I remember seeing a talk by Cliff Click where he explains how a small change to the CPU can greatly assist with developing pauseless GCs (though his former employer, Azul Systems, has moved on to implement the whole thing in software, probably at some cost to performance).
Regardless of the undertaking, the question remains -- as we move into a many-core era -- whether we'll find a programming paradigm good enough to take advantage of many cores. I think that whatever the winning paradigm may be, it will require a "really good" GC.
Azul's GC is basically a "really good" GC. It can handle hundreds of CPUs with hundreds of GBs of heap with tens of GBs per second of allocation rate with pause times of 10-20 milliseconds. It's basically a solved problem with the proper overall system design.
At the moment they can run on x86_64 but its not as fast as it could be. The have a special build for every linux distribution.
I am not sure how good this all works if you have very few cores and not a lot of threads. All in all however the C4 defently shows the way of the future. Hardware developer should learn from Azul, the OS guys should work more on supporting managed runtimes and the programming language community should do so to. Managed Memory is the future exept in very, very special cases.
To give the programmer this flexibility, Go must support what we call interior pointers to objects allocated in the heap. The X.buf field in the example above lives within the struct but it is legal to capture the address of this inner field, for instance to pass it to an I/O routine. In Java, as in many garbage-collected languages, it is not possible to construct an interior pointer like this, but in Go it is idiomatic. This design point affects which collection algorithms can be used, and may make them more difficult, but after careful thought we decided that it was necessary to allow interior pointers because of the benefits to the programmer and the ability to reduce pressure on the (perhaps harder to implement) collector. So far, our experience comparing similar Go and Java programs shows that use of interior pointers can have a significant effect on total arena size, latency, and collection times.
Interior pointers may pose a problem for sophisticated GCs. Interestingly, this may not be a deal-breaker (see here: https://groups.google.com/forum/#!topic/golang-dev/GvA0DaCI2...), but because Go relies on interior pointers so much, and because the language does not distinguish between interior pointers and regular object pointers, this may prove to be a huge problem in practice.
Rust's type system can.
I really like Rust, but I wonder how it is going to get over this hurdle that previous languages like Cyclone never fully solved.
I would need to have read Andrei's original argument to be sure, but I'm guessing it was along the lines of "without a strong type system, garbage collection is necessary to prevent dangling pointers". And Rust's type system is strong enough to prevent dangling pointers; they cannot happen unless you are deliberately dropping into `unsafe` blocks. This is true regardless of if you have garbage collection built in. But that doesn't necessarily mean that the type system is powerful enough to encode all the patterns that GC allows.
To see how hard the implementation of a parallel and concurrent garbage collector can be, take a look at the Garbage-First Garbage Collection paper. Try to really understand all the moving parts in there and how they interact. Once you're done wiping your brains off the wall, realize that this is the extent that GC engineers need to go to remove most bottlenecks on scaling, regardless of your opinion of G1 itself. This is a serious and hard-core engineering effort that absorbs several man years of expert effort (and I emphasize the expert part) to do properly. Rust would probably have to become the size and importance of the Java world to ever get this level of attention. Despite this, the JVM's collectors still routinely throw roadblocks in front of the mutator.
And that ignores the other tradeoffs that GCs present. E.g. most GCs need at least 3x working set to work reasonably efficiently. That's 3x less working set you could be keeping in a local memory bank, or using for something else (like a disk cache, because page faults sure are crazy expensive...).
The best way to avoid this problem is to not play the game at all. The reason I became interested in Rust in the first place is Graydon and co wisely did not follow that pied piper. Optional thread-local garbage collectors are a much simpler problem.
As an unrelated aside, I recall Andrei Alexandrescu making the argument that GC is important for type safety, unless the type system can prove that dangling pointers are impossible.