A first hand look from the .NET engineering teams
Helge Mahrt has published an interesting blog post on his experience using .NET Native on his Windows Store app, 499+1px. The post includes a first-level performance analysis.
Check out Helge's post!
@check this out: Thanks for the link. I don't know that this is immediately actionable by the .NET team. I agree with many of the comments on this Stack Overflow thread that point out that this is a single, embarrassingly parallel test case that parallelizes better on Java than .NET. I'm sure there are many cases that perform better on Java and many cases that perform better on .NET. This only implies to me that each runtime makes different decisions. What matters is real-world code performance, scalability and robustness.
@Andrew: That code (C# version) represents a real world application issue that I had and since I could not show the whole source code of the application, I created that sample code.
I would suggest to Microsoft .Net / C# team to focus a bit more on real world applications performance rather than just adding more and more syntactic sugar to the language.
BTW that .NET Native stuff are irrelevant to most of the real world applications.
@check this out: Thank you for the suggestion. The .NET team focuses on many real world scenarios, including those addressed by .NET Native. You might check out our new 64-bit JIT (RyuJIT) or the announcements we made at Tech Ed regarding the Katana/OWIN project.
I maintain that a single example is not indicative of .NET performance vs. Java performance as a whole. I'm sorry that your particular sample doesn't scale as well on .NET as it does on Java. I've passed it on to a performance dev here to investigate and we'll see if we can't improve.
By the way, most of today's technology is irrelevant to the real world. The real world existed for millennia and did quite well before computers came around.