Hey all,
I am developing a PDG TOPS Graph that utilizes an array of HDA Processors to generate game assets (in phases) over a render farm.

My question is, since I am doing this over a farm and tasks will be ran in parallel, should I be saving the working geometry from one HDA Processor and reloading them to the next? I am a bit confused as just calling a file merge from one HDA to another while running over a network may cause further issues. I have been running tests to determine but they seem to be taking forever so a quick tip would be super helpful.

Thanks,
JP

Hi JP,

I think that I need a bit more information to understand clearly. Do you have multiple HDAs that produce output in a row and each processes the result of the previous HDA? And then you want to run this chain in parallel?

Thanks BrandonA,
I had a chain of HDA Processors in a TOPNET after 2 nodes to collect CSV and JSON Data (attached SS_02.jpg)….What I was trying to do was pipe the output geometry from one HDA to the Next (to break apart the geometry manipulation into steps so that tasks could be done in parallel). At the end I wanted to save out the geometry from that output (separate each group into its own obj file).

What I didn't realize is that the ROP Nodes with Use External SOP selected where running the HDA files as well…so a top net to save out .obj files from a series of Top nodes was taking way too long.

I am performing a few tests to figure out how to do this in the most efficient manner as obviously some very big mistakes were made.

I would like to save the geometry within an HDA Processor but I can't seem to figure that out…even with exposing the save button and file path to the HDA's Typeface settings and setting it to editable. I think that would make the end result (saving out the final geometry .obj files separated by prim groups)much easier.

I assume what I need to do is cache the geo at every step in the PDG and then load that result into the next phase…but saving the end result as multiple .obj files (one for each prim group) without re-running the SOP for every obj section is something I haven't figured out yet.

Any tips or direction to specific areas of the documentation that I should be looking at for these tasks would be much appreciated.

Thanks!
JP