Thanks, Bennie. These are good comments, and of course I hadn't thought
about the hardware age. But there is one thing you said that I don't agree
with:

On Wed, Oct 16, 2013 at 5:48 AM, Bennie Kloosteman <[email protected]>wrote:

> - it did exceed the adjustable time cap due to nursery collection . There
> are way better Nursery sweep techniques for pauses since then  as eg by
> modern cocurent GC techniques or evern simple using more memory and
> swapping nuseries  , sweeping the nursery in the background and using a
> barrier while it is active.
>

I understand the proposal above, and I don't buy it. If we want to let the
mutator run concurrently, the key words here are "...using a barrier...".
If objects are in the process of being forwarded, we need to be clear that
this is a READ barrier. And in contrast to the block clearing behavior on
RC-immix, this is a read barrier on very hot objects.

I'm going to resume this on the object forwarding thread, so that the
discussion about forwarding stays in one place.


Jonathan
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to