1 GB hardware with 1min per move might have comparable strength of pentium III running in 15 years ago.Current SF with standard time control and optimum opening book, such as TCEC hardware is extremely unlikely to be beaten ( >95% draw rate against best available engines). In fact current top engines reach to almost perfection (100% draw would be perfect play by both sides)
Alpha-zero-stockfish (number of moves per second vs hardware debate)
Sort:
You don't need access to alpha, simply run the fish at pick against the crippled one and see what happen.
the thing is that machine learning is huge and will keep growing, Elon Musk for instance just resign from the OpenAl board not to have a conflict with the tesla team.
so I find it very odd for the alpha team to tinker with the fish at all, and wonder WHY they did? that's all to it.
Feb 22, 2018
0
#44
drmrboss wrote:
1 GB hardware with 1min per move might have comparable strength of pentium III running in 15 years ago.
No, not when it is a 64 thread monster (probably a 32 core hyperthreading machine). This is comparable to 1 thread running for 64 minutes.
Fast forward 8 years and now we understand why deepmind's tool was shortly incorporated in stockfish. Alphazero's approach was just superior.
That said, they are both search tree tools. It is just that stockfish ran an algorithm as dictated by humans , alphazero had a certain autonomy, it learnt an algorithm by itself . All this said I no longer consider stockfish, stockfish, but a hybrid that was forced to adapt to the news ways to stay relevant.
Also, and most important as I have said they are tools, they can't play chess. Yes they are perfect for finding the next best move, but:
1. They are not aware they are playing chess , in fact they are not aware they do anything
2. They don't know what chess is and that in its digital form it is a representation of material chess
3. They don't know the pieces their role and symbolism
4. They do not have an awareness of the chessboard
5. They don't know to explain why a line is best than the other . They can't take an initiative based on the positions of the pieces of the board based on logic. They can't feel fear when the queen is endangered in one or a couple of moves in the future.
They just need to run the algorithm. If it says the queen is endangered they could sweat, probably not.
This is not an organic understanding of reality. No more than a calculator could understand math. Yeah if you gave it some instructions on what rules to follow on predefined problems , it would ace some math tests , compared to humans. But this is not doing math, by far.
meh