No bishop magics with fixed shift 8

[niklasf]

Here's an iterim result of my search for magic factors, that may be interesting: There is no magic factor with shift 8 for the bishop square d5. Therefore fixed shift magics will always have to use shift 9.

Using the trick from http://www.talkchess.com/forum/viewtopic.php?t=65187 I had to test only 2^50 candidates instead of all 2^64. This yielded a list of 364854 magics with shift 9: https://raw.githubusercontent.com/nikla ... hop-d5.txt

None of them works for shift 8 and the index range is from 0 to 511 for all of them.

Here's the program I used for this search: https://github.com/niklasf/magics/blob/ ... c/magics.c. It uses one more trick I haven't seen before: Try to order the relevant occupancies so that candidates are refuted as quickly as possible. Any other ideas for speeding this up are most welcome. Moving on to rooks now, after this warmup.

[ZirconiumX]

This line looks ripe for an OpenMP pragma, and the code looks to be thread-safe. You might find that helps.

[niklasf]

Yes, thanks. So far I've split the range into chunks and used GNU parallel, which requires a bit of bookkeeping.

[ZirconiumX]

This line looks ripe for an OpenMP pragma, and the code looks to be thread-safe. You might find that helps.

I think I spoke too soon; `refs` and `magic` are shared, and some bookkeeping that seems irrelevant is done. Still, your method of GNU Parallel works fine.

EDIT: you could add `-flto -fwhole-program` to CFLAGS, for optimisation purposes.