JupyterLab C++ ROOT

ATLAS Software
1

JupyterLab C++ ROOT does not suport multicore. Passing 10fb to 39fb very long to run with labtop

More events less speed to run

2

Hi @guest_bernard.chaura ,

What were you running exactly? There are a handful of ways to parallelize to speed things up if you would like to.

Best,
Zach

3

C++ Root in jupyterlab are used to run four C++ analysis . have you example of prallel analysis in Docker and jupyterlab

4

Hi @guest_bernard.chaura ,

Sure! In our November tutorial we included a lesson on using RDataFrame, which supports multi-threading for analysis and which runs in a notebook. This particular one used pythonic ROOT, but it is straightforward to translate the same into C++, since it’s “native” ROOT:

Cheers,
Zach

5

IT is tedious . some jpynb files will be usefull in the pdf you have to retye all the code

6

I use you RDataframe . with time out with the example of HIggs it run a long time and stop with time out of waiting a file. Why you do not use the multicore option of ROOT.

7

Mistral propose to me solutions with multicore option of root for reading data in C++ Analysis programs

8

PROOF est il accessible dans la version de ROOT de jupyterlab

9

Hi @guest_bernard.chaura ,

PROOF is no longer supported and is not recommended:

Jupyterlab does not have a version of ROOT internal to it; you might have an older version installed in your cluster?

If RDataFrame times out waiting for an input file, multi-processing will not solve your problem — the problem is with the data input. RDataFrame nevertheless allows you to multithread:

See the section beginning with “In a nutshell”.

Cheers,
Zach

10

Time out can be solved loading localy the files. Multicore reduce drasticly execution time. I want to use your 4 examples of C++, eventually with small changes. PROOF permits this. I am a retired scientific. I do not want to rewrite the proposed C++ examples. Is PROOF activated in jupytertlab?. I have a multicore labtop. The virtualbox was very usefull for 2020 data with the multicore option. for C++ .

11

Hi @guest_bernard.chaura ,

As I mentioned above, jupyterlab does not come with an installation of ROOT. There is no notion of “Is PROOF activated in Jupyterlab”. This is like asking if I can play a tape in your car. If I don’t know what your car is, or if it has a tape player, there is no way to answer the question. If you are running Jupyterlab on your own laptop, then you are free to install an old version of ROOT that still supports PROOF.

If what you would prefer is to use the 10/fb legacy release, then you can simply use this tag:

With that tag and an old version of ROOT, PROOF will be supported.

We’ll discuss re-writing a couple of the C++ analyses with RDataFrame so that multi-threading can be natively supported.

Cheers,
Zach

12

Thanks for these explications

13

which ROOT version are compatible with PROOF to run 10fb_legacy

14

Hi @guest_bernard.chaura ,

The README suggests that v6.10.04 should work.

Best,
Zach

15

Thanks,with what Ubuntu version

16

Hi @guest_bernard.chaura ,

The ROOT version page:

Says that Ubuntu 14 or 16 should work.

Cheers,
Zach

17

Is there a ghcr file to run examples 10fb in Docker, like the four examples of 39fb . with good ubuntu ROOT

18

docker pull Package notebooks-collection-opendata · GitHub for 39fb

19

or docker pull ghcr.io/atlas-outreach-data-tools/atlas-outreach-cpp-framework-13tev:latest` | Télécharger l’image Docker |

20

Hi @guest_bernard.chaura ,

Perhaps you could try:

docker run -it -p 8888:8888 ghcr.io/atlas-outreach-data-tools/atlas-outreach-cpp-framework-13tev:10fb-legacy /usr/bin/bash

?

Cheers,
Zach