https://bugs.documentfoundation.org/show_bug.cgi?id=148094
--- Comment #31 from Patrick Luby (volunteer) <guibomac...@gmail.com> --- (In reply to Patrick Luby (volunteer) from comment #30) > Created attachment 195272 [details] > Allocations before and during hang using Xcode's Instruments application So I ran the Leaks option in the Instruments application connected to my local Mac Silicon build. In attachment #195272 the graph at the top shows net memory allocated. The left have is before I dragged the fill rectangle and it is about 140 MB of memory usage. After dragging, the memory usage jumps to about 200 MB and stays level while ScInterpreter::Interpret() grinds away. I used Skia/Metal during my runs so memory usage is higher with Skia/Raster but still level after dragging the fill rectangle. So, I don't see any increasing memory usage. Maybe that problem was fixed by one of my recent memory leak fixes? Anyway, what I did notice that is that in the LibreOffice preferences, Calc has an "Enable multi-threaded execution" preference. I have been running with this preference enabled (i.e. checkbox is checked). So I disabled (unchecked) this preference and reran a CPU Profile in the Instruments app. Interesting, the highest CPU usage stack wass the same as in attachment #195212 and, like that run, ScInterpreter::Interpret() is grinding away in the main thread. I am not familiar with the internal workings of the Calc formula interpreter so is it unexpected that ScInterpreter::Interpret() does all its work in the main thread when the "Enable multi-threaded execution" preference is enabled? In other words, if you disable the "Enable multi-threaded execution" preference on Linux or Mac Intel, do anyone else see a hang? If yes, then maybe multi-threaded execution is not occurring on Mac Silicon machines? So my questio -- You are receiving this mail because: You are the assignee for the bug.
