Search - User list
Full Version: rop cook slow on the first cook
Root » PDG/TOPs » rop cook slow on the first cook
Hello there!

I am working with tops and I don't understand if I am getting an issue or I am missing something.

I have a very simple setup with a grid and a noise on it.
I create something like 100 wedge and an attribute wedge on the height.
After that I put a rop geo and as soon I try to cook it seems to be quite slow.

now, if I change the wedge attribute it cooks instantly all the wedges. As soon I delete or increase the wedges count in the first wedge top it takes a little bit to cook down again all the wedges.
Is this a right behavior?

also, why there is no flipbook top? can we just use the mantra one?

Hi Paolo,

Can you attach your .hip file?
On the ROP fetch tab of the ROP Geometry node is a “Cache Mode” parm which is set to automatic in your HIP file. The first time that you cook the TOP graph, none of the geometry files have been produced, so the ROP geometry will cook each work item. The second time you are running it, because Cache mode is set to automatic, the ROP Geometry node will first look on disk to see if the output files already exist. If it finds them, it will mark the work item as cooked instantly. If you increase the wedge count, and recook the ROP geometry (and cache mode is set to automatic) it will use the geometry files it finds on disk and mark those corresponding work items as cooked. For the other work items where geometry hasn't been produced yet, it will cook the work item.

If you don't want it to use the cache, you can set cache mode to “Write Files” and it will always cook the output.

so a Top Rop output generation is basically taking more time due to it's creation, taht's it right?
Beacause I saw that if I put down some frames instead of just one is blazing fast!
There's some overhead associated with spawning your jobs in the ROP geometry node. It has to call a Hython script which sets up the job to cook the SOP network (this all happens out of process). Since it's a very simple SOP network that takes a negligible amount of time to cook, that associated overhead time seems far larger than the actual amount of time to cook.

That being said, I'm looking at your HIP file, and it seems like maybe it's not doing exactly what you intend. I see that the SOP network references $F, but the ROP geometry is set to only generate a single frame. So for each wedge, you're only cooking the first frame. If you set “Evaluate Using” to Frame Range, it will cook the entire frame range for *each* wedge, and you will see similar cooking times as if you were scrubbing the timeline.


Beacause I saw that if I put down some frames instead of just one is blazing fast!

Oops, I missed this part–it looks like you already discovered what I mentioned.
ok! this is clarifying the process! thanks Brandon!

and btw guys, I love tops, you did such a great job! amazing!

THANKS SIDEFX !!! whooop whooop!!
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Powered by DjangoBB