Completely bizarre times

joehuha

I see times in seconds that are at odds with reality.

A function containing mostly a call to System.Runtime.Serialization.Formatters.Binary.BinaryFormatter.Serialize shows up as taking 30s. The wall-clock time measured by a timer inside my application is about a thousandth of that, and that is likely a lot closer to the truth. (I have double-checked I am not confused between seconds and milliseconds, and the overall execution time is a similar order of magnitude)

Another run of my application takes about 90 minutes. However, the longest "time with children (seconds)" is 72. I'd expect my equivalent of main to take pretty much as long as the whole application?

I have turned off "Simplify very complex stack traces to save memory". I am using "Highest profiling overhead" and all additional options are active except I/O. I am using 9.3.

What could I be doing wrong? Many thanks.

Jessica R

Hi there,

Sorry to hear you're getting some strange results!

To help us rule out some common issues, can I check a few more things:

- In case the weird timings are due to overhead, can you kindly see if you get the same results if you profile using either "Line-level and method-level timings; only methods with source" or "Method-level timings; only methods with source"?
- Does toggling the "Adjust timings to compensate for overhead added by the profiler" option (from Help>Advanced options) make any difference?
- When the main method is shorter to the application runtime, can I just double check if it may have been due to the selected region--was the entire timeline selected at the time?
- Would you be able to share your results file with us? (you can send it to support@red-gate.com with "Ticket #55252" as the subject)

Thank you!