GlowCode vs. AQTime C++ profiling performance in the real world?

I am a user of AQTime Pro and while the tool is pretty nice, it does have a horrible performance impact on the application under test if you're not careful. (Even if you are careful, the performance impact is often high for the app I'm mostly profiling.)

I've recently stumbled over GlowCode (found it in a few answers on SO) and while it'll be easy to just download the trial and see how it works on my App, I was wondering if other users could confirm their boasting wrt. profiling performance.

So, I'm looking for 开发者_C百科real world assessments of the performance impact of GlowCode (vs. AQTime) for native C++ of people who regularly use these products. (I only fire up the profiler every odd month, therefore any assessment on my part will be very limited.)

---

I have a GlowCode license and in my experience it has very minimal performance impact compared to the other profilers I've used (SciTech .NET Memory Profiler and Visual Studio Ultimate profiler). Though like you, I only fire it up when needed.

I will say that GlowCode's UI is abysmal IMO. Once you understand enough of it to discover the bottlenecks it's okay but getting there is a hurdle. I did exchange email with GC devs any they were grateful for the feedback and even changed one thing for me. They did mention that they are working on a UI revamp and maybe the latest version has that, I'm not sure (I have GC 7).

I have never used AQTime Pro so can't offer a comparison there.

---

You may try out MicroProfiler (there is a performance comparison): it's impact is 5-6 time less than AQTime's and it is OpenSource (free; source code here).

It is realtime as Glowcode and easily integrates with VisualStudio (2005-2014). But unlike Glowcode it is less fragile (for instance, I couldn't enable Glowcode to profile STL classes and algorithms - always have bad hook (instrumentation) status for them).

To enable profiling of a particular DLL/EXE just click 'Enable Profiling' in the project's context menu. Or, you may fine grain the area you need to profile, by manually setting '/Gh /GH' command line options to specific files.