Re: [Bug-gnubg] Ply analysis does not give equal GWC and MWC in 1 point
Re: [Bug-gnubg] Ply analysis does not give equal GWC and MWC in 1 point match.
Mon, 2 Feb 2009 17:59:19 +0100
Yes, Jon's hack (zeroing something ni
CacheAdd()) worked for me: he was able to reproduce
the #Q0 issue on his machine with my
executables and, as soon as I put his extra line in,
he said issue was gone.
Interestnig thing I tried:
- start gnubg, set analysis&eval&tutor
at expert (0ply).
- enter the posID (2+4OAADb7gYEAA:cAknAAAAAAAA)
and ask for a hint
- the 0ply analysis and MWC/equities
are correct (!!++)
- but if I select all/some the moves
and click on the "0" button, the MWC get messed up
wrote on 02/02/2009 17:00:21:
> Same gcc version here, so that is unlikely to be the issue. I was
> bit quick about it, and the std makefile doesn't echo the actual
> commands used, but I tried using -O3 as well without any effect.
> Does Jon's cache workaround fix either of the cache problems for you?