I was wrong… vops are multi-thread and Houdini’s performance monitor showed that the tessellate asset was slow because of a for-each sop node. I found a way to avoid it and now tessellating point clouds is lightning fast. I did some testing with a couple of flip fluid simulations and here are the numbers.
From one million particles to 25 million and 2 minutes per frame.
From one hundred thousand particles to 10 millions and 10 seconds per frame.
All tests on a i7 processor machine with 24 Gb of RAM.
You can find the new version of the asset here.