Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I think it's still easier to ensure performant code as reference counting is more deterministic than GC. I assume you can turn off ARC for compilation unit if you want, so performance critical code can be manually managed.


Naieve reference counting incurs a bus lock across all processors on every assignment, more or less. That's quite painful.


95% of code can run an order of magnitude slower than "optimal" with no discernable degradation of the user experience.

If a bus lock degrades your performance too much, either change your algorithm, or isolate that section of code and do it in another language that doesn't suffer the same performance issues.


How many processors (cores) do most iOS devices have?

GC is still supported on MacOS X.


One core, but the Apple A5 in the iPad 2 and iPhone 4GS/5 (the flagship devices for iOS 5) has two cores.


Also autorelease only applies to NS objects, not plain C data.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: