Conversation
There was a problem hiding this comment.
you are allocating upfront 320 MB, do you expect so much data in single TF? The clear preserves the capacity of the vector, so the actual allocation will be done only in the first few TFs, once it reached the asymptotic size, it will reuse the space.
There was a problem hiding this comment.
This was because I didn't know the possible size of this vector. I did some measure on real data and actually I reduced it by a factor 2000.
There was a problem hiding this comment.
actually, resetting of such large array is much faster to do via memset(mCounters, 0, sizeof(int)*Geo::NCHANNELS)
There was a problem hiding this comment.
ReadoutWindowData is POD, why do you use vector instead of the span?
There was a problem hiding this comment.
There is not a specific reason... moved to gsl::span
|
@noferini thanks, I've also edited/committed on spot to remove unnecessary conversion of span to pointer and using |
|
FullCI failure is unrelated, merging. |
* fix for TOF cosmic calib * suppress unneeded dereferencing, [] instead of at()
Reserve memory to avoid allocation when emplacing_back elements (causing problem for high message rates)