How to get more performance


do you have any suggestions how we can transform 60 quads, which should be controlled seperatly, with the greatest performance?! There are several connections and queries with each single quad are running through and therefore its very slow. We need all this queries for contolling the quads.

Are there any posibilities for VVVV to check every connection only every second sec, or less? At the moment VVVV is checking as fast it could.
We are looking forward to get some ideas and solutions,

ehm. what kind of connections and queries do you have in mind?

just transforming 60 quads is no problem, right?

We made some pictures… we have 3 patches like the 1. picture… we need some values for every quad and for a better overview we put them in extra patches… but now there are TO many connections and we think thats the reason for running so slow.
On picure 2 we have so check many conditions.
I can send the patches for better understanding, but maybe its to complex to understand right now?!

Now we try to remove the unnessesary connections but the most we need.

bsp_1.jpg (165.3 kB)

picture 2… no text …

bsp_2.jpg (41.0 kB)

Are you using 60 sub patches 1 for each quad, or are you using spreads to transform 1 quad 60 times?
60 sub patches will be slower than using spreads, although it may be harder to patch using spreads!

We need 60 different quads, unfortunately.
And each one needs an independent transformation.
We already reduced the whole patches by doing some more efficient queries and using spreads.

But finally we are connecting 60-times the different x and y-values to the transform patch (with the quad). Is there is any better solution for that?

If you can stallone or cons all the queries and transform values into 1 quad subpatch, its more effiecient, I think the DX GUI modules are the best example of what I mean.
It can get difficult keeping tabs on which slice is which sometimes.
But there cases where you cant do that easily, where grouping them down into smaller numbers can help.

well, you should definitely try to do all things as one spread. in addition to catweasels advice, i´d say that you should try to have only one collision subpatch calculating all things in one spread with 60 slices, connected to one transform and one quad.

its a little difficult to help without knowing what your patch is supposed to do (and i dont know what is hidden in your Object subpatch) - but i think with some more learning, you should be able to accomplish doing all calculations in the same spread within only one instance of the nodes - and after you mastered it, your patch would be much easier and faster.

so keep asking…

i would imagine spreadcounts of a few thousands as not a problem at all.

It worked out very good.
We have now one subpatch for calculation spreads with 60 slices and this is connected to one subpatch where one quad is. We didn’t know that you can use spreads also on DX.
Thanks for your help. :-)