Gpt and parallel execution


browsing through the GPT documentation I see that GPT can be invoked on several source files at the same time, e.g.:

$ gpt graph-file [options] [ source-file-1 source-file-2 …]

does it mean it can apply the same processing to every source file, concurrently?
If so, how would I configure the write operator in the graph in order to have different output files?
I mean … now I would run the following commands:
$ gpt graph-file -t output-file1 source-file-1
$ gpt graph-file -t output-file2 source-file-2

Is there any magic I can do with the write operator in order to run one single gpt command?
$ gpt graph-file source-file-1 source-file-2
but still having both the output-file-1 and output-file-2 created in the same way?

Or not?

Thanks for any hint!



Hi Mario

multiple source files are supported if one graph/operator need multiple sources, like binning, mosaic and others.
The graph is not applied to all the inputs.
The processing is done concurrently by splitting the source into tiles and processing them in parallel.

You will need to call gpt for each of your sources independently.