| [ATL] performance tips [message #70523] |
Sat, 05 January 2008 09:55  |
Eclipse User |
|
|
|
Hi all, I have some problem of performance during the execution of ATL
higher-order transformations: they take about one/two minutes to finish.
My ATL transformations are composed of:
- amw and emf handlers for input models
- approximately 20 rules
- 20 helpers
- 1 entrypoint rule
- 1 endpoint rule (with 20/30 calls to the inject("OUT", "ebnf2", ...)
method)
- about 30/35 calls to thisModule.ResolveTemp
- almost every rule contains a call to some element of the input metamodels
(i.e. MOF!EModelElement.getInstanceById(modelName,classifierID))
- about 10 calls to the inject("OUT", "ebnf2", ...) method
Does anyone know how ATL virtual machine handles such type of
transformation, where is the bottleneck, and how to make a better
performance tuning of my transformation?
Thank you, Ivano
|
|
|
|
|
Powered by
FUDForum. Page generated in 0.03735 seconds
Home