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
Is your feature request related to a problem? Please describe.
Currently if we can't map an exec to a stage, it falls into a general bucket where we map it into any stages left. This can obviously be inaccurate. The problem is we map execs to stages based on the metrics and not all execs have metrics that allow us to map.
perhaps we can change our algorithm to look at the execs around execs without stages to say that if the one before and after it are in a stage then this one is likely in that same stage.
The text was updated successfully, but these errors were encountered:
Note this could have more effect now that we are looking more at adding transition times. If an exec isn't mapped and its one that fell back to CPU it has potential to affect that stages overall time more.
Is your feature request related to a problem? Please describe.
Currently if we can't map an exec to a stage, it falls into a general bucket where we map it into any stages left. This can obviously be inaccurate. The problem is we map execs to stages based on the metrics and not all execs have metrics that allow us to map.
perhaps we can change our algorithm to look at the execs around execs without stages to say that if the one before and after it are in a stage then this one is likely in that same stage.
The text was updated successfully, but these errors were encountered: