Author: Eric Oldre
Date: 13:00:43 07/11/04
Go up one level in this thread
On July 11, 2004 at 15:36:54, Dieter Buerssner wrote:
>On July 11, 2004 at 14:35:50, Eric Oldre wrote:
>
>>a) did I just not let it search deep enough?
>>b) does the fact that it took 3:51 to reach depth 21 indictate a issue with my
>>tranposition table?
>
>Yes. I tried this pos with only 1000 hash entries. It was solved in no time.
>Independent of the hashing scheme you are using (say depth preferred or always
>replace or more sophisticated), your hashing probably does not work well. I
>assume, you have null move disabled.
I did not originally have null move disabled, but it should be trying null moves
on this position due to the lack of material, just to be safe i tried compiling
with no null move. but i got the same results.
>
>>it was running at 1,200,000 nodes/sec.
>
>From this speed, I guess you are generating the hash keys incrementally (in your
>makemove/undomove?). A first start might be to check if they are correct, by
>comparing the incremental keys with ones you generate from start. You might also
>want to describe your hash struct, how you store/probe, when you cutoff, ...
>
yes i am updating the hash keys incrementally. I'll add some code to also
generate them from scratch from a given position and compare to the
incrementally generated one.
I'll report some results when i have them.
>>if there is a issue with my transposition table, could it be due to not having
>>random enough keys? here is the code i'm using to generate the numbers.
>>
>>U64 rand64(){
>> U64 retval = 0;
>> retval = (retval <<15) | rand();
>> retval = (retval <<15) | rand();
>> retval = (retval <<15) | rand();
>> retval = (retval <<15) | rand();
>> retval = (retval <<15) | rand();
>> return retval;
>>};
>
>This will work with many implementations, and will be very poor with other
>implementations (all those, which have a RAND_MAX >= 2^15-1). A fast fix is to
>use xor instead of or, to paste together the rand() returns. I prefer to use my
>own pseudo random number generator. Not necessarily, because I expect better
>results. It makes debugging easier, when you compare runs in different
>environments (the C-library can have different implementations of rand(), and
>then you cannot reproduce node counts, search trees, ...)
I'm using MSVS.Net. I've also tried xor'ing the results from 5 calls to this
function together, to see if that might help any problems i'm having, hasn't
made a difference though.
Thanks for the ideas Dieter!
Eric
>
>Regards,
>Dieter
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.