Re: [Bug-gnubg] Question about higher plies

[Joern Thyssen]

On Mon, Jul 26, 2004 at 10:51:21PM +0200, Jim Segrave wrote
> 
> I think the classification is probably cheap in CPU time compared to
> the evaluation. What might be profitable is trying to save some of the
> information from the classification for setting up the inputs to the
> neural net. since I'd guess there's a lot of similar work being done.

Yes, I've thought about this as well. 

CalculateHalfInputs is the expensive function, and the function is
unfortunately two-sided, so it's not possible to do one-sided caching
which would be very efficient. 

Another idea would be to do some one-sided caching storing the first
matrix-vector multiply inside the neural net evaluation.

Joern

pgp6uOyAUm6eX.pgp
Description: PGP signature