Bryce Kampjes wrote:
> I get 22% of the time inside the GC with 369ms with the 0 at 0s and 101ms
> with the 0 at 0 commented out so on my machine with my VM and current
> compile flags so 30% of the object creation overhead is GC and 70% is
> allocation. That's running interpreted code. So using my measurements,
> I could get realistically get a two to three times gain in compiled
> code.
> 
> I'm measuring GC time with TimeProfileBrowser. Unless you're measuring
> with a more accurate tool then I guess platform matters for relative
> times.

Just one question: What's the timer accuracy on your platform? On 
Windows (where I took my measures) I'm guaranteeing an effective 1ms 
resolution but I'm not sure you get this on Linux (there have been 
issues with that in the past because, stupidly, many kernels are 
compiled with timer resolutions of >20ms). And timer resolution matters 
a lot for these benchmarks.

> Modifying the GC to use a copying collector for new space will speed
> up GC performance noticably.

Now *that* would be a thing that I'd like to see... I would expect that 
to have far more of a real-world impact.

> Thanks for the interesting emails. And yes, this is enough theory
> without proper profiling.

You're welcome!

Cheers,
   - Andreas