Overhead During Profiling
Brian Donahue
Posts: 6,590 Bronze 1
Hi Josh,
Yep, it was right there in the FAQ all along. By the way, ANTS Profiler
will make your application run noticably (understatement here!) slower than
normal. The Profiler will try to compensate for the overhead incurred as
best it can to try to deliver 'real-world' timings for the profiling
session. I've been told that it's 99% accurate, but not perfect.
Regards,
Brian Donahue
Red Gate Technical Support
"Josh" <jcarlisle@nospam-removeme.viewfusion.com> wrote in message
news:SJGT9e9LEHA.2488@server53...
> I'm currently evaluating the Ants Profiler. I must so far I'm extremly
> impressed and I wish we would have been using it a few months ago to save
us
> the hassle we are currently experiencing. My question is related to the
> overhead involved during profiling. We are currently using your profiler
to
> try to find bottlenecks in a windows based application that is based in a
> Citrix Environment. During profiling in our citrix environment and on our
> local development machine there seems to be a good bit more overhead when
it
> comes to the processor (haven't evaluated the memory) during profiling
then
> during normal execution without profiling. For example in one of our
problem
> areas I've noticed a CPU spike of around 10-15% but during profiling that
> CPU spike is around 25%. Normal flow of the application is noticably
slower
> also. It seems reasonable that the profiling process contains overhead, is
> the amount I've noticed normal and within normal limits and is this
overhead
> included in the metrics returned or is it just the overhead of profiling
but
> doesn't affect the validity of the results?
>
> Thanks for your time.
>
> Josh Carlisle
>
>
Yep, it was right there in the FAQ all along. By the way, ANTS Profiler
will make your application run noticably (understatement here!) slower than
normal. The Profiler will try to compensate for the overhead incurred as
best it can to try to deliver 'real-world' timings for the profiling
session. I've been told that it's 99% accurate, but not perfect.
Regards,
Brian Donahue
Red Gate Technical Support
"Josh" <jcarlisle@nospam-removeme.viewfusion.com> wrote in message
news:SJGT9e9LEHA.2488@server53...
> I'm currently evaluating the Ants Profiler. I must so far I'm extremly
> impressed and I wish we would have been using it a few months ago to save
us
> the hassle we are currently experiencing. My question is related to the
> overhead involved during profiling. We are currently using your profiler
to
> try to find bottlenecks in a windows based application that is based in a
> Citrix Environment. During profiling in our citrix environment and on our
> local development machine there seems to be a good bit more overhead when
it
> comes to the processor (haven't evaluated the memory) during profiling
then
> during normal execution without profiling. For example in one of our
problem
> areas I've noticed a CPU spike of around 10-15% but during profiling that
> CPU spike is around 25%. Normal flow of the application is noticably
slower
> also. It seems reasonable that the profiling process contains overhead, is
> the amount I've noticed normal and within normal limits and is this
overhead
> included in the metrics returned or is it just the overhead of profiling
but
> doesn't affect the validity of the results?
>
> Thanks for your time.
>
> Josh Carlisle
>
>
