Performance of Formulas in Families

[iru69]

Is there any performance issues of using the results of a formula from one parameter in another formula parameter.

Example:
Parameter A: Parameter B - Parameter C
Parameter B: Parameter D + Parameter E
Parameter C: Parameter E - Parameter D
Parameter D: 10 (example value)
Parameter E: 20 (example value)

I've always tried to make every formula non-dependent on the results of another formula, so in the case of parameter A, I'd typically write that formula as:
Parameter A: (Parameter D + Parameter E) - (Parameter E - Parameter D)

But I'm wondering if there's any reason to continue doing that from a performance standpoint. I've kept the example extremely simple, but the concern is that as the formulas are processed, when processing hits a formula parameter referencing another formula parameter that hasn't been calculated yet, it might interrupt the processing, skip over it until the results of the referenced formula parameter are calculated, and then loop back to process it - causing a performance penalty. Theoretically, these calcs shouldn't take more than a fraction of a second, but I'm wondering if anyone's noticed any real world implications to performance or other issues?

[twiceroadsfool]

Ive noticed real world implications on some families that have a LOT of formulas in them, and- where there isnt a benefit- i dont do it. That said, theres usually a benefit. The Armstrong Serpentina Ceiling Component Families ive built are the same way. Theyre all the same, just with different mathematical dimensions that are running off formulas with trig. But, i have them nested in to a Line Based family to array them. You change the type and they move realllllllly slow.

But having a different family for each size is stupid. Im not here to serve Family Performance as the end goal, im here to do architecture.

So i wont build an all in one door with 1200 parameters with formulas, but if the multiple conditions dependant on a value, like you have above, i use a parameter in a formula in a formula, etc.

[Silleke]

Time to evaluate first the trade-off using formulas. It will really slow down your work.

[twiceroadsfool]

No way. ive got hundreds of families with tons of formulas. The ONLY ones that start to actually affect work, are the radial arrays, where families are nested four levels deep, and have formulas nested through all 4 levels.

The rest of my families have tons of complex formulas, all relating and calculating based on one another, NO performance issues whatsoever. And we model way more and with way more detail than most places, based on those formulas.

[iru69]

So i wont build an all in one door with 1200 parameters with formulas, but if the multiple conditions dependant on a value, like you have above, i use a parameter in a formula in a formula, etc.

Thanks Aaron! So in conclusion, at least for modest families, you haven't noticed any performance issues with including formula parameters nested within formula parameters, i.e. dependent formulas. Good to hear - I keep most of my "complex" families to no more than a couple dozen parameters, but some of my formulas get really long, and this would help ease the length.

[twiceroadsfool]

LOL, ive got some formulas that- when copied and pasted in to Word at 11 point font... Are 12 lines long.

No worries.

i find its much mroe critical how many times the formula is daisychained through levels of NESTED families. IE, the parametric Radial Arrays- 4 levels deep, if its a formulas with other parameters, and theyre all tied to the other 3 levels of families, they get reallllllll sloooooow.