[knem-devel] The unexpected performance of DMA memcopy with knem
zhengda1936 at gmail.com
Tue Jul 30 13:35:19 CEST 2013
Thank you, Brice.
On Mon, Jul 29, 2013 at 11:11 AM, Brice Goglin <Brice.Goglin at inria.fr>wrote:
> > The way to fix this is to stop using the netdma channels and directly
> > acquire physical channels from KNEM.
> KNEM actually currently uses the "general purpose" channels, those are
> improperly distributed per cpu as explained above. One ugly way to solve
> the issue would be to manually migrate to another CPUs so that we get a
> channel with correct locality from the "general purpose" allocator. But
> that's ugly.
> There's another low-level acquire interface but it doesn't work for the
> channels we are considering here. So I don't think there's any clean
> solution for now. I'll ask the kernel guys and report the locality
> issue. We'll see what they say.
Could you keep me informed of the progress? I'm very interested in how the
kernel guys will fix the problem.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the knem-devel