[ovs-dev] [PATCH v6 0/6] DPCLS Subtable ISA Optimization

Van Haaren, Harry harry.van.haaren at intel.com
Mon Jul 6 11:57:39 UTC 2020


> -----Original Message-----
> From: William Tu <u9012063 at gmail.com>
> Sent: Sunday, July 5, 2020 3:06 PM
> To: Van Haaren, Harry <harry.van.haaren at intel.com>
> Cc: ovs-dev <ovs-dev at openvswitch.org>; Stokes, Ian <ian.stokes at intel.com>; Ilya
> Maximets <i.maximets at ovn.org>; Federico Iezzi <fiezzi at redhat.com>
> Subject: Re: [PATCH v6 0/6] DPCLS Subtable ISA Optimization
> 
> On Thu, Jul 2, 2020 at 10:42 AM Harry van Haaren
> <harry.van.haaren at intel.com> wrote:
> >
> > v6 work done:
> > - Fix as --64 unrecognized option
> > - Fix build issues with avx512 library changes
> > - Fix files left in build dir after distclean
> > - Fix CPU arch dependant RTE_CPUFLAG_ usage
> >
> > Thanks William & Ian for review & help on v5.
> > All known issues are fixed and working here.

<snip cover letter details>

> > Harry van Haaren (6):
> >   dpif-netdev: implement subtable lookup validation.
> >   dpif-netdev: add subtable lookup prio set command.
> >   dpif-netdev: add subtable-lookup-prio-get command.
> >   dpdk: enable cpu feature detection.
> >   dpif-lookup: add avx512 gather implementation.
> >   docs/dpdk/bridge: add datapath performance section.
> >
> 
> The v6 looks good to me. It would be great if there is another pair of
> eyes on this series.
> It's a little hard for me to understand the avx512 gather
> implementation (patch5),
> but the idea of using autovalidator and it passes our test cases give
> me more confidence
> that it is correct.
> 
> Acked-by: William Tu <u9012063 at gmail.com>

I'm happy to see the Autovalidator concept in practice working well,
it is likely a development/testing model to re-use in future for similar
types of work to ensure consistency and validity of SIMD code.

Thanks for your multiple reviews of the series over the last months.
-Harry


More information about the dev mailing list