I don't think you can record/probe the main transposition table during
quiescence search, because the evaluation is fundamentally different,
taking only captures etc into account.
Would it make sense to add another transposition table just for the
quiesence searches? Does anyone do this, tried it?

(pd42) writes:

I don't think you can record/probe the main transposition table during
quiescence search, because the evaluation is fundamentally different,
taking only captures etc into account.

You can definitely use the main transposition table in the quiescence
search, without any sort of problems. I have always done this
myself. There is no reason to worry about the things you do
differently in the qsearch. You can easily determine whether a
transposition table entry is from the main search or the qsearch by
looking at the depth. Transposition table entries with depth = 0 are
from the qsearch, while entries with depth 0 are from the main
search.

Would it make sense to add another transposition table just for the
quiesence searches? Does anyone do this, tried it?

A lot of people do this. It reduces the size of the tree a bit, but
at the same time the speed (as measured by nodes/second) is reduced.
Whether it is worth the price depends on the engine. In my engine, it
is a big win.

--
Tord Romstad