RDataFrame parallel status


_ROOT Version:6.14
_Platform:ubuntu
_Compiler:gcc


Hi all:
I use EnableImplicitMT(20);
but there seems no parallel using if I ‘top’, only 1 CPU using 100%
how can I see if the MT parallel already in use?

Hi,

implicit parallelism is there and works very well but the cpu usage depends on some factors. For example, if the input of your analysis is a ROOT file, the unit of parallelisation is the cluster within the ROOT file. What is your input?

Cheers,
D

1 Like

To add to @Danilo’s answer, the size of RDataFrame’s threadpool is whatever ROOT::GetImplicitMTPoolSize says at RDF construction time.

Another common reason why you might see only one CPU running is that you might have a long time spent just-in-time compiling large template invocations before the event loop starts – and then a very short event loop. You can check by adding a printout inside the event loop (e.g. a Filter that prints something and then returns true).

2 Likes

well it helps, thanks

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.