I have installed WebODM in a system with 251GB RAM and 4TB HDD, processed multiple datasets sequentially, and processed parallelly by changing the --parallel_queue_processing parameter value to 5.
In both cases, RAM consumption is not the fullest but the processing time of parallelly processed tasks is much higher than the sequentially processed tasks (Note: Data sets are the same).
What is the difference between the two configurations?
Why the processing time is different even there is enough RAM available?
What are the other options available to achieve parallel processing of tasks with less time?
Much of OpenDroneMap is multi-threaded, but there are large portions that aren’t. So, if you aren’t running out of RAM, throwing a bunch of projects at it will run faster, especially with that kind of thread count.
My apologies for the late reply, busy with some other works. I tried processing multiple projects instead of multiple tasks and seems it is working. I didn’t compare processing timings with other approaches, I will process a few more datasets and Let you know the findings.