- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
I'm using VTune 9.1 with the remote collectors. After my runs are over I try and drill down through the process I'm interested in trying to look at counts for particular functions. This one function I tried to get to calculates 2 values and then calls a series of library calls. The library functions are in the list of functions called with their associated counts, but the specific function I was looking for is not listed.
It is like this function was "optimized" out, even though with the calculations there would cause floating point counts to be reflected.
Thanks.
Les
I'm using VTune 9.1 with the remote collectors. After my runs are over I try and drill down through the process I'm interested in trying to look at counts for particular functions. This one function I tried to get to calculates 2 values and then calls a series of library calls. The library functions are in the list of functions called with their associated counts, but the specific function I was looking for is not listed.
It is like this function was "optimized" out, even though with the calculations there would cause floating point counts to be reflected.
Thanks.
Les
Link Copied
1 Reply
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
It's often necessary to make a static link in order to get library events included in a convenient way with your module of interest. Even then, if the library were stripped, the function might not show up with proper identification.
Even if everything is set up OK, if your run doesn't take enough samples, it's possible to miss something. My usual experience with even a run of a few minutes is that functions with less than 1% of the events don't show up reliably. Sometimes, it's possible to make an artificial case which spends enough time in the function of interest to get useful data.
Even if everything is set up OK, if your run doesn't take enough samples, it's possible to miss something. My usual experience with even a run of a few minutes is that functions with less than 1% of the events don't show up reliably. Sometimes, it's possible to make an artificial case which spends enough time in the function of interest to get useful data.

Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page