Cannot filter types/methods when working with the call graph

krzysztofkrzysztof Posts: 8
edited September 27, 2016 12:05PM in ANTS Performance Profiler 9
Is there any way to do the filtering while working with the call graph?

Comments

  • Hi and thanks for your post!

    Regarding your question, it is not currently possible to filter the call graph, but there is the option to collapse stack traces in the call graph--is that along the lines of what you're looking for?

    Jessica Ramos | Product Support Engineer | Redgate Software

    Have you visited our Help Center?


  • Hi, we try make analysis of performance and found that some parts of the code hide others. The call graph is very usefull but we are unable to see some information because of not being able to filter methods/types. Say we have assembly A and B. And we are not interested in improving assembly B but it remains on call graph as the major thing which most of available resources.
  • Ah I see, many thanks for those details!

    If I'm understanding correctly then, the issue is that with a very large call graph, it can be difficult to view all the relevant parts of the graph at one time if there are too many method calls to other assemblies you're not interested in?

    In that case, I've gone ahead and added a feature request for filtering the call graph to help with this, logged internally as PP-3954.

    Thanks very much for your feedback and apologies there isn't a way to hide the calls from certain assemblies at the moment!

    Jessica Ramos | Product Support Engineer | Redgate Software

    Have you visited our Help Center?


  • Hi, There are many scenarios where the graph filtering can be used. Too many method calls can be one of them. Out is a bit different. We analyse, find issues and improve the performance. After a while we see the most overhead is in assembly B in method B1. But we know it will be (it's some kind of sync mechanism) and would like to search for other parts which need improvement. This one (B) cannot be improved for now. Currently we are unable to analyse other parts because B is hidding others. If you could improve this it would help us a lot.
  • Ah, thanks very much for clarifying that! Just to be completely sure I'm following what you mean by methods being hidden though (so I can be sure to create the correct feature request for you)-- do you mean that currently, method B1 is showing as taking up the most CPU, but since you're not interested in this method, you want to be able to completely exclude that from the call graph so that the next method taking up the most CPU will be shown as the most expensive?

    (Also, I don't think these are related in this case, but in case others are running into a different issue where some methods are completely excluded from the results, this is likely from the 'Hide insignificant methods' option which will hide methods that contribute less than 1% of your application's total CPU time, and also a profiling option in Tools -> Advanced Options called "Avoid profiling extremely trivial methods" which will automatically exclude very short methods from results.)

    Jessica Ramos | Product Support Engineer | Redgate Software

    Have you visited our Help Center?


  • Hi Jessica. Yes, that's the case :)
  • Thank you for confirming that. :)

    In that case, we do have a feature request for filtering out the actual timings for certain types/methods logged as PP-784 so I have +1'd that on your behalf. Thank you again for your feedback!

    Jessica Ramos | Product Support Engineer | Redgate Software

    Have you visited our Help Center?


Sign In or Register to comment.