Stockfish solving times question

[Patishi]

Hi,
is it normal that it can solve a certain position in 49 seconds the first time and 1:30 the second time? and so forth..
Sometimes it reaches that certain depth that it solved it the first time but still doesn't come to the solution.
And each time I completely close the program and open it again to reset the transposition table.

[Ajedrecista]

Hello:

Could be if you are analysing with more than one thread/core. I expect SF to be determinist in single core (always gives the same outputs with the same nodes, times would depend on the speed of each run), while what you described has been called SMP luck. With multithreaded searches, each run is different and sometimes there are lucky runs with fast solutions and other times you get unlucky runs with slower solutions or even without solutions in reasonable times.

Try with single core and the analysis should be the same once and again under your conditions (close the programme and open it again to reset the transposition table). Speeds and times would change a bit depending on the run (imagine that something is running in the background and slows SF, while other time SF can run at full speed).

Regards from Spain.

Ajedrecista.

[Patishi]

Thank you very much for the great explanation. Yes, indeed I am using multiple cores.
I didn't know that there could be such a great difference in solving times.
Very interesting.