[Cuis-dev] Problems in class Number

[Andres Valloud]

Do you see any way to improve upon it?  If not... maybe it's time to 
integrate the thing already.

On 10/10/19 20:02, Agustín Sansone via Cuis-dev wrote:
> With this last improvement:
> Time millisecondsToRun:
>         [10000 timesRepeat: [1 to: 1600 do: [:e | e isPrimeFast2]] ]  3502
> 
> Without it:
> Time millisecondsToRun:
>          [10000 timesRepeat: [1 to: 1600 do: [:e | e isPrimeFast]] ]  4982
> 
> It is indeed a 30% speed increase for numbers up to 1521.
> 
> El jue., 10 oct. 2019 a las 23:35, Andres Valloud via Cuis-dev 
> (<cuis-dev at lists.cuis.st <mailto:cuis-dev at lists.cuis.st>>) escribió:
> 
>     Hmmm, in here,
> 
>              self < 384 ifTrue: [
>                      "Approximate sqrtFloor to avoid computational expense"
>                      3 to: (self bitShift: -3) + 1 by: 2 do:
>                              [:each | self \\ each = 0 ifTrue: [ ^false ]].
>                      ^true
>              ].
> 
> 
>     the loop should start at 7 because the factors 3 and 5 have already
>     been
>     excluded.  This gains noticeable speed for smallish integers (because
>     384 sqrtFloor is just 19).  Looking into this further led to another
>     30%
>     speed increase.  Would you mind checking?
> 
>     On 10/10/19 18:40, Agustín Sansone via Cuis-dev wrote:
>      >     Here's something really important: doing too much
>     micro-optimization
>      >     detracts from the motivation to implement better things.  I'd
>     say we
>      >     leave things with the round of 30.
>      >
>      >
>      > Okay. This is the 30*k version.
>      >
>     -- 
>     Cuis-dev mailing list
>     Cuis-dev at lists.cuis.st <mailto:Cuis-dev at lists.cuis.st>
>     https://lists.cuis.st/mailman/listinfo/cuis-dev
> 
>