#1154: Timers which do not exist on certain processes should not be included in reductions -----------------------------------+---------------------------------------- Reporter: hinder | Owner: Type: enhancement | Status: new Priority: major | Milestone: Component: EinsteinToolkit thorn | Version: Keywords: TimerReport | -----------------------------------+---------------------------------------- The thorn TimerReport reduces Cactus timers across processes to provide minimum, maximum, total, average values etc. However, if a timer does not exist on some of the processes, TimerReport will fail. The solution so far has been to ensure that all timers exist on all processes. However, since the timed value will be 0 on some processes, this makes the minimum and average reductions of limited usefulness.
I propose that TimerReport be modified to allow different timers on different processes, and to only include those that actually exist in the reductions. This will make the timing of different multipatch maps useful (see #1149).
#1154: Timers which do not exist on certain processes should not be included in reductions ------------------------------------+--------------------------------------- Reporter: hinder | Owner: Type: enhancement | Status: new Priority: major | Milestone: Component: EinsteinToolkit thorn | Version: Resolution: | Keywords: TimerReport ------------------------------------+---------------------------------------
Comment (by eschnett):
I thought that TimerReport already supports this?
#1154: Timers which do not exist on certain processes should not be included in reductions ------------------------------------+--------------------------------------- Reporter: hinder | Owner: Type: enhancement | Status: closed Priority: major | Milestone: Component: EinsteinToolkit thorn | Version: Resolution: duplicate | Keywords: TimerReport ------------------------------------+--------------------------------------- Changes (by hinder):
* status: new => closed * resolution: => duplicate
Comment:
You are right. Duplicate of #814 where this was implemented.
trac@lists.einsteintoolkit.org