What usage of the trtexec option "threads"

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

I currently use tensorrt 8.6.1.6.

What usage of the trtexec option “threads”?
The trtexec help descripts it as “Enable multithreading to drive engines with independent threads or speed up refitting (default = disabled)”. I can’t understand it.
What difference is with/without “threads” option? Is this option used for trt engine conversion? Or it is only used in trtexec to inference?
Does it have any effect on the trt engine inference in multiple OS threads? Or it just impact the trt engine conversion?

Thanks

3

I’m also interested in the clarification for the usage of the threads argument

4

The --threads option helps to load an engine into multiple threads and enqueue parallelly.

5

So it is only used for trtexec inferencing? And does it mean that multiple threads are created for parallel trt inferencing?

6

Yes, this option is for trtexec.