ROOT I/O Meeting

Friday 2 Mar 2018, 16:00 → 17:40 Europe/Zurich

32/1-A24 (CERN)

Brian Paul Bockelman (University of Nebraska-Lincoln (US)), Philippe Canal (Fermi National Accelerator Lab. (US))

Oksana, Danilo, Guilherme, Jim, Zhe

Oksana: submitted a few PRs with enabled LZ4 and updated reference files for tests for a case when a default ROOT compression algorithm is LZ4

Danilo: working on serialization/CMS

Guilherme: planning to add lzma as a ROOT new builtins

Jim: testing a parallel reading performance(with EnableImplicitMT()) to able to see if there is any performance improvement by using parallel decompression (test case with TTree with a big amount of branches ~1000). Jim doesn't see any improvements yet.
Danilo recommended to contact Enric, as a one of original authors.
Zhe reported that he had a similar problem, but a test case was a bit different (TTree with few branches and with many subranches), while in Jim's case all branches are top branches. Zhe reported that in his case perf showed that most of CU consumsion was caused by "tbb_receive_work" calls, while trying to to access subranches.

Zhe: submitted PR on a signal handling, blocked by ongoing work of Axel to finish backtrace function for OSX

16:00 → 16:20 Round Table