You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The --profile-time option of criterion is very convenient for profiling, but in some cases you don't want to automatically scale the number of iterations, such as when you compare profile of different benchmarks executions.
In such a case, having the same number of iterations for all benchmarks is very convenient, as it allows to easily compare profiling metrics such as instruction count (and other performance counters).
I think that this could be achieved by simply adding a command-line parameter --profile-iter that would behave identically to --profile-time, except that it would set a fixed number of benchmark iterations instead of fixed time.
Besides, it would also be helpful if --profile-time could print the number of iterations it ran.
The text was updated successfully, but these errors were encountered:
The
--profile-time
option of criterion is very convenient for profiling, but in some cases you don't want to automatically scale the number of iterations, such as when you compare profile of different benchmarks executions.In such a case, having the same number of iterations for all benchmarks is very convenient, as it allows to easily compare profiling metrics such as instruction count (and other performance counters).
I think that this could be achieved by simply adding a command-line parameter
--profile-iter
that would behave identically to--profile-time
, except that it would set a fixed number of benchmark iterations instead of fixed time.Besides, it would also be helpful if
--profile-time
could print the number of iterations it ran.The text was updated successfully, but these errors were encountered: