On Mon, Jul 15, 2013 at 8:48 PM, Jonathan S. Shapiro <[email protected]>wrote:

> My guess is that the C4 collector will outperform malloc/free in your
> example, with *no* performance degradation in the mutator.
>

I 100% agree, for workloads with spare CPU. CPU bound workloads, no.


> Setting that aside, kicking straw dogs is not a good use of time. Unless
> there are meaningful programs that actually exhibit this behavior, it's an
> exercise in intellectual masturbation. One can, of course, create
> pathological cases for *any* allocator, including fully manual storage
> management.
>

There are tons of meaningful programs that exhibit this behavior. They are
currently written in C/C++. I think computer-games make a good case study,
specifically because they *are* using GC (often Lua, sometimes Mono), they
just can't use it for the majority of data -- which is kept in C++. One
also doesn't have to throw a stone very far to find a JVM webserver which
has had to go on a heap-diet.

With respect and humor, kicking straw dogs seems a saner approach than
legistlation to ban use of tools that meet there needs without proof the
alternatives are sufficient.

The onus is not on C/C++ users to prove it can be done. The onus is on us
(proponents of GC). So far GC has done well taking over less latency
centric tasks with constrained heaps. No so well on large heaps or latency
centric software.

I'm confident Azul-like kernel changes will *eventually* make it into
kernels. I guess I just think our pressure is better applied here, where it
increases GC's ability to achieve our goals, than on advocating the end of
software systems which pausing-collectors have no hope of rivaling.

@Bennie - The C4 kernel changes are not about allocations, but bulk MMU
page protection changes. They use MMU page-protection for a hardware
assisted concurrent write-barrier, and to do this they have to change tons
of MMU page protection bits. The standard kernel interfaces don't make this
efficient enough.  (or at least that is my understanding/memory of it, the
paper tells the real details)
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to