Stockfish hash tables and evaluation display

[lucasart]

In the book I mentioned, the author stated that engines used to clear the hash after every move. Perhaps this is just an antiquated method.

Well, the author of that book is very misinformed then. I don't think any serious engine would do that, neither today not 20 years ago.

It's amazing how, when something is written in a book, people are convinced that it must be true. I used to think the same about the television when I was a kid, and have arguments with my father about it

Anyway, use which ever engine you like for analysis, and do not fear that "reset hash after move".

[michiguel]

In the book I mentioned, the author stated that engines used to clear the hash after every move. Perhaps this is just an antiquated method.

Well, the author of that book is very misinformed then. I don't think any serious engine would do that, neither today not 20 years ago.

Robin Smith was very well informed and a distinguished member of this forum. Some engines did clear hash a while ago between moves. One of the reasons is that they were pre-calculating a lot of stuff before each move, so the hashtable got in the way.

Miguel

It's amazing how, when something is written in a book, people are convinced that it must be true. I used to think the same about the television when I was a kid, and have arguments with my father about it

Anyway, use which ever engine you like for analysis, and do not fear that "reset hash after move".

[lucasart]

In the book I mentioned, the author stated that engines used to clear the hash after every move. Perhaps this is just an antiquated method.

Well, the author of that book is very misinformed then. I don't think any serious engine would do that, neither today not 20 years ago.

Robin Smith was very well informed and a distinguished member of this forum. Some engines did clear hash a while ago between moves. One of the reasons is that they were pre-calculating a lot of stuff before each move, so the hashtable got in the way.

Miguel

You mean precalculating the eval before the search in the form of a piece/square table, to be probed at the leaves of the qsearch ?
You're right, some engines did that, but that was in the paleolithic, or neolithic, I can't remember. Wasn't GNU Chess 4, one of the last dinausore doing precalc eval ?