 50 core Sjeng had scored a similar victory.

[bob]

http://www.taccl.org/ACCAWCRCC/2009ACCA ... CPart.html

I see the list pretty good so far. Rybka on its cluster and the rest on pretty average hardware. If i was Vas i would enter on bog standard hardware and probably win what is the point of the world champion winning on 52 cores?

Funny! Personally I would do everything I could to guarantee victory. It's kind of chumpy to play in a championship and and not try to win. Of course if daring your opponents to play at disadvantage is what you do to try to win, so be it.

But surely if Rybka wins on similar hardware to the rest there is no argument - if it wins on 52 cores who cares?! if the engine that came 2nd had <=8?

There will probably be other competitiors on Clusters, like if Sjeng enters. Championship chess programs can not continue to try to live in the past on hardware that I bought years ago. They need to be ready to take advantage and be a good investment for the next computer people will buy. People are buying computers with more than 8 cores this year.

Who is buying a Cluster and who has a program that can run on 1? When they are common, and simple to install and run by anyone who has more than 1 machine, of course they should be allowed. I would love to see the support emails when someone 1st releases a Cluster program.

That is completely illogical. Don't allow clusters until many programs can use them. But who is going to expend the effort to make their program run on a cluster until they can be used. This makes absolutely no sense.

When I did the parallel version of Crafty, everyone was on the _same_ bandwagon. Why do that when the average user will never have a multiple-cpu machine? The average pundits were wrong, and now _everyone_ has multiple CPUs. One can program for the past, or for the future. Clusters are the future. How many households today have more than one computer? Mine does. I'll bet the majority do. Which means the majority already _have_ clusters.

[Harvey Williamson]

http://www.taccl.org/ACCAWCRCC/2009ACCA ... CPart.html

I see the list pretty good so far. Rybka on its cluster and the rest on pretty average hardware. If i was Vas i would enter on bog standard hardware and probably win what is the point of the world champion winning on 52 cores?

Funny! Personally I would do everything I could to guarantee victory. It's kind of chumpy to play in a championship and and not try to win. Of course if daring your opponents to play at disadvantage is what you do to try to win, so be it.

But surely if Rybka wins on similar hardware to the rest there is no argument - if it wins on 52 cores who cares?! if the engine that came 2nd had <=8?

There will probably be other competitiors on Clusters, like if Sjeng enters. Championship chess programs can not continue to try to live in the past on hardware that I bought years ago. They need to be ready to take advantage and be a good investment for the next computer people will buy. People are buying computers with more than 8 cores this year.

Who is buying a Cluster and who has a program that can run on 1? When they are common, and simple to install and run by anyone who has more than 1 machine, of course they should be allowed. I would love to see the support emails when someone 1st releases a Cluster program.

That is completely illogical. Don't allow clusters until many programs can use them. But who is going to expend the effort to make their program run on a cluster until they can be used. This makes absolutely no sense.

When I did the parallel version of Crafty, everyone was on the _same_ bandwagon. Why do that when the average user will never have a multiple-cpu machine? The average pundits were wrong, and now _everyone_ has multiple CPUs. One can program for the past, or for the future. Clusters are the future. How many households today have more than one computer? Mine does. I'll bet the majority do. Which means the majority already _have_ clusters.

But buy a multi cpu machine - easy to run and install multi cpu program - always has been always will be. If a household has more than 1 computer and I agree many do. How many know anything about networking? How many would know where to start with installing and running a cluster program without serious help? Us geeks here would do ok but the average person with more than 1 computer?!

[bob]

http://www.taccl.org/ACCAWCRCC/2009ACCA ... CPart.html

I see the list pretty good so far. Rybka on its cluster and the rest on pretty average hardware. If i was Vas i would enter on bog standard hardware and probably win what is the point of the world champion winning on 52 cores?

Funny! Personally I would do everything I could to guarantee victory. It's kind of chumpy to play in a championship and and not try to win. Of course if daring your opponents to play at disadvantage is what you do to try to win, so be it.

But surely if Rybka wins on similar hardware to the rest there is no argument - if it wins on 52 cores who cares?! if the engine that came 2nd had <=8?

There will probably be other competitiors on Clusters, like if Sjeng enters. Championship chess programs can not continue to try to live in the past on hardware that I bought years ago. They need to be ready to take advantage and be a good investment for the next computer people will buy. People are buying computers with more than 8 cores this year.

Who is buying a Cluster and who has a program that can run on 1? When they are common, and simple to install and run by anyone who has more than 1 machine, of course they should be allowed. I would love to see the support emails when someone 1st releases a Cluster program.

That is completely illogical. Don't allow clusters until many programs can use them. But who is going to expend the effort to make their program run on a cluster until they can be used. This makes absolutely no sense.

When I did the parallel version of Crafty, everyone was on the _same_ bandwagon. Why do that when the average user will never have a multiple-cpu machine? The average pundits were wrong, and now _everyone_ has multiple CPUs. One can program for the past, or for the future. Clusters are the future. How many households today have more than one computer? Mine does. I'll bet the majority do. Which means the majority already _have_ clusters.

But buy a multi cpu machine - easy to run and install multi cpu program - always has been always will be. If a household has more than 1 computer and I agree many do. How many know anything about networking? How many would know where to start with installing and running a cluster program without serious help? Us geeks here would do ok but the average person with more than 1 computer?!

There's not anything "to do" for a cluster program. First, I assume all the machines are on a local network so that all can access the internet. Given that, the battle is won before it starts. One doesn't have to run around and twinkle around with each machine, it is all run from whatever you choose to call the main or head node.

When you bring up a windows box you are already using a cluster. DNS is cluster-based for example, and you don't have to do any configuration at all to make it work.

I hope to have one ready to release before the end of the year. We will see how hard it is. For unix machines, it will be invisible. WIndows is another question. But it always is.

[Harvey Williamson]

http://www.taccl.org/ACCAWCRCC/2009ACCA ... CPart.html

I see the list pretty good so far. Rybka on its cluster and the rest on pretty average hardware. If i was Vas i would enter on bog standard hardware and probably win what is the point of the world champion winning on 52 cores?

Funny! Personally I would do everything I could to guarantee victory. It's kind of chumpy to play in a championship and and not try to win. Of course if daring your opponents to play at disadvantage is what you do to try to win, so be it.

But surely if Rybka wins on similar hardware to the rest there is no argument - if it wins on 52 cores who cares?! if the engine that came 2nd had <=8?

There will probably be other competitiors on Clusters, like if Sjeng enters. Championship chess programs can not continue to try to live in the past on hardware that I bought years ago. They need to be ready to take advantage and be a good investment for the next computer people will buy. People are buying computers with more than 8 cores this year.

Who is buying a Cluster and who has a program that can run on 1? When they are common, and simple to install and run by anyone who has more than 1 machine, of course they should be allowed. I would love to see the support emails when someone 1st releases a Cluster program.

That is completely illogical. Don't allow clusters until many programs can use them. But who is going to expend the effort to make their program run on a cluster until they can be used. This makes absolutely no sense.

When I did the parallel version of Crafty, everyone was on the _same_ bandwagon. Why do that when the average user will never have a multiple-cpu machine? The average pundits were wrong, and now _everyone_ has multiple CPUs. One can program for the past, or for the future. Clusters are the future. How many households today have more than one computer? Mine does. I'll bet the majority do. Which means the majority already _have_ clusters.

But buy a multi cpu machine - easy to run and install multi cpu program - always has been always will be. If a household has more than 1 computer and I agree many do. How many know anything about networking? How many would know where to start with installing and running a cluster program without serious help? Us geeks here would do ok but the average person with more than 1 computer?!

There's not anything "to do" for a cluster program. First, I assume all the machines are on a local network so that all can access the internet. Given that, the battle is won before it starts. One doesn't have to run around and twinkle around with each machine, it is all run from whatever you choose to call the main or head node.

When you bring up a windows box you are already using a cluster. DNS is cluster-based for example, and you don't have to do any configuration at all to make it work.

I hope to have one ready to release before the end of the year. We will see how hard it is. For unix machines, it will be invisible. WIndows is another question. But it always is.

No point releasing something almost no one will buy - your presumption that most know how to network i think is wrong.

[Spock]

Well - for the kids coming out of school today it would probably be a breeze. Ten times the computer knowledge usually of their parents. So, I don't think a cluster will be an issue

[Harvey Williamson]

Well - for the kids coming out of school today it would probably be a breeze. Ten times the computer knowledge usually of their parents. So, I don't think a cluster will be an issue

How many of them are interested in buying a chess program?

[Gian-Carlo Pascutto]

No point releasing something almost no one will buy - your presumption that most know how to network i think is wrong.

They might not know how to network, as long as they can access the internet, they'll be fine.

[Harvey Williamson]

No point releasing something almost no one will buy - your presumption that most know how to network i think is wrong.

They might not know how to network, as long as they can access the internet, they'll be fine.

you go 1st - release a 'commercial' sjeng for cluster - good luck answering all the emails. If internet access is all thats needed - fine but you will still need to tell them what a router is.