Ok, I know from the inner circle that Vas owns China, Japan and Spain. Not so much Mexico. But nobody ever got the impression that the tournaments he won there, had been rigged by the fabulous Vas & Fish.Tom Barrister wrote: Odd how he isn't winning the tournaments that don't belong to him.
A Very Novel Idea Concerning Vas- BE FAIR
Moderator: Ras
-
- Posts: 6081
- Joined: Fri Mar 10, 2006 11:14 pm
- Location: Munster, Nuremberg, Princeton
Re: A Very Novel Idea Concerning Vas- BE FAIR
-Popper and Lakatos are good but I'm stuck on Leibowitz
-
- Posts: 6081
- Joined: Fri Mar 10, 2006 11:14 pm
- Location: Munster, Nuremberg, Princeton
Re: A Very Novel Idea Concerning Vas- BE FAIR
Ok, I know from the inner circle that Vas owns China, Japan and Spain. Not so much Mexico. But nobody ever got the impression that the tournaments he won there, had been rigged by the fabulous Vas & Fish.Tom Barrister wrote: Odd how he isn't winning the tournaments that don't belong to him.
-Popper and Lakatos are good but I'm stuck on Leibowitz
-
- Posts: 227
- Joined: Tue Oct 05, 2010 5:29 pm
Re: A Very Novel Idea Concerning Vas- BE FAIR
How could anybody not get that impression? The Ippolit series wasn't (and at some places still isn't) allowed in. That's "rigging".Rolf wrote:But nobody ever got the impression that the tournaments he won there, had been rigged by the fabulous Vas & Fish.Tom Barrister wrote: Odd how he isn't winning the tournaments that don't belong to him.
This production is being brought to you by Rybka: "The engine made from scratch.™"
-
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: A Very Novel Idea Concerning Vas- BE FAIR
I did not "chicken out". The discussion was not going to be open, with others allowed to participate. That made no sense. And in a forum with extremely biased moderation philosophy? Perfect place to settle everything, right? We offered to do this here. No response. Yet I am the one that "chickened out"???Rolf wrote:But why the talks now, Fabien goes to court and everything will be solved. As you know I wanted to mediate, Vas had agreed, you at first too, but then you chickened out because Rybkaforum is zoo dangerous for you as it seemed. However on Open forum you go on promenade, that is no problem for the propaganda. Again Vas was willing to talk to you.bob wrote:No, although I was specifically talking about clearing up this fruit/rybka mess, as opposed to anything else...Tom Barrister wrote:You made a good point, and I'm curious about it. Has Mr. Rajlich contributed anything in the forums (in any forums, not just this one) regarding chess programming, re: technical discussions, ideas, etc.?bob wrote:
BTW, what have we learned in the past 5 years? Certainly not one thing from Vas.
-
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: A Very Novel Idea Concerning Vas- BE FAIR
Yes, but there has to be some sort of financial justification, and someone with enough credibility to raise the issue. Comparing a binary to a Source program is a challenge, to say the least. If it were Microsoft doing this, I am sure FSF would be more than willing. But court costs would be substantial. Lawyers are not free, nor are the expert witnesses required. Money thrown into a black hole with little hope to get it back since computer chess doesn't exactly make millionaires out of authors.Tom Barrister wrote:So Mr. Rajlich hasn't contributed a thing to chess programming, other than Rybka itself? Of course, that would make sense, since he makes his living, or a substantial part of it, from the sale and license of Rybka. Or it might be that he has nothing original to contribute.
I don't understand why it's Mr. Letouzey's choice to go to court anyway. Wasn't it established that he gave the license (or whatever it's called) for Fruit to the Free Software Foundation? Wouldn't it be the prerogative of the FSF to pursue litigation?
-
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: A Very Novel Idea Concerning Vas- BE FAIR
Here's a deal for you. I will likely retire in 2 years. At that point, offer me $100,000 to disassemble any two commercial engines of your choice and compare them to an open source program of your choice. I'll take the job and do it. That is a year's effort, for a year's pay. That is a pretty good estimate of the time it would require.Rolf wrote:Bob, thanks for your never tiring eloquence, but I must correct a delusion in computerchesscode once and for all. Your comparison with literature is a lame duck. The animal wont fly anymore.bob wrote:There are two ways to do things. Nobody is not doubting the programming skills of Vas. But you can write your own program, or you can copy another and save 1-2-3 years of effort, and start your own effort on top of the program you copied. You might get to the same point, skill-wise. But not time-wise. And not effort-wise.Rolf wrote:It's not just a question of looking and seeing. IMO.bob wrote:Not is it is not so convenient and you now don't bother with his opinion at all...
OK, you watch someone kill somebody, with your own two eyes you see the entire act. Do you consider him innocent until he is found guilty in court? I do not. I _know_ what I saw. Same with the fruit/rybka issue. I _know_ what I saw.
Go for it if you are so sure about it. I am the last who could contradict you because I am a technical layman.
But I know what I saw, namely the titles Vas got after making his program always stronger and stronger. Surely not through always finding new tricks in the old Fruit on Chrismas Eve. That is the reason why all the other programmers dont support the group of 4 or 5. 295 are just watching what is going on and possibly are shocked like me because it doesnt make sense to insinuate that Vas simply copied. Because from where came the material for all the titles???
It doesnt make sense in my eyes. I still beleive in Vasik, that's for sure. It's just not fair how his character has been torn through the mud. NB not every wrong that is done on this planet was intention to betray others. The history of the past five years should have told you a better story than what you are now so much focussed on.
So just because Rybka became #1, one can't use that to justify copying the code of others, any more than an author could copy the plot from one book, the setting from another, the characters from a third, and form that into a best seller..
Say take the Matt Reilly book "Ice Station", change the location to South America, change the main character to willie makeit, and you would have a book that would sell. And be a knock-off at the same time.
BTW, what have we learned in the past 5 years? Certainly not one thing from Vas. He could have cleared this up and everyone could have moved on. But silence keeps it going. In this case, it was Fabien that started the discussion for the N+1th time.
1) you have no plot at all - chess is always the same outcome and since Vas is a good chessplayer he quickly understood the usual tone
2) everything in CC is technical setting. It's all thought and done. Also here no surprise. For all it cant be stolen or copied, because it's always the same. Proof: go and get a Shredder or Junior and then tell me about copying. But you dont want to do this because it would destroy your holy argument against Vas.
Who, in their right mind, would want do that for free? Invest a year of time for something that won't help the person at all? I work on Crafty because I enjoy it. I do other things because I am paid for doing them (teaching classes, research, committees, advising students, etc). I hunt, fish, fly model airplanes, play the guitar, build things, etc. all for free, because I enjoy doing them. I don't enjoy taking a microscope to a program to disassemble it and try to track asm back to C. I know how, certainly. But it is a lot of work. It isn't fun. It pays nothing. So what is the incentive? If there was a legitimate question raised about them someone would likely take the time, myself included. But with no red flags waving, one could spend the time and discover absolutely nothing.
I'd rather spend that year working on Crafty. Or doing things I get paid for, or things I enjoy doing.
It really is that simple, and there is no dark motive lurking in the background.
Programming is a _creative_ task. Not a mechanical one. Ditto for writing a novel, a screenplay, a poem, a song. Designing a building. Or a rocket to go to the moon. Things can be stolen or copied. In any of those. There is a great book "The eagle has landed" which is not about Apollo 11, but is instead about some Data General engineers that openly broke into the DEC plant in Massachusetts and openly went thru the design/hardware of the Vax 11/780 while they were designing the Data General MV series of machines. Copying happens. It saves a ton of time, which when selling things that evolve quickly is a major concern.
3) Yes, you might have characters too but they are also interchangable. Nothing new under the sun. Look, a language has emotions that it can express, but your code must play vchess and not cause you weep some tears.
You miss the point completely, whether that is intentional or not I can't tell. But the idea is about time. If you do things yourself, it takes a lot of time to catch the front-runners. I did this and it took me 11-12 years (I tied for first at the 1982 ACM event and won the 1983 ACM/WMCC event, where my program played its first move in 1968 and started competing in 1976). Time. If you can short-cut the development cycle, you deliver a product quicker. You reap the financial benefits quicker, and therefore over a longer period of time.
4) Now the most difficult part for me as a tech lay. Yes, I think I can imagine what code means. It's a permanent linkage of ordering stuff and here comes something of an ideal effectness into play because what could cause speed and depth that will end in chesswise more correct and deep moves and taken, the chess positions allow something and are not drawish then minimalism beats ranting.
5) Vas cannot have done just copying because that wouldnt have made his code win, he won because he's the better programmer in exploiting the little possibilities where programmers could differ. Everything else is the same and copied more or less.
So this is about time, and money. And saving some of the former to make more of the latter. Wouldn't surprise me if Vas were able to write a world-champion program completely from scratch. But it would have taken far longer. Or one can cut a few ethical corners and get there quicker.
But I dont want to keep you away from court trials. Let Fabien learn his lessons. And you coach him. I know that you cant win such a case.
What makes you so omniscient that you can say that with certainty? A trial by jury is anything but a certainty. Juries are human.
It is. You copy a few years worth of effort and expend a few minutes of effort. You just saved several years of writing, debugging, testing, tuning, and such.Still I would prefer to applaud you and your teamsters because I want to know who the real hypocrits are in the story. It's so funny. You are talking about copying as if that were easier than writing from scratch.
I think the assumption about hardware is way too speculative. There are good systems out there. Cray makes one. As does Sun. And others. The cluster Rybka is using is a toy compared to some real shared-memory hardware platforms that are around. But they are expensive. However, I proved, years ago, that one could gain access to a machine that sold for way over $30 million, and use it to play chess. So the cluster superiority is not guaranteed. It is not even convincing me it is that much stronger than a good 16 core box anyway... way more hype than substance, IMHO.But layman Rolf says, it must be much more difficult. Proof: why not many walked on that same highway and became Champions? Houdini wont make it because on a single machine for two programs we have no hardware problemsolving. But in all his tournaments Vas had the best combination of software and hardware. Therefore it's boring what Martin does. Look, if I would fill you up with tons of whisky, I would be the better teacher of us in computerchess! It's a bad example but you get the drift. gringrin
-
- Posts: 3726
- Joined: Thu Mar 16, 2006 7:10 pm
Re: A Very Novel Idea Concerning Vas- BE FAIR
bob wrote:Here's a deal for you. I will likely retire in 2 years. At that point, offer me $100,000 to disassemble any two commercial engines of your choice and compare them to an open source program of your choice. I'll take the job and do it. That is a year's effort, for a year's pay. That is a pretty good estimate of the time it would require.Rolf wrote:Bob, thanks for your never tiring eloquence, but I must correct a delusion in computerchesscode once and for all. Your comparison with literature is a lame duck. The animal wont fly anymore.bob wrote:There are two ways to do things. Nobody is not doubting the programming skills of Vas. But you can write your own program, or you can copy another and save 1-2-3 years of effort, and start your own effort on top of the program you copied. You might get to the same point, skill-wise. But not time-wise. And not effort-wise.Rolf wrote:It's not just a question of looking and seeing. IMO.bob wrote:Not is it is not so convenient and you now don't bother with his opinion at all...
OK, you watch someone kill somebody, with your own two eyes you see the entire act. Do you consider him innocent until he is found guilty in court? I do not. I _know_ what I saw. Same with the fruit/rybka issue. I _know_ what I saw.
Go for it if you are so sure about it. I am the last who could contradict you because I am a technical layman.
But I know what I saw, namely the titles Vas got after making his program always stronger and stronger. Surely not through always finding new tricks in the old Fruit on Chrismas Eve. That is the reason why all the other programmers dont support the group of 4 or 5. 295 are just watching what is going on and possibly are shocked like me because it doesnt make sense to insinuate that Vas simply copied. Because from where came the material for all the titles???
It doesnt make sense in my eyes. I still beleive in Vasik, that's for sure. It's just not fair how his character has been torn through the mud. NB not every wrong that is done on this planet was intention to betray others. The history of the past five years should have told you a better story than what you are now so much focussed on.
So just because Rybka became #1, one can't use that to justify copying the code of others, any more than an author could copy the plot from one book, the setting from another, the characters from a third, and form that into a best seller..
Say take the Matt Reilly book "Ice Station", change the location to South America, change the main character to willie makeit, and you would have a book that would sell. And be a knock-off at the same time.
BTW, what have we learned in the past 5 years? Certainly not one thing from Vas. He could have cleared this up and everyone could have moved on. But silence keeps it going. In this case, it was Fabien that started the discussion for the N+1th time.
1) you have no plot at all - chess is always the same outcome and since Vas is a good chessplayer he quickly understood the usual tone
2) everything in CC is technical setting. It's all thought and done. Also here no surprise. For all it cant be stolen or copied, because it's always the same. Proof: go and get a Shredder or Junior and then tell me about copying. But you dont want to do this because it would destroy your holy argument against Vas.
Who, in their right mind, would want do that for free? Invest a year of time for something that won't help the person at all? I work on Crafty because I enjoy it. I do other things because I am paid for doing them (teaching classes, research, committees, advising students, etc). I hunt, fish, fly model airplanes, play the guitar, build things, etc. all for free, because I enjoy doing them. I don't enjoy taking a microscope to a program to disassemble it and try to track asm back to C. I know how, certainly. But it is a lot of work. It isn't fun. It pays nothing. So what is the incentive? If there was a legitimate question raised about them someone would likely take the time, myself included. But with no red flags waving, one could spend the time and discover absolutely nothing.
I'd rather spend that year working on Crafty. Or doing things I get paid for, or things I enjoy doing.
It really is that simple, and there is no dark motive lurking in the background.
Programming is a _creative_ task. Not a mechanical one. Ditto for writing a novel, a screenplay, a poem, a song. Designing a building. Or a rocket to go to the moon. Things can be stolen or copied. In any of those. There is a great book "The eagle has landed" which is not about Apollo 11, but is instead about some Data General engineers that openly broke into the DEC plant in Massachusetts and openly went thru the design/hardware of the Vax 11/780 while they were designing the Data General MV series of machines. Copying happens. It saves a ton of time, which when selling things that evolve quickly is a major concern.
3) Yes, you might have characters too but they are also interchangable. Nothing new under the sun. Look, a language has emotions that it can express, but your code must play vchess and not cause you weep some tears.
You miss the point completely, whether that is intentional or not I can't tell. But the idea is about time. If you do things yourself, it takes a lot of time to catch the front-runners. I did this and it took me 11-12 years (I tied for first at the 1982 ACM event and won the 1983 ACM/WMCC event, where my program played its first move in 1968 and started competing in 1976). Time. If you can short-cut the development cycle, you deliver a product quicker. You reap the financial benefits quicker, and therefore over a longer period of time.
4) Now the most difficult part for me as a tech lay. Yes, I think I can imagine what code means. It's a permanent linkage of ordering stuff and here comes something of an ideal effectness into play because what could cause speed and depth that will end in chesswise more correct and deep moves and taken, the chess positions allow something and are not drawish then minimalism beats ranting.
5) Vas cannot have done just copying because that wouldnt have made his code win, he won because he's the better programmer in exploiting the little possibilities where programmers could differ. Everything else is the same and copied more or less.
So this is about time, and money. And saving some of the former to make more of the latter. Wouldn't surprise me if Vas were able to write a world-champion program completely from scratch. But it would have taken far longer. Or one can cut a few ethical corners and get there quicker.
But I dont want to keep you away from court trials. Let Fabien learn his lessons. And you coach him. I know that you cant win such a case.
What makes you so omniscient that you can say that with certainty? A trial by jury is anything but a certainty. Juries are human.
It is. You copy a few years worth of effort and expend a few minutes of effort. You just saved several years of writing, debugging, testing, tuning, and such.Still I would prefer to applaud you and your teamsters because I want to know who the real hypocrits are in the story. It's so funny. You are talking about copying as if that were easier than writing from scratch.
I think the assumption about hardware is way too speculative. There are good systems out there. Cray makes one. As does Sun. And others. The cluster Rybka is using is a toy compared to some real shared-memory hardware platforms that are around. But they are expensive. However, I proved, years ago, that one could gain access to a machine that sold for way over $30 million, and use it to play chess. So the cluster superiority is not guaranteed. It is not even convincing me it is that much stronger than a good 16 core box anyway... way more hype than substance, IMHO.But layman Rolf says, it must be much more difficult. Proof: why not many walked on that same highway and became Champions? Houdini wont make it because on a single machine for two programs we have no hardware problemsolving. But in all his tournaments Vas had the best combination of software and hardware. Therefore it's boring what Martin does. Look, if I would fill you up with tons of whisky, I would be the better teacher of us in computerchess! It's a bad example but you get the drift. gringrin
I am pretty sure the cluster is not hype. I have looked at some games and it is incredibly strong. Of course you could probably argue that the use of the hardware is not optimal, and so many cores could be put to better use by using a different cluster setup ... but at the moment the cluster is definitely much stronger than a 16 core box. Some time back Lukas from the Rybka forum posted that he tested a 40 core cluster against a 4 core box and I think he mentioned an over 200 ELO advantage. I am not sure if that was against identical Rybka executables, but if it was, then that is pretty impressive ... especially since the test games were done at fast time controls where I would think the latencies of the cluster would be at a disadvantage.
-
- Posts: 227
- Joined: Tue Oct 05, 2010 5:29 pm
Re: A Very Novel Idea Concerning Vas- BE FAIR
Understood, which is why I mentioned elsewhere that it was unlikely that anything would come of this legally. The FSF would have more resources than Mr. Letouzey for litigation. Unfortunately, they have bigger fish to fry.bob wrote:Yes, but there has to be some sort of financial justification, and someone with enough credibility to raise the issue. Comparing a binary to a Source program is a challenge, to say the least. If it were Microsoft doing this, I am sure FSF would be more than willing. But court costs would be substantial. Lawyers are not free, nor are the expert witnesses required. Money thrown into a black hole with little hope to get it back since computer chess doesn't exactly make millionaires out of authors.Tom Barrister wrote:So Mr. Rajlich hasn't contributed a thing to chess programming, other than Rybka itself? Of course, that would make sense, since he makes his living, or a substantial part of it, from the sale and license of Rybka. Or it might be that he has nothing original to contribute.
I don't understand why it's Mr. Letouzey's choice to go to court anyway. Wasn't it established that he gave the license (or whatever it's called) for Fruit to the Free Software Foundation? Wouldn't it be the prerogative of the FSF to pursue litigation?
This production is being brought to you by Rybka: "The engine made from scratch.™"
-
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: A Very Novel Idea Concerning Vas- BE FAIR
I do not believe the +200 Elo at all. Why? 200 Elo represents at least 3 doublings of speed, perhaps a bit more. 4 to 40 is 10x, or 3+ doublings. No way the cluster gets an optimal speedup.M ANSARI wrote:bob wrote:Here's a deal for you. I will likely retire in 2 years. At that point, offer me $100,000 to disassemble any two commercial engines of your choice and compare them to an open source program of your choice. I'll take the job and do it. That is a year's effort, for a year's pay. That is a pretty good estimate of the time it would require.Rolf wrote:Bob, thanks for your never tiring eloquence, but I must correct a delusion in computerchesscode once and for all. Your comparison with literature is a lame duck. The animal wont fly anymore.bob wrote:There are two ways to do things. Nobody is not doubting the programming skills of Vas. But you can write your own program, or you can copy another and save 1-2-3 years of effort, and start your own effort on top of the program you copied. You might get to the same point, skill-wise. But not time-wise. And not effort-wise.Rolf wrote:It's not just a question of looking and seeing. IMO.bob wrote:Not is it is not so convenient and you now don't bother with his opinion at all...
OK, you watch someone kill somebody, with your own two eyes you see the entire act. Do you consider him innocent until he is found guilty in court? I do not. I _know_ what I saw. Same with the fruit/rybka issue. I _know_ what I saw.
Go for it if you are so sure about it. I am the last who could contradict you because I am a technical layman.
But I know what I saw, namely the titles Vas got after making his program always stronger and stronger. Surely not through always finding new tricks in the old Fruit on Chrismas Eve. That is the reason why all the other programmers dont support the group of 4 or 5. 295 are just watching what is going on and possibly are shocked like me because it doesnt make sense to insinuate that Vas simply copied. Because from where came the material for all the titles???
It doesnt make sense in my eyes. I still beleive in Vasik, that's for sure. It's just not fair how his character has been torn through the mud. NB not every wrong that is done on this planet was intention to betray others. The history of the past five years should have told you a better story than what you are now so much focussed on.
So just because Rybka became #1, one can't use that to justify copying the code of others, any more than an author could copy the plot from one book, the setting from another, the characters from a third, and form that into a best seller..
Say take the Matt Reilly book "Ice Station", change the location to South America, change the main character to willie makeit, and you would have a book that would sell. And be a knock-off at the same time.
BTW, what have we learned in the past 5 years? Certainly not one thing from Vas. He could have cleared this up and everyone could have moved on. But silence keeps it going. In this case, it was Fabien that started the discussion for the N+1th time.
1) you have no plot at all - chess is always the same outcome and since Vas is a good chessplayer he quickly understood the usual tone
2) everything in CC is technical setting. It's all thought and done. Also here no surprise. For all it cant be stolen or copied, because it's always the same. Proof: go and get a Shredder or Junior and then tell me about copying. But you dont want to do this because it would destroy your holy argument against Vas.
Who, in their right mind, would want do that for free? Invest a year of time for something that won't help the person at all? I work on Crafty because I enjoy it. I do other things because I am paid for doing them (teaching classes, research, committees, advising students, etc). I hunt, fish, fly model airplanes, play the guitar, build things, etc. all for free, because I enjoy doing them. I don't enjoy taking a microscope to a program to disassemble it and try to track asm back to C. I know how, certainly. But it is a lot of work. It isn't fun. It pays nothing. So what is the incentive? If there was a legitimate question raised about them someone would likely take the time, myself included. But with no red flags waving, one could spend the time and discover absolutely nothing.
I'd rather spend that year working on Crafty. Or doing things I get paid for, or things I enjoy doing.
It really is that simple, and there is no dark motive lurking in the background.
Programming is a _creative_ task. Not a mechanical one. Ditto for writing a novel, a screenplay, a poem, a song. Designing a building. Or a rocket to go to the moon. Things can be stolen or copied. In any of those. There is a great book "The eagle has landed" which is not about Apollo 11, but is instead about some Data General engineers that openly broke into the DEC plant in Massachusetts and openly went thru the design/hardware of the Vax 11/780 while they were designing the Data General MV series of machines. Copying happens. It saves a ton of time, which when selling things that evolve quickly is a major concern.
3) Yes, you might have characters too but they are also interchangable. Nothing new under the sun. Look, a language has emotions that it can express, but your code must play vchess and not cause you weep some tears.
You miss the point completely, whether that is intentional or not I can't tell. But the idea is about time. If you do things yourself, it takes a lot of time to catch the front-runners. I did this and it took me 11-12 years (I tied for first at the 1982 ACM event and won the 1983 ACM/WMCC event, where my program played its first move in 1968 and started competing in 1976). Time. If you can short-cut the development cycle, you deliver a product quicker. You reap the financial benefits quicker, and therefore over a longer period of time.
4) Now the most difficult part for me as a tech lay. Yes, I think I can imagine what code means. It's a permanent linkage of ordering stuff and here comes something of an ideal effectness into play because what could cause speed and depth that will end in chesswise more correct and deep moves and taken, the chess positions allow something and are not drawish then minimalism beats ranting.
5) Vas cannot have done just copying because that wouldnt have made his code win, he won because he's the better programmer in exploiting the little possibilities where programmers could differ. Everything else is the same and copied more or less.
So this is about time, and money. And saving some of the former to make more of the latter. Wouldn't surprise me if Vas were able to write a world-champion program completely from scratch. But it would have taken far longer. Or one can cut a few ethical corners and get there quicker.
But I dont want to keep you away from court trials. Let Fabien learn his lessons. And you coach him. I know that you cant win such a case.
What makes you so omniscient that you can say that with certainty? A trial by jury is anything but a certainty. Juries are human.
It is. You copy a few years worth of effort and expend a few minutes of effort. You just saved several years of writing, debugging, testing, tuning, and such.Still I would prefer to applaud you and your teamsters because I want to know who the real hypocrits are in the story. It's so funny. You are talking about copying as if that were easier than writing from scratch.
I think the assumption about hardware is way too speculative. There are good systems out there. Cray makes one. As does Sun. And others. The cluster Rybka is using is a toy compared to some real shared-memory hardware platforms that are around. But they are expensive. However, I proved, years ago, that one could gain access to a machine that sold for way over $30 million, and use it to play chess. So the cluster superiority is not guaranteed. It is not even convincing me it is that much stronger than a good 16 core box anyway... way more hype than substance, IMHO.But layman Rolf says, it must be much more difficult. Proof: why not many walked on that same highway and became Champions? Houdini wont make it because on a single machine for two programs we have no hardware problemsolving. But in all his tournaments Vas had the best combination of software and hardware. Therefore it's boring what Martin does. Look, if I would fill you up with tons of whisky, I would be the better teacher of us in computerchess! It's a bad example but you get the drift. gringrin
I am pretty sure the cluster is not hype. I have looked at some games and it is incredibly strong. Of course you could probably argue that the use of the hardware is not optimal, and so many cores could be put to better use by using a different cluster setup ... but at the moment the cluster is definitely much stronger than a 16 core box. Some time back Lukas from the Rybka forum posted that he tested a 40 core cluster against a 4 core box and I think he mentioned an over 200 ELO advantage. I am not sure if that was against identical Rybka executables, but if it was, then that is pretty impressive ... especially since the test games were done at fast time controls where I would think the latencies of the cluster would be at a disadvantage.
As I said, _way_ more hype than fact.
They could show some 4 core and 40 core comparisons. But all we get is "the cluster is for playing chess, not running tests..." That says a lot, because if the numbers were impressive, they would be shown...
Going from 4 shared memory nodes, to 40 cluster nodes, I would be quite happy to see a +50 Elo gain. that is worth something. And possibly reachable.
-
- Posts: 4190
- Joined: Wed Nov 25, 2009 1:47 am
Re: A Very Novel Idea Concerning Vas- BE FAIR
But Lucas said it, then it must be true, since Lukas is a right hand of the God Vas...bob wrote:I do not believe the +200 Elo at all. Why? 200 Elo represents at least 3 doublings of speed, perhaps a bit more. 4 to 40 is 10x, or 3+ doublings. No way the cluster gets an optimal speedup.
As I said, _way_ more hype than fact.
They could show some 4 core and 40 core comparisons. But all we get is "the cluster is for playing chess, not running tests..." That says a lot, because if the numbers were impressive, they would be shown...
Going from 4 shared memory nodes, to 40 cluster nodes, I would be quite happy to see a +50 Elo gain. that is worth something. And possibly reachable.
Some ppl posts are really nothing but an insult for the intelligence...