Norm Wattenberger: CVData V2 Testing Program

[Norm Wattenberger]

The testing program for the next version of CVData will start in a couple of weeks. Interested CVData V1 users please contact me.

regards,
norm

QFIT

[Cacarulo]

> The testing program for the next version of
> CVData will start in a couple of weeks.
> Interested CVData V1 users please contact
> me.

I'm always interested in your excellent programs.

Sincerely,
Cacarulo

[Don Schlesinger]

[Franz Joseph]

What additions and changes will appear in CVData V2?

[Norm Wattenberger]

> What additions and changes will appear in
> CVData V2?

Sorry, I avoid pre-announcing features. At least until the features have been completed and are well into testing. I don't want to be in the position of promising something and not delivering, for some reason, for a long period.

QFIT

[Franz Joseph]

Include the rules, playing conditions, etc., which are available on Casino Verite V3.0. We all know that CVData V3.0 will be a superior product.

[Norm Wattenberger]

The problem with including the vast number of rules & conditions available in CVBJ into a high-speed simulator is the impact on speed resulting from handling so many variations. Most odd-ball rules are not tested by commercial sims. I do have a solution designed; but have not yet decided if it will go into V2 or V3.

BJStats

[Franz Joseph]

> The problem with including the vast number
> of rules & conditions available in CVBJ
> into a high-speed simulator is the impact on
> speed resulting from handling so many
> variations. Most odd-ball rules are not
> tested by commercial sims. I do have a
> solution designed; but have not yet decided
> if it will go into V2 or V3.

Wouldn't it slow down the simulation only if one or more such rule is actually in the set of rules tested? Isn't there some way to bypass all the conditional branches by using only the common rules variations?

[Norm Wattenberger]

> Wouldn't it slow down the simulation only if
> one or more such rule is actually in the set
> of rules tested? Isn't there some way to
> bypass all the conditional branches by using
> only the common rules variations?

Yes, it is possible to avoid the many conditional branches when the rules are not in effect. The problem lies in gathering the information in case the rules are in effect. That code must be buried in the heart of the simulator. Placing conditional branches around that code doesn't save anything since branches can be the most time-consuming operations in modern CPUs. There is a solution - but it increases maintenance effort. My intention is to implement it after I have completed a couple of more dramatic changes scheduled for the core code.

Speed is not my first concern. In fact it is fourth behind accuracy, flexibility and maintainability. But, it is factored into every design issue. Running at millions of hands a second means that every single aspect of the core sim requires its own elegance. If one area (RNG, dealing, shuffling, hand resolution, stat gathering, TC calculation, decision handling, betting) is inefficient - all the work in the other areas is wasted. Much like real play

QFIT