Laskos wrote: ↑Fri Aug 10, 2018 12:49 pm
Price-wise Nvidia 1060 is a pretty fair comparison to Ryzen 7 1700. A bit lower price for GPU, but higher consumption power-wise. I still think that Lc0 on GTX 1060 performs somewhat better against regular engines on 8 core Ryzen 7 1700 than what TCEC shows. I think that two CPU cores allocation for Lc0 is a bit bogus, one can allocate 2 threads of the same core with very little loss, a large part of CPU cycles Lc0 needs being anyway garbage. In fact I might check the GPU load with affinites set to 2 threads of the same CPU core.
That is probably true with 1060 being the borderline case. However, if you use higher NN cache values and start approaching late middle game IMHO impact of CPU could become noticeable.
But that was not my point anyway. My point was that you need a CPU no matter how weak in order to run Lc0. GPU alone is not enough. And it is a flawed comparison to compare only price and power consumption of GPU vs. some CPU price/power for A/B engine. You have to take the whole package into comparison.
1060 6GB is around 300$ which is identical price to Ryzen 7 1700. You need almost another 100$ for CPU that runs that GPU even if you take the cheapest i3 or Ryzen 3. And it is gonna spend at least additional 30W of power to those 120W of 1060, while Ryzen 7 1700 is at 65W.
So 150W vs 65W and 100$ more whichever way you take it .
And in that configuration strength-wise it is 400-500Elo (depending on TC) advantage of SFdev vs Lc0 with newest test net.
I don't think even die-hard NN fans believe 400-500Elo is something that is realistically achievable on the current 20x256 net...
Gian-Carlo Pascutto wrote: ↑Fri Aug 10, 2018 10:47 am
Will they withdraw Leela Zero from TCEC if DX plays? If they don't, IMHO the statement is without teeth. By playing in the tournament you are supporting the status quo.
Just a bit of my oil to throw on the fire.
If TCEC doesn't respect Lc0 team request to disqualify DeusX, why do you think TCEC would respect their request to withdraw Lc0???
Lc0 is open source, there is literally no way authors can prevent it from being used in a private tournament.
You think Ivanhoe authors would agree to it playing crippled to a single core just to bring some more clicks money to Anton and Chessdom?
TCEC can do, and is doing whatever the hell they want and legally they are entitled to it. Morally, they are a joke, but they obviously don't care.
Gian-Carlo Pascutto wrote: ↑Fri Aug 10, 2018 10:47 am
Will they withdraw Leela Zero from TCEC if DX plays? If they don't, IMHO the statement is without teeth. By playing in the tournament you are supporting the status quo.
Just a bit of my oil to throw on the fire.
If TCEC doesn't respect Lc0 team request to disqualify DeusX, why do you think TCEC would respect their request to withdraw Lc0???
Lc0 is open source, there is literally no way authors can prevent it from being used in a private tournament.
You think Ivanhoe authors would agree to it playing crippled to a single core just to bring some more clicks money to Anton and Chessdom?
TCEC can do, and is doing whatever the hell they want and legally they are entitled to it. Morally, they are a joke, but they obviously don't care.
Lc0 team can send TCEC intentionally a version with bugs so Leela demote as a protest if TCEC does not respect their request to withdraw Lc0
A request to withdraw a program is different than a request to disqualify a program.
A request to withdraw is only about yourself and not about other(assuming all authors of Lc0 agree about it)
A request to disqualify Deus is against Albert silver(I support it but it does not change the fact that it is a request by part of the authors of Deus against one of the authors of Deus)
whereagles wrote: ↑Fri Aug 10, 2018 9:21 pm
ok, let me rephrase that: everybody there is having great fun
and yes I can speak for everybody because that's what I see
While I find aspects of Deus X rather objectionable, Div 4 was really a lot of fun to watch, and I for one am looking forward to the continuation of the tournament.
Indeed the Lc0 copyright was infringed if the engine was mailed to the TCEC admins (or anyone else) without an accompanying offer to provide the sources (or at least being open about those sources being available from Lc0's github).
Now that it is clear that the engine is really an unmodified Lc0 with a new net, everything seems legal from a copyright point of view.
An interesting question is whether there is copyright on the neural net. In my view there is not, since no creativity went into creating it. The only possible creativity would be in selecting the games to feed it, but this selection was made on the basis of functional criteria.
syzygy wrote: ↑Fri Aug 10, 2018 10:44 pm
An interesting question is whether there is copyright on the neural net. In my view there is not, since no creativity went into creating it. The only possible creativity would be in selecting the games to feed it, but this selection was made on the basis of functional criteria.
I think there is definite creativity in creating it, as there are many training parameters that will affect the final output, as well as the choice of games. There's not a single obvious correct procedure to follow. I think Albert said it took 7 attempts on the same data to get something out that played good chess.
On the other hand, the weights are literally a shorthand for a specific algebraic function F(x). I'm not sure you can copyright a math function, no matter how complicated it is, since it is just math. y= 5*a + 2*b - 3*exp(c)*d, just a ton more detailed.
Robert Pope wrote: ↑Fri Aug 10, 2018 11:34 pmI think there is definite creativity in creating it, as there are many training parameters that will affect the final output, as well as the choice of games. There's not a single obvious correct procedure to follow. I think Albert said it took 7 attempts on the same data to get something out that played good chess.
On the other hand, the weights are literally a shorthand for a specific algebraic function F(x). I'm not sure you can copyright a math function, no matter how complicated it is, since it is just math. y= 5*a + 2*b - 3*exp(c)*d, just a ton more detailed.
Not all the creativity is copyrightable. And you are right, weights have only meaning as parameters of math function and are not copyrightable.
The only thing Albert could do, is patent (not really put a copyright since it is not expressed in code) the training algorithm/method.
.