In principle, Parallel Chunk Loop in combination with Port to URI and URI to Port allow the execution of OpenMS/Knime workflows with multiple threads. But the implementation of Parallel Chunk Loop has problems (NullPointerException) and is incomplete (multiport start missing).
IMO, it would be best to implement a parallelised version of ZipLoop which replaces the old one. We will then have the same behaviour as in TOPPAS now.
Below the start of a loop in order to run a multi-threaded workflow in the Knime workflow engine.
IMO, it's bad UX and a major hurdle for adoption.
In comparison, this is what the equivalent workflow in TOPPAS looks like. Parallel execution is entirely managed by the TOPPAS workflow engine in the background.
In principle,
Parallel Chunk Loopin combination withPort to URIandURI to Portallow the execution of OpenMS/Knime workflows with multiple threads. But the implementation ofParallel Chunk Loophas problems (NullPointerException) and is incomplete (multiport start missing).IMO, it would be best to implement a parallelised version of
ZipLoopwhich replaces the old one. We will then have the same behaviour as inTOPPASnow.Below the start of a loop in order to run a multi-threaded workflow in the
Knimeworkflow engine. IMO, it's bad UX and a major hurdle for adoption.In comparison, this is what the equivalent workflow in
TOPPASlooks like. Parallel execution is entirely managed by theTOPPASworkflow engine in the background.