Does Jamovi use multiple cores while processing data? Specifically the SEMLj package.
If not, is there a way to make it so? (Except rewritting the package
)Thank you for your help
Raphael
Multicore processing?
-
RaphaelHuber
- Posts: 6
- Joined: Mon Nov 30, 2020 2:16 pm
Multicore processing?
Hi
Does Jamovi use multiple cores while processing data? Specifically the SEMLj package.
If not, is there a way to make it so? (Except rewritting the package )
Thank you for your help
Raphael
Does Jamovi use multiple cores while processing data? Specifically the SEMLj package.
If not, is there a way to make it so? (Except rewritting the package
)Thank you for your help
Raphael
Re: Multicore processing?
jamovi will assign separate analyses to separate cores, but analyses themselves typically don't use multiple cores.
whether semlj can be made to support multiple cores will depend on the underlying R packages (lavaan?). if they don't support multiple cores, then you're out of luck.
R itself doesn't provide a parallelism model, so its up to packages to do parallelism in their own way ... which makes things difficult for us to support parallelism. ideally, jamovi would be aware of when and how analyses are doing parallelism so we can clean up, etc. if something goes wrong with the analysis, etc. so we can place reasonable safe guards on how much memory and CPU the analysis consumes, etc. these things all become more difficult when R packages are implementing parallelism in their own way.
if analyses were written in python, python provides a really great parallelism model which would allow us to support multiple cores in a simple and consistent way. alas.
jonathon
whether semlj can be made to support multiple cores will depend on the underlying R packages (lavaan?). if they don't support multiple cores, then you're out of luck.
R itself doesn't provide a parallelism model, so its up to packages to do parallelism in their own way ... which makes things difficult for us to support parallelism. ideally, jamovi would be aware of when and how analyses are doing parallelism so we can clean up, etc. if something goes wrong with the analysis, etc. so we can place reasonable safe guards on how much memory and CPU the analysis consumes, etc. these things all become more difficult when R packages are implementing parallelism in their own way.
if analyses were written in python, python provides a really great parallelism model which would allow us to support multiple cores in a simple and consistent way. alas.
jonathon
-
mcfanda@gmail.com
- Posts: 462
- Joined: Thu Mar 23, 2017 9:24 pm
Re: Multicore processing?
Hi
some modules, such as GAMLj3 and jAMM employ parallel processing. SEMLj does not yet, but we are testing a new version that does. However, parallel processing is used only for operations that can be easily parallelized, mainly bootstrap confidence intervals. Is that what you were looking for?
some modules, such as GAMLj3 and jAMM employ parallel processing. SEMLj does not yet, but we are testing a new version that does. However, parallel processing is used only for operations that can be easily parallelized, mainly bootstrap confidence intervals. Is that what you were looking for?