[dpdk-dev] [PATCH 5/5] examples/l3fwd: add neon support for l3fwd

[Jianbo Liu]

On 5 May 2017 at 12:24, Sekhar, Ashwin <Ashwin.Sekhar at cavium.com> wrote:
> On Thu, 2017-05-04 at 16:42 +0800, Jianbo Liu wrote:
>> Hi Ashwin,
>>
>> On 3 May 2017 at 13:24, Jianbo Liu <jianbo.liu at linaro.org> wrote:
>> >
>> > Hi Ashwin,
>> >
>> > On 2 May 2017 at 19:47, Sekhar, Ashwin <Ashwin.Sekhar at cavium.com>
>> > wrote:
>> > >
>> > > Hi Jianbo,
>> > >
>> > > I tested your neon changes on thunderx. I am seeing a performance
>> > > regression of ~10% for LPM case and ~20% for EM case with your
>> > > changes.
>> > > Did you see improvement on any arm64 platform with these changes.
>> > > If
>> > > yes, how much was the improvement?
>> > Thanks for your reviewing and testing.
>> > For some reason, I have not done much with the performance testing.
>> > I'll send a new version later after tuning the performance.
>> >
>> Can you tell me how did you test?
> Built with following commands.
> make config T=arm64-thunderx-linuxapp-gcc
> make -j32
>
> Tested LPM with
> sudo ./examples/l3fwd/build/l3fwd -l 9,10  --master-lcore 9  -- -p 0x1 --config="(0,0,10)"
>
> Tested EM with
> sudo ./examples/l3fwd/build/l3fwd -l 9,10  --master-lcore 9  -- -p 0x1 --config="(0,0,10)" -E
>

Only one port? What's the network topology, and lpm/em rules? How did
you stress traffic...?

>> My testing shows that EM case is much better, while LPM is almost the
>> same as before.
> Could you please tell on which arm64 processor/platform you tested.
> Also how much was the percentage increase in performance for EM ?
>

I'm sorry I can't tell you what's arm64 platform I tested on. But I
can get a ThunderX, and replicate your testing environment if you can
tell me more...

Thanks!
Jianbo