Hi,
I have a FME 2016.1 workbench that contains a Custom Loop transformer and within that I also have a PythonCaller with some Python Code.
The workbench outputs the expected result but it takes over an hour to run. I need it to run ideally in less than 20 minutes.
I have attached some screen shots of the main workbench and the custom transformer. Does anyone have ideas on things I could do to speed this up?
Thanks,
Like @davidrich said: without the actual workbench and test data it's hard to give much specific advice.
One thing you can try is have the custom transformer openen when you run the workspace, by looking at the live feature counts you can get an idea of where time is lost
And just so you know, it may be that "an hour" is how long this takes.
looks like the iteration addition, seeming to pose as a index, is not synched with the process.
Is the loop connection not totally superfluous? Looks like you can do this with just one input.
I suspect by the looks of it, that a lot of lists are created not in sequence but simultaneously, which might account for the performance issue.
It is not actually processing "one at a time" afacs (as far as I know...) and can see.
But, you are processing groups. I don't think you need to use a loop at all.
deductive comment based on process structure analysis.
looks like the iteration addition, seeming to pose as a index, is not synched with the process.
Is the loop connection not totally superfluous? Looks like you can do this with just one input.
I suspect by the looks of it, that a lot of lists are created not in sequence but simultaneously, which might account for the performance issue.
It is not actually processing "one at a time" afacs (as far as I know...) and can see.
But, you are processing groups. I don't think you need to use a loop at all.
deductive comment based on process structure analysis.
looks like the iteration addition, seeming to pose as a index, is not synched with the process.
Is the loop connection not totally superfluous? Looks like you can do this with just one input.
I suspect by the looks of it, that a lot of lists are created not in sequence but simultaneously, which might account for the performance issue.
It is not actually processing "one at a time" afacs (as far as I know...) and can see.
But, you are processing groups. I don't think you need to use a loop at all.
deductive comment based on process structure analysis.
The time taken to process the whole file is now only 10.8 seconds!!
Thanks so much for pointing this out @gio!