Fathom
Fathom copied to clipboard
basil00
Optimized FEN-parsing a bit: +/- 50% faster.
Tested by invoking it 100 million on a test-FEN ("1nbqkbnr/1Bpppppp/r7/p7/6P1/8/PPPPPP1P/RNBQK1NR b KQk - 10 1") and then comparing the value of white/black/kings/etc with those of the original code.
Hi, I am wondering what is the use case or motivation behind this optimization? I did not consider there was any need to optimize the parse FEN function.
I made the code into a library which I call from my chess-program. I give it the position as a fen-string because I'm not using bitboards.
On Thu, Dec 3, 2015 at 1:16 PM, basil00 [email protected] wrote:
Hi, I am wondering what is the use case or motivation behind this optimization? I did not consider there was any need to optimize the parse FEN function.
— Reply to this email directly or view it on GitHub https://github.com/basil00/Fathom/pull/1#issuecomment-161618003.
www.vanheusden.com www.slimwinnen.nl | www.smartwinning.info
Generating and then (re)parsing FEN strings sounds very inefficient! Would it not be better to generate the bitboards directly? It probably should not be that hard to do.
I would have to look into that.
But as your program already accepted a fen-string on the command-line, I also for that reason found a fen-string to most logical for a library.
I did not submit the changes for the library, can do if you like.
On Thu, Dec 3, 2015 at 1:21 PM, basil00 [email protected] wrote:
Generating and then (re)parsing FEN strings sounds very inefficient! Would it not be better to generate the bitboards directly? It probably should not be that hard to do.
— Reply to this email directly or view it on GitHub https://github.com/basil00/Fathom/pull/1#issuecomment-161618937.
www.vanheusden.com www.slimwinnen.nl | www.smartwinning.info
Hi,
Please merge https://github.com/flok99/Fathom/commit/1015eceaa54658e45f0e7b8bb5eee86ba9a726bc
It allows you to compile it on non-x64.
On Thu, Dec 3, 2015 at 1:16 PM, basil00 [email protected] wrote:
Hi, I am wondering what is the use case or motivation behind this optimization? I did not consider there was any need to optimize the parse FEN function.
— Reply to this email directly or view it on GitHub https://github.com/basil00/Fathom/pull/1#issuecomment-161618003.
www.vanheusden.com www.slimwinnen.nl | www.smartwinning.info
Using gcc extensions instead of x86 assembly is an improvement. But ideally we should make the code also portable to other compilers, e.g. msvc. I need to investigate the best way to do this.