Re: [gnugo-devel] cosmic patch

[Stéphane Nicolet]

> Dan wrote:
> > After the patch, the regression results are ALMOST the same as
> > the current CVS. There is one exception, and unexpected pass in
> > nngs3:
> > 390 unexpected PASS!
> > An unexpected pass is of course not bad but do we understand
> > why the results are different?

Arend replied :
> Are you sure you used otherwise identical versions? I get that PASS
> in CVS, too.


It seems that the fail/passed status of that test in essentialy
random. I think we are in front of a weird cached memory issue
there, or maybe hitting a floating point precision limit.

With the *same* executable (current cvs, non cosmic, without
recompiling), I get different results when I issue the two
identical commands in a row in a console window :


gnugo --quiet -T -l 
regression/games/nngs/Lazarus-gnugo-3.1.31-200204140254.sgf --until 90
gnugo --quiet -T -l 
regression/games/nngs/Lazarus-gnugo-3.1.31-200204140254.sgf --until 90

The problem is that the opponent white P10 stone is sometimes
considered critical, and sometimes not, by the initial dragon
calculations.

If it is not critical, gnugo plays L11.
If it is, gnugo plays O11.


On my system, the bug seems to be dependant on wether I'm
connected to Internet, wether I'm running lots of applications,
etc. Quite strange.

Stephane.
































... /// ... \\\ ...