[Free]

When I want to watch games from myhands, and I want to check out if there's a killing lead, GIB usually hangs. No result is found! From trick 2 there's no problem, but when no card has been played I rarely see a solution. At BBO I have similar problems, but not as much as with myhands...

Is there anything I can do? Or is this a bug?

Note: sorry if someone else mentioned this before, I didn't find it

[inquiry]

The further you are into the hand, the faster gib solver (I assume this is what you mean) works, as less lines are possible. Trick one is often slow especially on low level contracts that just make or just off one. Best is to open a new window and let the one not working solve in background. Another option is to click "hide double dummy analysis" then reclick slow dummy analysis. I have found if you wait long enough, the solution will show up.

[Walddk]

Regarding the double dummy analysis, GIB is more useful than DF because GIB tells you how many tricks declarer can make on best declarer play and defence. DF only tells you if a contract can be made or not.

However, DF is better when it comes to speed. As soon as the auction is over, DF has the solution, GIB has not. As Free pointed out, it takes ages for GIB to announce anything before the opening lead has been made.

Roland

[hotShot]

Within the submenu called when you press the GIB button, remove the upper and the lower option. You will get a DF like answer much faster.

I have had deals that took GIB 2-3 minutes to analyse, other needed only seconds. If you have a slower computer it can be even longer. But it's worth waiting.

I assume GIB is not just running a single double dummy analysis, but one analysis for each possible lead.

[pigpenz]

i too at times have thought that it didnt work, but sometimes it might take GIB up to 5 minutes to come up with a solution....once there it starts working faster.

[Free]

Tnx all. In this day and age, you'd suspect computers can handle a simple card game in seconds, but apparently 13 levels deep is a lot. 12 levels is a lot faster. The search tree is obviously exponential, so that might explain the problems.

I'll need more patience

[uday]

Was the older DF always very fast or did it also run into this issue? GIB does do more work each time you ask it for help. Maybe we should have it also support the simpler DF 'y/n' functionality

[pigpenz]

it seems like the old DF was faster but again it was only saying yes or no wether contract could be made...GIB shows +/- on the contracts

[fred]

DF is definitely faster than GIB (sometimes much faster), because it answers simpler question than GIB does (which requires less computation and therefore less time).

Changing the options on the GIB menu will not impact GIB's speed - it will only change what GIB displays.

I will consider adding a "fast DD analysis" option in a future version.

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com

[DenisO]

inquiry, on Apr 27 2006, 12:14 PM, said:

Another option is to click "hide double dummy analysis" then reclick show double dummy analysis. I have found if you wait long enough, the solution will show up.

This is a very good tip. I've found that when GIB appears to be frozen, this tip gives me a result very quickly

[barmar]

Is there really a different DD algorithm depending on whether you want to know yes/no or how many? I figured that DF actually computed how many tricks could be taken, and then determined yes/no by comparing this with the contract. I know DF can calculate the number of tricks, because ACBL uses DF to produce the little charts on hand records showing the maximum contracts that can be made in each suit.

[fred]

barmar, on Apr 28 2006, 03:01 PM, said:

Is there really a different DD algorithm depending on whether you want to know yes/no or how many? I figured that DF actually computed how many tricks could be taken, and then determined yes/no by comparing this with the contract. I know DF can calculate the number of tricks, because ACBL uses DF to produce the little charts on hand records showing the maximum contracts that can be made in each suit.

Yes it is different.

The difference is that the depth of search is less for the yes/no problem than it is for the "how many tricks?" problem (because in the former case you can stop parsing as soon as the contract has been made or defeated).

I believe that the "how many tricks" answers that DF provides on ACBL hand records are generated by multiple calls to DF's yes/no engine.

Fred Gitelman
Bridge Base Inc.
www.bridgebase.com