ROOT Parallelisation, Performance and Programming Model Meeting

Thursday 5 Oct 2017, 16:00 → 18:25 Europe/Zurich

40/R-D10 (CERN)

Danilo Piparo (CERN)

CERN number: 71400 Extension: 109284483#

Present: Enrico, Omar, Lorenzo, Philippe, Enric, Guilherme, Xavi, Danilo, Enrico, Jim

ACTIONS

- Redo the scaling plot several times to understand what error we have on the measured points, say 10 times, with a bit less events. Also we need to specify the type of the columns of the histograms. Also have more chunks of works to have a better balancing. Another point: measure the time needed for a single event.

Xavi

- TDF does not add a constant overhead wrt the original code, which is nice.

- Strong scaling: TDF and Rene's code - the scaling is identical

- Has an almost working numa aware TThreaded executor

 

16:00 → 17:00 Round table

17:00 → 17:20 Update on Nuons Model TDF implementation

Speaker: Xavier Valls Pla (University Jaume I (ES))

Nuon.pdf