[dpdk-dev] [PATCH v2] net/memif: optimized with one-way barrier

[David Marchand]

On Wed, Oct 9, 2019 at 1:15 PM Jakub Grajciar -X (jgrajcia - PANTHEON
TECHNOLOGIES at Cisco) <jgrajcia at cisco.com> wrote:
>
>
>
> > -----Original Message-----
> > From: Phil Yang <phil.yang at arm.com>
> > Sent: Wednesday, October 9, 2019 4:05 AM
> > To: Jakub Grajciar -X (jgrajcia - PANTHEON TECHNOLOGIES at Cisco)
> > <jgrajcia at cisco.com>; ferruh.yigit at intel.com; dev at dpdk.org
> > Cc: thomas at monjalon.net; Damjan Marion (damarion)
> > <damarion at cisco.com>; Honnappa.Nagarahalli at arm.com;
> > gavin.hu at arm.com; nd at arm.com
> > Subject: [PATCH v2] net/memif: optimized with one-way barrier
> >
> > Using 'rte_mb' to synchronize the shared ring head/tail between producer and
> > consumer will stall the pipeline and damage performance on the weak
> > memory model platform such like AArch64. Meanwhile update the shared
> > ring head and tail are observable and ordered between CPUs on IA.
> >
> > Optimized this full barrier with the one-way barrier can improve the
> > throughput. On AArch64 n1sdp server this patch make testpmd throughput
> > boost 2.1%. On Intel E5-2640, testpmd got 3.98% performance gain.
> >
> > Signed-off-by: Phil Yang <phil.yang at arm.com>
> > Reviewed-by: Gavin Hu <gavin.hu at arm.com>
> > ---
> > v2:
> > Upgrade 'MEMIF_VERSION_MAJOR' number to 3. (Jakub Grajciar)
> >
> > v1:
> > Initial version.
>
> I jumped the gun with the version bump. The change doesn't break compatibility. I'm putting reviewed label on v1.
>
> Sorry for the inconvenience.

Ok, marking v2 as rejected then.


-- 
David Marchand