Not having issues at all, I'm just curious about the multithreading nature of the twos.
This is a the scenario:
1) A point wrangle node that prints to console the i@id of the points.
2) And a compiled for-loop iterating over points with the very same wrangle (attached picture).
In the first case the IDs will be printed in ascending order.
In the latter case the IDs order will be slightly*** random (but still ascending), and every new cook will be different.
Given the results it seems that the compiled block does a truer parallel cooking, while the operations in the wrangle are still parallel but returned so that they follow the point number order. Does this produce some kind of overhead in the wrangle?
I guess I should do some profiling.
Anyway, I don't really have questions, but would like somebody to tell me more about it and/or correct the above guessing work.
cheers
*** it's just “slightly” random in my poor 4 threads notebook, I guess the randomness would increase in cores rich machines.
wrangle vs compiled block. Order of operations?
855 0 1- Andr
- Member
- 900 posts
- Joined: Feb. 2016
- Offline
-
- Quick Links