I have no idea what's available for MSVC. GCC builds can use gprof (Pioneer has a configure option for it) or you can use valgrind/cachegrind if you've got years to spare.

VS2010 profiler should be the first one to try (Analyze -> Launch performance wizard...) but it might only be available from pro upwards.

It should and probably will be but I thought it was Pro and Team editions (and better) only of Visual Studio? I mean I know you can use profile guided optimisation but that's done by instrumented analysis by VS2010 automagically.

I use Instruments (osx) - still learning how it works actually.

You can use AMD CodeAnalyst or Intel V-Tune. But you shouldn't profile a debug build. A debug build contains a lot of checks which make it slow. You can profile a release build.

Oh I know, I'm a developer by trade but we have suites of profiling software for what I work on professionally. It's finding _good_ free profiling tools outside work that is difficult!

The Debug build is the one I'm interested in, there's no point it even existing if it's so slow that it's non-functional. Something happened to it between Alpha's 13 and Pre-Alpha 15 which means that it is too slow to actually use for debugging now with Visual Studio 2010/2008. This appears to be due to some horrific list thrashing which gets so heavily optimised in Release builds that it doesn't make too much difference on a powerful PC... but on the lower-end machines it might still have a negative impact.

Profiling it was just to find out the cause, because I could see the thrashing but not where it was coming from.

Ive also noticed that the prelease build that was added in the last MSVSC fix ups, has also got really slow too, still a bit faster then the debug build is is horribly slow, but the prelease is certainly slowed down a lot since then, and sol when stardreaming (when it doesn't crash) is horrible slow

so to find this cause of the slowdown would help testing the debug and prelease builds a lot better, I do have a good system, and even that doesn't really help much, so on low end systems it would be horrible

Yeah I added the PreRelease version to get some speed back. I've used a similar system in my day job so that you're usually running at 90% of full speed but with most of the debug info.

It's pretty bad if that's slowing down too. Something must be destroying or reorganising some pretty massive amounts of data stored in a very inappropriate data structure every frame.

The performance comes back if you pause the game but even in an almost empty star system in the middle of space with no other ships, cities, planets nearby it still performs very badly, until paused when the framerate returns.

I've no problems running a debug build with the current code. But I've no compare to the previous version.

The game looks great but not the thing I'm looking for. 🙁

Then you should give AMD CodeAnalyst a try. No matter if you're using and AMD or Intel cpu.

The only "problems" are very poor performance which given some of the code I'm writing at the moment (in my "Vavatch" branch if you want to take a look) can bring the release build to it's knees, so what would be a 3-5 fps slideshow in Debug in the normal game becomes 15-seconds-per-frame when I have an Orbital 600,000kms across etc.

I've been giving CodeAnalyst a try, it's limited on my Intel machine to pretty crap profiling, but it's a little better on my laptop. I'm away during the week so not getting much time to try it again.

Should do once more this weekend.

What is it that you're looking for?