ICT 9 - Leiden 2009 - Hiarcs Wins!

This forum is for general discussions and questions, including Collectors Corner and anything to do with Computer chess.

Moderators: Harvey Williamson, Steve B, Watchman

Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
gerold
Member
Posts: 55
Joined: Tue Jul 31, 2007 2:05 pm

Post by gerold »

Harvey Williamson wrote:So far only 7 entries in the main tournament Hiarcs Diep and Ktulu the top players. I hope the list will grow. It seems that although Rybka and Sjeng do not enter there is a proposed Cluster match Sjeng v Rybka - personally I do not see the point and see this as an attempt to undermine the main tournament and all the hard work of the volunteers who run the CSVN.

Sadly the small number of entries I think is largely due to the unavoidable clash with the dates of the WCCC. But if, a few, others do not enter because they fear a hardware limit and claim it handicaps them that is their loss.
Be looking forward to this match.
I don't know if they are trying to undermine it or not. I think it will
be interesting for both to be there. It will create more fans for the
events.

Good Luck,

Gerold.

P.S. I do like the idea of the handicaps.
To me it gives a good idea of who has
the strongest program.
User avatar
Watchman
Hiarcs Team Member
Posts: 874
Joined: Fri Aug 31, 2007 12:51 am
Location: Indianapolis

Post by Watchman »

Watchman wrote:Every time I hear this word "handicap" from someone...
gerold wrote:P.S. I do like the idea of the handicaps.
To me it gives a good idea of who has
the strongest program.
Ok let me re-phrase / clarify what I mean... not "every time" I hear it. I think the way Gerold says "handicap" he could also the word "restriction." Not trying to put words in your mouth Gerold... this is just how I perceive your meaning. And even if you do really mean handicap... k fine glad to see your opinion here.

My frustration with the use of this word... it has been used over and over by bellyachers in the Rybka Forum that somehow feel they are being targeted and cheated; it seems to always have a negative connotation associated with it.

If the ICGA ruled that they could use only an 8 core system max AND Rybka must set in the engine params "UCI_LimitStrength" to such and such elo... ok sure that's a handicap and I might even join in their chorus of crying. But that is not what has happened. Everyone had the same restriction placed on them. Heck... Rybka could have used a cluster (k an 8 core one but still...) however they chose not to keep abreast of ICGA rulings.
User avatar
turbojuice1122
Senior Member
Posts: 2315
Joined: Thu Aug 23, 2007 9:11 pm

Post by turbojuice1122 »

Watchman wrote:Everyone had the same restriction placed on them. Heck... Rybka could have used a cluster (k an 8 core one but still...) however they chose not to keep abreast of ICGA rulings.
What?! They knew about this as soon as it occurred, and it sparked outrage immediately. Also, you seem to have an extremely short memory: Rybka won the handicap event, too.
Watchman wrote:If there is any handicap it is self-imposed: Vas & Lukas (e.g.) pursued cluster development over MP development. They attempted a little innovation and were caught flat-footed by a tournament rule change. That is no one's fault but their own.
Sorry, but this is garbage. Rybka 3 can use all eight cores of an octacore just as well as Zappa (at least, any differences are negligible now, quite unlike versions before Rybka 3). There is no more MP improvement to attain that would be useful within the current rule limits anyway.

The use of the word "handicap" is meant as a joke to parody the fact that the ruling is stupid and takes away from the computer innovation and "levels the playing field" for those who do not have the necessary innovations that will be standard in a few years.
User avatar
Watchman
Hiarcs Team Member
Posts: 874
Joined: Fri Aug 31, 2007 12:51 am
Location: Indianapolis

Post by Watchman »

turbojuice1122 wrote:
Watchman wrote:Everyone had the same restriction placed on them. Heck... Rybka could have used a cluster (k an 8 core one but still...) however they chose not to keep abreast of ICGA rulings.
What?! They knew about this as soon as it occurred, and it sparked outrage immediately.
Aha... I see... that is why I see things like Lukas saying: So, why did nobody tell us?
We would have used a small cluster of 2 4.2 GHz Nehalem quads, and I could have saved €4000.

turbojuice1122 wrote:Also, you seem to have an extremely short memory: Rybka won the handicap event, too.
So what? What is your point? Well?
turbojuice1122 wrote:Rybka 3 can use all eight cores of an octacore just as well as Zappa (at least, any differences are negligible now, quite unlike versions before Rybka 3). There is no more MP improvement to attain that would be useful within the current rule limits anyway.
Wow Turbo I am impressed... you speak as if you are Vas himself. Tell me tho (since you are not Vas and there is a chance you are just regurgitating his words) you own an Octa? Please tell me you do. And tell me you have been using Zappa for the last year+ on this octa and R3 on it since its release. Sounds to me you are quite the programming expert (with first-hand knowledge of the code) and work as a beta tester.
turbojuice1122 wrote:The use of the word "handicap" is meant as a joke to parody the fact that the ruling is stupid
It's no joke to hear your incessant whining about it. Btw... who taught you to debate? To employ an impotent attack like "it's stupid" is unimaginative and demonstrates absolutely zilch in the critical thinking department.
turbojuice1122 wrote:and takes away from the computer innovation
Wow you went against the party line here. Vas certainly doesn't think so: "No tournament rule is going to stifle cluster development"

turbojuice1122 wrote: and "levels the playing field" for those who do not have the necessary innovations that will be standard in a few years.
I have already said "levels the playing field" is a good thing for these tourneys.

But your vision of this (cluster) innovation becoming a standard? First let me do an "I Dream of Jeanie Boing" so that all PCs have attained the highly prized (and long sought for) status of "appliance." Ok, What application(s) for the home user do you envision benefiting from this technology?
User avatar
turbojuice1122
Senior Member
Posts: 2315
Joined: Thu Aug 23, 2007 9:11 pm

Post by turbojuice1122 »

Watchman wrote:
turbojuice1122 wrote:
Watchman wrote:Everyone had the same restriction placed on them. Heck... Rybka could have used a cluster (k an 8 core one but still...) however they chose not to keep abreast of ICGA rulings.
What?! They knew about this as soon as it occurred, and it sparked outrage immediately.
Aha... I see... that is why I see things like Lukas saying: So, why did nobody tell us?
We would have used a small cluster of 2 4.2 GHz Nehalem quads, and I could have saved €4000.
The topic at hand was the knowledge about the use or non-use of large clusters like the current Rybka cluster. They knew immediately of the ruling against systems with more than 8 cores, and that was the point that sparked outrage. Whether a small 8-core cluster rather than an octacore machine was used is small peanuts in a handicap tournament. :-)
Watchman wrote:
turbojuice1122 wrote:Also, you seem to have an extremely short memory: Rybka won the handicap event, too.
So what? What is your point? Well?
My point was responding to this statement of yours in the previous post:
Watchman wrote:If there is any handicap it is self-imposed: Vas & Lukas (e.g.) pursued cluster development over MP development. They attempted a little innovation and were caught flat-footed by a tournament rule change. That is no one's fault but their own.
They weren't caught "flat-footed"--they had the superior product in any case--it's just that the rule change took away some of the opportunity to test it against the best computer opposition available.
Watchman wrote:
turbojuice1122 wrote:Rybka 3 can use all eight cores of an octacore just as well as Zappa (at least, any differences are negligible now, quite unlike versions before Rybka 3). There is no more MP improvement to attain that would be useful within the current rule limits anyway.
Wow Turbo I am impressed... you speak as if you are Vas himself. Tell me tho (since you are not Vas and there is a chance you are just regurgitating his words) you own an Octa? Please tell me you do. And tell me you have been using Zappa for the last year+ on this octa and R3 on it since its release. Sounds to me you are quite the programming expert (with first-hand knowledge of the code) and work as a beta tester.
I trust what Vas and Lukas say concerning testing, since this is what they do specifically, particularly Lukas. They were forthright about Rybka's disadvantage when Zappa had the lead in this department by a large margin, so they have no reason not to be here. No, I don't own an octacore--I defer to Vas and Lukas, i.e. those who have made MP implementation and testing a very high priority in the development of Rybka 3, as well as the results of many other people who have octacores. The kn/s numbers, which make sense more for Rybka than for any other engine in terms of gauging scalability, show that the difference in this department now between Rybka and Zappa is small. I seem to recall Lukas saying sometime ago that Rybka may even have a small advantage in this now; I'm sure that overall, the small differences depend on the system.
Watchman wrote:
turbojuice1122 wrote:The use of the word "handicap" is meant as a joke to parody the fact that the ruling is stupid
It's no joke to hear your incessant whining about it. Btw... who taught you to debate? To employ an impotent attack like "it's stupid" is unimaginative and demonstrates absolutely zilch in the critical thinking department.
I was simply explaining why the word "handicap" is used by those putting a humorous spin on a stupid rule change. Obviously if the debate is directly over the rule change itself, then I wouldn't lightly dismiss the argument if I'm taking so much time to argue about it and discuss points of view from both sides. However, I wasn't doing that here--that is in a different thread. Here, I was discussing the use of the word "handicap" and why those who use it use it like that.
Watchman wrote:
turbojuice1122 wrote:and takes away from the computer innovation
Wow you went against the party line here. Vas certainly doesn't think so: "No tournament rule is going to stifle cluster development"
He's talking about the development of the Rybka cluster, and I think that he is mistaken here. The rule change prevented Rybka from being able to play against as many strong opponents with good preparation by taking away the incentive for them to come with a cluster version (having a separate "Olympiad" tournament simply adds to complications if it's not considered part of the main tournament). It also takes away from the motivation of others developing clusters if there is no publicity in the games of these clusters in an official venue, such as there has been in the case of Hydra, GridChess, and Rybka in recent years.
Watchman wrote:
turbojuice1122 wrote: and "levels the playing field" for those who do not have the necessary innovations that will be standard in a few years.
I have already said "levels the playing field" is a good thing for these tourneys.
Then it's changing what much of the philosophy (and indeed the name itself, "World Computer Chess Championship) that has been in the tournament in previous years. The "separate tournament" should be the one that is hardware limited, and should have a name similar to something similar that has been used in the past: the World Microcomputer Chess Championship. There is no reason not to resurrect this and make this separate from what people consider the "main event" instead of creating this new event and calling it the "main event".
Watchman wrote:
But your vision of this (MP) innovation becoming a standard? First let me do an "I Dream of Jeanie Boing" so that all PCs have attained the highly prized (and long sought for) status of "appliance." Ok, What application(s) for the home user do you envision benefiting from this technology?
Notice the change in one word above compared with what you wrote. :-) With the same logic (which I believe is flawed), this would have been considered a legitimate question a few years ago.
User avatar
Dr.Wael Deeb
Member
Posts: 143
Joined: Fri Nov 16, 2007 10:33 pm
Location: Amman,Jordan

Post by Dr.Wael Deeb »

turbojuice1122 wrote:In many cases, isn't working on a version that works on a cluster simply a good long-term investment? Of course, no special work was needed for some of the programs out there, so if such work is needed, one could consider that to be something that handicaps the program, particularly since machines with very high numbers of cores will be on the market within a couple of years.
Creating a cluster version of a particular chess engine is not that trivial....it's a major restructure of the search algorithm plus special techniques to get the damn thing work....
I don't say that Mark can't do it,but why he should :!: :?:
Dr.D
User avatar
Ted Summers
Member
Posts: 269
Joined: Tue Jul 31, 2007 2:20 pm
Location: Marietta, GA
Contact:

Post by Ted Summers »

Dr.Wael Deeb wrote:
turbojuice1122 wrote:In many cases, isn't working on a version that works on a cluster simply a good long-term investment? Of course, no special work was needed for some of the programs out there, so if such work is needed, one could consider that to be something that handicaps the program, particularly since machines with very high numbers of cores will be on the market within a couple of years.
Creating a cluster version of a particular chess engine is not that trivial....it's a major restructure of the search algorithm plus special techniques to get the damn thing work....
I don't say that Mark can't do it,but why he should :!: :?:
Dr.D
Really, I agree with Wael, how many people would he sell it to? How many people have clusters of old PC's just hanging around, let alone have the knowledge of setting up clusters. Supporting Windows alone is enough, but to support someone's cluster is just asking to much for such a small market.
"Good decisions come from experience, and experience come from bad decisions."
User avatar
turbojuice1122
Senior Member
Posts: 2315
Joined: Thu Aug 23, 2007 9:11 pm

Post by turbojuice1122 »

As always with technology, it's not about today, but about tomorrow. The machines of a few years from now that will cost the same as quad-cores today will probably have in the range of 16-32 cores, and the programs that would work on these machines would perhaps run efficiently with the cluster technology of today. In either case, it seems like obviously a good medium-term to long-term idea to work on making a Hiarcs version that can use machines with more than 8 cores (and the same goes for other teams such as Junior). It will have the added benefit of allowing them to compete effectively in non-handicap tournaments. ;-) :lol:
User avatar
Harvey Williamson
Site Admin
Posts: 6079
Joined: Sun Jul 29, 2007 6:57 am
Location: Media City, UK
Contact:

Post by Harvey Williamson »

turbojuice1122 wrote:As always with technology, it's not about today, but about tomorrow. The machines of a few years from now that will cost the same as quad-cores today will probably have in the range of 16-32 cores, and the programs that would work on these machines would perhaps run efficiently with the cluster technology of today. In either case, it seems like obviously a good medium-term to long-term idea to work on making a Hiarcs version that can use machines with more than 8 cores (and the same goes for other teams such as Junior). It will have the added benefit of allowing them to compete effectively in non-handicap tournaments. ;-) :lol:
My Hiarcs runs very nicely already on 16 cores. So this will almost certainly be included in the next release. Supporting a Cluster version would be a nightmare. A customer reads somewhere that if he has more than 1 computer he can Cluster them together. He buys the product. He has no knowledge of networking etc... You should see some of the emails we get now from in-experienced users....
User avatar
Ted Summers
Member
Posts: 269
Joined: Tue Jul 31, 2007 2:20 pm
Location: Marietta, GA
Contact:

Post by Ted Summers »

turbojuice1122 wrote: It will have the added benefit of allowing them to compete effectively in non-handicap tournaments. ;-) :lol:
To "Handicap" means to place at a disadvantage. So please explain to us how any advantage or disadvantage is being imposed when everyone has to play by the same rules. I don't see any one player being singled out here, do you? I really like definition 2a. Does playing by the rules makes achievement unusually difficult for Rybka? I think we all know the the answer to this question is no.

Now just in case you need or want the Merriam-Webster definition.

http://www.merriam-webster.com/dictionary/handicap

1 a: a race or contest in which an artificial advantage is given or disadvantage imposed on a contestant to equalize chances of winning b: an advantage given or disadvantage imposed usually in the form of points, strokes, weight to be carried, or distance from the target or goal

2 a: a disadvantage that makes achievement unusually difficult b sometimes offensive : a physical disability
"Good decisions come from experience, and experience come from bad decisions."
Soren Riis
Member
Posts: 110
Joined: Wed Jun 11, 2008 8:58 pm

Post by Soren Riis »

To "Handicap" means to place at a disadvantage. So please explain to us how any advantage or disadvantage is being imposed when everyone has to play by the same rules. I don't see any one player being singled out here, do you? I really like definition 2a. Does playing by the rules makes achievement unusually difficult for Rybka? I think we all know the the answer to this question is no.
Rybka Cluster has a quite impressive score against Rybka 8-core. So according to your definition "Rybka" (="Rybka Cluster") was definitely handicapped and there is no doubt that at the present the odds that Rybka 8-core will win a tournament is smaller than Rybka Cluster will win the a corresponding "open" tournament.

I see nothing controversial about this. Personally I would have loved to see Hiarcs 16 core against the cluster. This would have been a tough fight that might have ended in a draw, the same result as the Hiarcs 8core vs Rybka 8core game ended in the world championship 2009.

Also clusters are not infallible. In Pamplona, Deep Sjeng 52 core lost to Shredder running on 8 cores, while Deep Sjeng on only 4 cores got a draw against Shredder 8core in the world champion.
Last edited by Soren Riis on Sat May 30, 2009 9:39 am, edited 1 time in total.
User avatar
Ted Summers
Member
Posts: 269
Joined: Tue Jul 31, 2007 2:20 pm
Location: Marietta, GA
Contact:

Post by Ted Summers »

Soren Riis wrote:
To "Handicap" means to place at a disadvantage. So please explain to us how any advantage or disadvantage is being imposed when everyone has to play by the same rules. I don't see any one player being singled out here, do you? I really like definition 2a. Does playing by the rules makes achievement unusually difficult for Rybka? I think we all know the the answer to this question is no.
Rybka Cluster has a quite impressive score against Rybka 8-core. So according to your definition Rybka Cluster is definitely handicapped and there is no doubt the the odd Rybka 8-core winning a tournament is smaller than Rybka Cluster winning the tournament.

I see nothing controversial about this. Personally I would have loved to see Hiarcs 16 core against the cluster. This would have been a tough fight that might have ended in a draw as the Hiarcs 8core vs Rybka 8core game ended in the world championship 2009.
You seem to forget or ignore the fact that everyone has to play by the same rules. Therefore it is a fair match, no handicaps to either side. Rybka Cluster is not being Handicaped, it is not being allowed to play. There is a difference ...
"Good decisions come from experience, and experience come from bad decisions."
Soren Riis
Member
Posts: 110
Joined: Wed Jun 11, 2008 8:58 pm

Post by Soren Riis »

I was just applying your definition to the situation. According to each of the definitions you gave any engine that have worse odds after the rule change, has been handicapped by the new rules.

Personally I actually think the 8core rule might have handicapped Hiarcs to!
User avatar
Ted Summers
Member
Posts: 269
Joined: Tue Jul 31, 2007 2:20 pm
Location: Marietta, GA
Contact:

Post by Ted Summers »

Soren Riis wrote:I was just applying your definition to the situation. According to each of the definitions you gave any engine that have worse odds after the rule change, has been handicapped by the new rules.

Personally I actually think Hiarcs might have been handicapped to!
This is not about odds, it's about the same rules being applied to everyone. By your definition then there is already a handicap since all engines do not use the same algorithms. As I said before, Rybka Cluster is not being Handicaped, it is not being allowed to play. I think Uri Blass said it best.
Uri Blass wrote:The word handicap is not about handicapping rybka but about reducing the level of the tournament.
"Good decisions come from experience, and experience come from bad decisions."
User avatar
turbojuice1122
Senior Member
Posts: 2315
Joined: Thu Aug 23, 2007 9:11 pm

Post by turbojuice1122 »

I don't use the word "handicap" seriously in these discussions, and I think that most other people on the side of the Rybka and Deep Sjeng people don't, either. It's simply a cute saying to get across frustration with these rules.

On the other hand, one can say that it handicaps the entire tournament and handicaps many entire teams. No GridChess. Nothing like Hydra. No Crafty cluster. No Achilles (though they've never competed in these things, anyway, which is unfortunate, since that piece of software, per core, is nearly as strong as Rybka 2.3.2a). The result is that it makes things a lot easier on those teams that have not been able to field software that is able to take advantage of the available technology.

One could actually say that blocking teams using more than 8 cores, but allowing those using 16 threads, is a handicap in the true sense of the word: it favors teams that are able to put together systems that take advantage of hyperthreading technology to "bend the rules". On the surface, I think that this is fine, but if they're going to do that, then to be fair, they should allow teams that can readily take advantage of other technology, too, such as those who can build a cluster that effectively runs 16 threads (whether true cores or not). The rule thus arbitrarily picks a particular technology of which teams are allowed to take advantage, and this does truly handicap teams that chose to go other directions in the past.
Post Reply