bob wrote:
So a few weak-kneed developers who want to restrict hardware because they either don't have the necessary skill, or else don't want to expend the effort, should have the say-so for an event that has 30+ years of history? It is not a "world championship" if the best players in the world don't compete. It is something less. Change the name to "uniform platform world computer chess championship" and be done with it. At least most will know that there are _strong_ players around that just were not allowed to compete. Or else get off the stick and do the development work rather than trying to propose rules that penalize those that do, to gain an advantage for those that do not.
It really is that simple...
This event is a _lot_ bigger than just a couple of lazy programming teams. It has 35 years of history behind it that should _not_ be undone. The old WMCCC events were limited hardware. The WCCCs never were. Go back to the WMCCC event if you insist on uniform hardware and let the WCCC mean that the strongest computer chess players available are allowed to play.
Unfortunately, commercial software authors whose engines are weaker than Rybka (and that is all of them actually) may be tempted to try to reduce Rybka's advantage as much as they can, for obvious monetary reasons. Most do not. With the WCCC being played with so few games, upsets can happen, and the temptation is definitely there to handicap Rybka as much as possible so as to be ready to take an advantage of any surprises. Most do not, most are sporting.
It is clear we need two events to try and keep most people happy. A WMCCC event for uniform hardware, and WCCC being unlimited, completely open. And it is also my view that the "World Champion" title should apply to the open, unlimited tournament and not the restricted one. The audience want to see the highest quality chess possible, and that is what something like Cluster Rybka gives.
bob wrote:
So a few weak-kneed developers who want to restrict hardware because they either don't have the necessary skill, or else don't want to expend the effort, should have the say-so for an event that has 30+ years of history? It is not a "world championship" if the best players in the world don't compete. It is something less. Change the name to "uniform platform world computer chess championship" and be done with it. At least most will know that there are _strong_ players around that just were not allowed to compete. Or else get off the stick and do the development work rather than trying to propose rules that penalize those that do, to gain an advantage for those that do not.
It really is that simple...
This event is a _lot_ bigger than just a couple of lazy programming teams. It has 35 years of history behind it that should _not_ be undone. The old WMCCC events were limited hardware. The WCCCs never were. Go back to the WMCCC event if you insist on uniform hardware and let the WCCC mean that the strongest computer chess players available are allowed to play.
Unfortunately, commercial software authors whose engines are weaker than Rybka (and that is all of them actually) may be tempted to try to reduce Rybka's advantage as much as they can, for obvious monetary reasons. Most do not. With the WCCC being played with so few games, upsets can happen, and the temptation is definitely there to handicap Rybka as much as possible so as to be ready to take an advantage of any surprises. Most do not, most are sporting.
It is clear we need two events to try and keep most people happy. A WMCCC event for uniform hardware, and WCCC being unlimited, completely open. And it is also my view that the "World Champion" title should apply to the open, unlimited tournament and not the restricted one. The audience want to see the highest quality chess possible, and that is what something like Cluster Rybka gives.
You are right about the monetary reasons I doubt any can afford Rybkas Cluster unless they have access to University hardware and have an academic project. Whatever the World Championship is about it is certainly not all about Rybka. I found on the Rybka forum by looking back in the archives to the Amsterdam tournament when others had Clusters and Rybka did not that the fanboys said Clusters = bad/unfair on Rybka.
Say a commercial program lets use Zappa as an eg. - it wins WCCC on fast 64bit hardware. When it won 3 or 4 years ago people said wow I must have that engine - they bought it and found on their single 32bit machines it was only mediocre - weaker than Junior and Shredder who also competed that year. Is this fair to customers/spectators & sponsors?
Harvey Williamson wrote:
Say a commercial program lets use Zappa as an eg. - it wins WCCC on fast 64bit hardware. When it won 3 or 4 years ago people said wow I must have that engine - they bought it and found on their single 32bit machines it was only mediocre - weaker than Junior and Shredder who also competed that year. Is this fair to customers/spectators & sponsors?
Can you tell me exactly the names of the people who bought Zappa directly as a result of that tournament, and ran it on their single CPU 32-bit machine and then complained ?
And anyway, an intelligent sensible person is unlikely in my view to use the result of one tournament as their sole point of reference for their purchasing decision. If they do, then they have only themselves to blame.
That is quite funny - you could say the exact same thing about the unaffordable for most W5580 machines that played in WCCC Forget clusters, 99% people could only dream about a W5580. A cluster of multiple Quad Core Intel boxes for example would be cheaper than the W5580, as pointed out many times by other people here
Harvey Williamson wrote:
Say a commercial program lets use Zappa as an eg. - it wins WCCC on fast 64bit hardware. When it won 3 or 4 years ago people said wow I must have that engine - they bought it and found on their single 32bit machines it was only mediocre - weaker than Junior and Shredder who also competed that year. Is this fair to customers/spectators & sponsors?
Can you tell me exactly the names of the people who bought Zappa directly as a result of that tournament, and ran it on their single CPU 32-bit machine and then complained ?
And anyway, an intelligent sensible person is unlikely in my view to use the result of one tournament as their sole point of reference for their purchasing decision. If they do, then they have only themselves to blame.
That is quite funny - you could say the exact same thing about the unaffordable for most W5580 machines that played in WCCC Forget clusters, 99% people could only dream about a W5580. A cluster of multiple Quad Core Intel boxes for example would be cheaper than the W5580, as pointed out many times by other people here
I know that as Zappa was an American program sales were huge, in the USA, after it won and the complaints followed soon after.
People keep using this Cluster is cheaper argument. But people do not turn up to tournaments with cheap Clusters - Sjengs was actually, allegedly, more expensive than Rybkas. I say allegedly because GCP will not confirm what was in his borrowed cluster. Neither was anyone allowed to benchmark the machine he played the main tournament with.
People may only be able to dream of owning the machines that Shredder, Hiarcs, Sjeng, Junior and Rybka used this year - but anyone can walk into a shop and order one knowing that any of the above programs will run on them out of the box.
Last edited by Harvey Williamson on Sun Jun 14, 2009 7:48 pm, edited 2 times in total.
Harvey Williamson wrote:
I know that as Zappa was an American program sales were huge, in the USA, after it won and the complaints followed soon after.
Well they have only themselves to blame as I said. They should have researched their purchase better. That isn't a reason for turning WCC on it's head.
Harvey Williamson wrote:
People keep using this Cluster is cheaper argument. But people do not turn up to tournaments with cheap Clusters - Sjengs was actually, allegedly, more expensive than Rybkas. I say allegedly because GCP will not confirm what was in his borrowed cluster.
The point, is the 8 core limit would *prevent* a cluster that is actually cheaper than a W5580, being used. That is why it was a nonsense in my view.
If something ain't broke, then don't fix it. WCC was not broken, now as a result of this foolishness, it is.
Harvey Williamson wrote:
I know that as Zappa was an American program sales were huge, in the USA, after it won and the complaints followed soon after.
Well they have only themselves to blame as I said. They should have researched their purchase better. That isn't a reason for turning WCC on it's head.
Harvey Williamson wrote:
People keep using this Cluster is cheaper argument. But people do not turn up to tournaments with cheap Clusters - Sjengs was actually, allegedly, more expensive than Rybkas. I say allegedly because GCP will not confirm what was in his borrowed cluster.
The point, is the 8 core limit would *prevent* a cluster that is actually cheaper than a W5580, being used. That is why it was a nonsense in my view.
If something ain't broke, then don't fix it. WCC was not broken, now as a result of this foolishness, it is.
Actually the rule does not prevent clusters as long as the total cores <= 8 or 16 with HT so i guess a 2x i7 cluster would be fine with HT on.
bob wrote:
So a few weak-kneed developers who want to restrict hardware because they either don't have the necessary skill, or else don't want to expend the effort, should have the say-so for an event that has 30+ years of history? It is not a "world championship" if the best players in the world don't compete. It is something less. Change the name to "uniform platform world computer chess championship" and be done with it. At least most will know that there are _strong_ players around that just were not allowed to compete. Or else get off the stick and do the development work rather than trying to propose rules that penalize those that do, to gain an advantage for those that do not.
It really is that simple...
This event is a _lot_ bigger than just a couple of lazy programming teams. It has 35 years of history behind it that should _not_ be undone. The old WMCCC events were limited hardware. The WCCCs never were. Go back to the WMCCC event if you insist on uniform hardware and let the WCCC mean that the strongest computer chess players available are allowed to play.
Unfortunately, commercial software authors whose engines are weaker than Rybka (and that is all of them actually) may be tempted to try to reduce Rybka's advantage as much as they can, for obvious monetary reasons. Most do not. With the WCCC being played with so few games, upsets can happen, and the temptation is definitely there to handicap Rybka as much as possible so as to be ready to take an advantage of any surprises. Most do not, most are sporting.
It is clear we need two events to try and keep most people happy. A WMCCC event for uniform hardware, and WCCC being unlimited, completely open. And it is also my view that the "World Champion" title should apply to the open, unlimited tournament and not the restricted one. The audience want to see the highest quality chess possible, and that is what something like Cluster Rybka gives.
Again, it is about development. Set your sights on improving your own program, rather than trying to cripple the opponents. This is getting overlooked. If someone spends time developing a quality parallel search, and can't use parallel hardware, then that effort is wasted. While those not considering parallel programming spend their time on something that will matter in a restricted hardware event. Ditto for cluster development. It is all about unevening the playing field, which is the opposite of what the WCCC should be about, which is to have the best play the best, period.
bob wrote:
So a few weak-kneed developers who want to restrict hardware because they either don't have the necessary skill, or else don't want to expend the effort, should have the say-so for an event that has 30+ years of history? It is not a "world championship" if the best players in the world don't compete. It is something less. Change the name to "uniform platform world computer chess championship" and be done with it. At least most will know that there are _strong_ players around that just were not allowed to compete. Or else get off the stick and do the development work rather than trying to propose rules that penalize those that do, to gain an advantage for those that do not.
It really is that simple...
This event is a _lot_ bigger than just a couple of lazy programming teams. It has 35 years of history behind it that should _not_ be undone. The old WMCCC events were limited hardware. The WCCCs never were. Go back to the WMCCC event if you insist on uniform hardware and let the WCCC mean that the strongest computer chess players available are allowed to play.
Unfortunately, commercial software authors whose engines are weaker than Rybka (and that is all of them actually) may be tempted to try to reduce Rybka's advantage as much as they can, for obvious monetary reasons. Most do not. With the WCCC being played with so few games, upsets can happen, and the temptation is definitely there to handicap Rybka as much as possible so as to be ready to take an advantage of any surprises. Most do not, most are sporting.
It is clear we need two events to try and keep most people happy. A WMCCC event for uniform hardware, and WCCC being unlimited, completely open. And it is also my view that the "World Champion" title should apply to the open, unlimited tournament and not the restricted one. The audience want to see the highest quality chess possible, and that is what something like Cluster Rybka gives.
You are right about the monetary reasons I doubt any can afford Rybkas Cluster unless they have access to University hardware and have an academic project. Whatever the World Championship is about it is certainly not all about Rybka. I found on the Rybka forum by looking back in the archives to the Amsterdam tournament when others had Clusters and Rybka did not that the fanboys said Clusters = bad/unfair on Rybka.
Say a commercial program lets use Zappa as an eg. - it wins WCCC on fast 64bit hardware. When it won 3 or 4 years ago people said wow I must have that engine - they bought it and found on their single 32bit machines it was only mediocre - weaker than Junior and Shredder who also competed that year. Is this fair to customers/spectators & sponsors?
No, it shows stupidity. Do you believe that the Ford you see in a nascar race will provide the same performance as a ford you buy at a local Ford dealer? Hardly anybody is that stupid. If they are, that really is their problem...
Harvey Williamson wrote:
Say a commercial program lets use Zappa as an eg. - it wins WCCC on fast 64bit hardware. When it won 3 or 4 years ago people said wow I must have that engine - they bought it and found on their single 32bit machines it was only mediocre - weaker than Junior and Shredder who also competed that year. Is this fair to customers/spectators & sponsors?
Can you tell me exactly the names of the people who bought Zappa directly as a result of that tournament, and ran it on their single CPU 32-bit machine and then complained ?
And anyway, an intelligent sensible person is unlikely in my view to use the result of one tournament as their sole point of reference for their purchasing decision. If they do, then they have only themselves to blame.
That is quite funny - you could say the exact same thing about the unaffordable for most W5580 machines that played in WCCC Forget clusters, 99% people could only dream about a W5580. A cluster of multiple Quad Core Intel boxes for example would be cheaper than the W5580, as pointed out many times by other people here
I know that as Zappa was an American program sales were huge, in the USA, after it won and the complaints followed soon after.
Huge = maybe total sales of 500 copies at best???
People keep using this Cluster is cheaper argument. But people do not turn up to tournaments with cheap Clusters - Sjengs was actually, allegedly, more expensive than Rybkas. I say allegedly because GCP will not confirm what was in his borrowed cluster. Neither was anyone allowed to benchmark the machine he played the main tournament with.
People also show up with ND hardware that is not yet commercially available, or hardware with wild overclocking and cooling... This doesn't stop the "arms race" it just pushes it into a different direction.
People may only be able to dream of owning the machines that Shredder, Hiarcs, Sjeng, Junior and Rybka used this year - but anyone can walk into a shop and order one knowing that any of the above programs will run on them out of the box.
Harvey Williamson wrote:
Actually the rule does not prevent clusters as long as the total cores <= 8 or 16 with HT so i guess a 2x i7 cluster would be fine with HT on.
Yes - but it would prevent say 5 overclocked Quad Q6600 boxes (20 cores) from competing. That is a cheap set-up. It is crazy to prevent that, but allow a W5580 which is perhaps twice the cost. What performance would be like, well who knows
Forget clusters, 99% people could only dream about a W5580. A cluster of multiple Quad Core Intel boxes for example would be cheaper than the W5580, as pointed out many times by other people here