I'd just like to congratulate all who have participated in the 'EGTs Online' initiative and reached the '3x+' milestone for all EGTs.
This success has required a combination of vision, commitment, technical skill, generosity, patience - and cash. Well done. g
Congratulations to all ...
-
Arpad Rusz
- Posts: 93
- Joined: Mon Mar 27, 2006 5:33 pm
- Sign-up code: 0
- Location: Romania/Hungary
- Contact:
-
guyhaw
- Posts: 489
- Joined: Sat Jan 21, 2006 10:43 am
- Sign-up code: 10159
- Location: Reading, UK
- Contact:
7-man bitbases
Certainly, Marc B could create 7-man bitbases from his 7-man DTZ EGTs.
There are choices here: one could create wtm/btm 2-bit-per position arrays for (win, draw, loss, other).
Or one could create wtm/btm 1-bit-per-position win/not-win and lose/not-lose arrays. The latter would appear to require less space _and_ time in use.
I also believe, on the basis of a theoretical argument, that it is quicker to generate a DTx EGT via bitbase-generation than directly.
SHREDDER 10 shows (I think) the number of accesses to its SHREDDERBASES and Nalimov EGTs: if so, it certainly shows the utility of the former.
g
There are choices here: one could create wtm/btm 2-bit-per position arrays for (win, draw, loss, other).
Or one could create wtm/btm 1-bit-per-position win/not-win and lose/not-lose arrays. The latter would appear to require less space _and_ time in use.
I also believe, on the basis of a theoretical argument, that it is quicker to generate a DTx EGT via bitbase-generation than directly.
SHREDDER 10 shows (I think) the number of accesses to its SHREDDERBASES and Nalimov EGTs: if so, it certainly shows the utility of the former.
g