20 Jul
2010
20 Jul
'10
8:36 a.m.
On 15/07/2010 00:26, John Meacham wrote:
On Wed, Jul 14, 2010 at 10:35:50AM +0100, Simon Marlow wrote:
Ah, this paper looks very interesting, I was wondering if you had experimented with prefetching just ahead of the allocation pointer. Looks like it helped :)
It helped back then, but had to be finely tuned for each CPU. I think automatic prefetching these days gets most of the benefit without any of the tuning or portability issues. I experimented with other kinds of prefetching in the GC, but failed to find anything that worked reliably enough. Cheers, Simon