Hi,
It seems I am now ready to generate partial db.
Before launching a partial generation on the 9 pieces db I tested may program on 4 slices of the 7 pieces db with the following results
3M1K vs 2M1K : 69,12% resolved, 29,81% partially resolved, 1,07% unknown
4M vs 2M1K : 84,49% resolved, 15,23% partially resolved, 0,28% unknown
3M1K vs 3M : 83,45% resolved, 16,42% partially resolved, 0,13% unknown
4M vs 3M : 89,01% resolved, 10,94% partially resolved, 0,04% unknown
Can Ed. or any other guy confirm these figures?
Partial database
-
Ed Gilbert
- Posts: 859
- Joined: Sat Apr 28, 2007 14:53
- Real name: Ed Gilbert
- Location: Morristown, NJ USA
- Contact:
Re: Partial database
Hi Gerard,
I do not have an easy way to confirm your counts. The numbers look reasonable. You can (sort of) verify that your partial db is correct by comparing each position against your full 7-piece db and checking that the values are consistent. I checked my partial 9-piece slices by doing a self-verification -- checking that the value of each position is consistent with the values of all of its successors. If you do both of these tests then I think you can be pretty confident that your build program is working correctly.
-- Ed
I do not have an easy way to confirm your counts. The numbers look reasonable. You can (sort of) verify that your partial db is correct by comparing each position against your full 7-piece db and checking that the values are consistent. I checked my partial 9-piece slices by doing a self-verification -- checking that the value of each position is consistent with the values of all of its successors. If you do both of these tests then I think you can be pretty confident that your build program is working correctly.
-- Ed
Re: Partial database
Hi Ed,
I intend to generate the 4 following partial db:
4M1K vs 3M1K
5M vs 3M1K
4M1K vs 4M
5M vs 4M
In absence of problems my first estimation for the time needed to generate these 4 db is about 7 months
For this 7 pieces partial database I did these two tests without any problem. I am currently running a last test : I simulate a crash of my program and I verify that I can restart the generation process from the last subslice handled. In my implementation a subslice last about one hour.Ed Gilbert wrote:Hi Gerard,
I do not have an easy way to confirm your counts. The numbers look reasonable. You can (sort of) verify that your partial db is correct by comparing each position against your full 7-piece db and checking that the values are consistent. I checked my partial 9-piece slices by doing a self-verification -- checking that the value of each position is consistent with the values of all of its successors. If you do both of these tests then I think you can be pretty confident that your build program is working correctly.
-- Ed
I intend to generate the 4 following partial db:
4M1K vs 3M1K
5M vs 3M1K
4M1K vs 4M
5M vs 4M
In absence of problems my first estimation for the time needed to generate these 4 db is about 7 months
Gérard
