[libdispatch-dev] libdispatch object caches

[Dave Zarzycki]

On May 20, 2011, at 4:26 AM, Joakim Johansson wrote:

> - Would a magazine-based object cache be interesting for libdispatch?

malloc(3) on Mac OS X is already magazine based.

> - Have any analysis of typical TSD continuation cache sizes been done already? (anyone who have seen any issues here in GCD-heavy applications?)

One has to define the applications that will be evaluated first. In any case, most programs are extremely bursty in nature, so the caches are cleaned regularly in practice.

> - What would be a reasonable test for performance validation?

The "starfish test" was setup to, as much as possible, take advantage of thread and continuation cache recycling. It isn't a great test, and depending on tuning knobs it can spillover into swap files, but at least it is something.

> - Would a relying on libumem be acceptable for people in general? 

Hard to say.

davez
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.macosforge.org/pipermail/libdispatch-dev/attachments/20110524/85ba92b2/attachment.html>