[ovs-dev] [PATCH v3 0/7] DPCLS Subtable ISA Optimization
Van Haaren, Harry
harry.van.haaren at intel.com
Tue Jun 16 17:26:10 UTC 2020
> -----Original Message-----
> From: William Tu <u9012063 at gmail.com>
> Sent: Tuesday, June 16, 2020 4:50 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 v3 0/7] DPCLS Subtable ISA Optimization
> On Wed, Jun 10, 2020 at 3:47 AM Harry van Haaren
> <harry.van.haaren at intel.com> wrote:
> > v3 Changes Summary:
> > - Added new "subtable lookup get" command for ease of use
> > - Changed set command to include "prio" aligning with other commands
> > - Improved output of "subtable lookup prio set" command
> > - Added documentation
> > - Minor code cleanups, #defines for magic numbers, typos etc
> > - Implement fix for hash-mismatch issue (reported by William Tu)
Thanks for the review & comments!
I've been fixing some of the changes you mentioned, looking to fix issues like
the robot noted on -mavx512f CC flag not supported etc.
I'll try get the v4 up this week, I believe it addresses most/all open issues,
and of course I'll incorporate your feedback on list in this thread.
> I benchmark v3 with EMC disable.
> Similar to the conclusion from v2, overall with AVX512 lookup enabled,
> the overall performance is slower, but the miniflow_lookup is faster.
Thanks for re-benchmarking, indeed no performance changes are expected
in the v3 or (next) v4 versions. Intention is to investigate other potential areas
of OVS for AVX512 optimization in future patchsets.
Thanks again for the detailed review, -Harry
More information about the dev