Program Profiling trouble

Alex Chupahin — Thu, 13 Oct 2011 11:30:13 GMT

A program written in C eats CPU. It is need to detect which function do this.

Of course, I use profiler PCA.

With small programs all ok. I can collect execution data and do queries to see any things.

But when I try to do the same things, but just sleep for a day with tonns of megabyte collecting data.

Is it normal setuation?

How can I detect a place of the program eating CPU?

Re: Program Profiling trouble

Ruslan Laishev — Thu, 13 Oct 2011 11:42:19 GMT

Hi, Alex!

Aligning ? Debug ?

Re: Program Profiling trouble

Hoff — Thu, 13 Oct 2011 13:55:05 GMT

There's unfortunately not enough detail in that posting for a response. Paraphrasing your posting, I know that you have a performance problem with an application, that you have a busy CPU of some unknown type, and that PCA is collecting an unspecified and reportedly large quantity of data.

Some additional and specific details, please? Please start with the hardware system model, the OpenVMS architecture and version, the C compiler version and (if this is an open-source application) the identity of the software involved, and how much data was logged by PCA. How big and how complex is this target C application?

Additional details such as the working set size as compared with the pagefault rates can help, too; there are many reasons and many different cases for an application to be slow and CPU-bound. Many of these cases are due to the application code. Some cases are due to quotas and settings and system resource limitations.

When the application (eventually) finishes its run, can you then profile it using the collected PCA data? You're probably not going to be running PCA in production, after all, so the collection process is a one-off or a few-off monitoring process.

And yes, if this is OpenVMS I64, alignment faults are always worth a look.

topic Program Profiling trouble in Operating System - OpenVMS

Program Profiling trouble

Re: Program Profiling trouble

Re: Program Profiling trouble