Singular extensions

[jd1]

Hi,

I have a question. If one implements singular extensions the way it is done in Stockfish, what percentage of the nodes would you expect to be from the zero window, reduced depth, search of all other moved except the trans_move?

Also, how much is time-to-depth affected?

I am getting 12% of nodes are in SE search and time-to-depth is up be a factor of 1.5 so I think I am doing something wrong.

Thanks for your help.

Jerry

[lucasart]

Hi,

I have a question. If one implements singular extensions the way it is done in Stockfish, what percentage of the nodes would you expect to be from the zero window, reduced depth, search of all other moved except the trans_move?

Also, how much is time-to-depth affected?

I am getting 12% of nodes are in SE search and time-to-depth is up be a factor of 1.5 so I think I am doing something wrong.

Thanks for your help.

Jerry

i tried once SE in a similar fashion to SF. the time to depth was only slightly increased, but there was no measurable elo increase in testing.

[jd1]

Thanks. Maybe it doesn't necessarily work for all/most engines.

I counted the number of normal extensions (checks, good see captures, etc.) compared with singular extensions and normal extensions greatly outnumber singular extensions. Does anyone have any idea why it might makes such a big difference for me?

Thanks,
Jerry