The 26 is the search ply. A ply is a 1/2 move -- so HIARCS is looking 13 moves ahead. Its pretty meaningless because some engines go deeper and some engines prune less and look more thoroughly within a shorter horizon. And sometimes the same engine varies a bit depending on the type of position. About all its good for is giving you a rough sense of what giving your engine an extra minute is actually doing. For example, Stockfish leaps to 20 ply really quickly, but getting much beyond 25 ply takes real increasing time.
What does the second part of the evaluation -0.14/26 mean?
Sort:
ArtNJ wrote:
The 26 is the search ply. A ply is a 1/2 move -- so HIARCS is looking 13 moves ahead. Its pretty meaningless because some engines go deeper and some engines prune less and look more thoroughly within a shorter horizon. And sometimes the same engine varies a bit depending on the type of position. About all its good for is giving you a rough sense of what giving your engine an extra minute is actually doing. For example, Stockfish leaps to 20 ply really quickly, but getting much beyond 25 ply takes real increasing time.
That's very interesting, thank you. I wonder if the algorithms you mention result in the phenomenon of the first suggested line varying on repeat analyses, even when using the same engine to the same depth?
Geordie1974 wrote:
ArtNJ wrote:
The 26 is the search ply. A ply is a 1/2 move -- so HIARCS is looking 13 moves ahead. Its pretty meaningless because some engines go deeper and some engines prune less and look more thoroughly within a shorter horizon. And sometimes the same engine varies a bit depending on the type of position. About all its good for is giving you a rough sense of what giving your engine an extra minute is actually doing. For example, Stockfish leaps to 20 ply really quickly, but getting much beyond 25 ply takes real increasing time.
That's very interesting, thank you. I wonder if the algorithms you mention result in the phenomenon of the first suggested line varying on repeat analyses, even when using the same engine to the same depth?
Not usually no. However, if you go ahead in another line, and come back, the engine might still have some of its analysis in memory, which could change its initial evaluation of the position. That happens with stockfish on my mac, anyway. And also different versions of the same engine, or even the same version on a different computer -- or even the same kind of computer but the engine is configured a little differently, can yield different results.
And just to clarify what I meant by varying, for example in an endgame an engine can often analyze more ply more quickly. So when I said Stockfish leaps to 20 ply very quickly and then tends to get meaningfully progressively slower after 25 ply, I was generalizing to a typical early opening/middlegame position
ArtNJ wrote:
Geordie1974 wrote:
ArtNJ wrote:
The 26 is the search ply. A ply is a 1/2 move -- so HIARCS is looking 13 moves ahead. Its pretty meaningless because some engines go deeper and some engines prune less and look more thoroughly within a shorter horizon. And sometimes the same engine varies a bit depending on the type of position. About all its good for is giving you a rough sense of what giving your engine an extra minute is actually doing. For example, Stockfish leaps to 20 ply really quickly, but getting much beyond 25 ply takes real increasing time.
That's very interesting, thank you. I wonder if the algorithms you mention result in the phenomenon of the first suggested line varying on repeat analyses, even when using the same engine to the same depth?
Not usually no. However, if you go ahead in another line, and come back, the engine might still have some of its analysis in memory, which could change its initial evaluation of the position. That happens with stockfish on my mac, anyway. And also different versions of the same engine, or even the same version on a different computer -- or even the same kind of computer but the engine is configured a little differently, can yield different results.
And just to clarify what I meant by varying, for example in an endgame an engine can often analyze more ply more quickly. So when I said Stockfish leaps to 20 ply very quickly and then tends to get meaningfully progressively slower after 25 ply, I was generalizing to a typical early opening/middlegame position
Fantastic, thanks. These details distract me if I don't have answers, so that's very helpful! All the best.
Hi there, I'm using HIARCS / Stockfish to analyse my games. For each move it throws up the evaluation, but also a second number after a forward slash. Does anyone know what this number means? Is it how deeply Stockfish has analysed the move? E.G: 1.e4 -0.14/26