[Click] [PATCH] e1000e-0.4.1.7 NAPI for click polling extension

[Joonwoo Park]

Hi,

I used Counter element for test.
Basically I have used 'rate' read handler of Counter element, I'm
pretty sure it's accurate.
The output of 'rate' handler was not various because the other linux
machine which was sending packet with pktgen was stable.
Moreover since mostly click received almost of sent packets, I could
compare pktgen report with 'count' handler of Counter element.

p.s I cc-ed click list.  Please cc list next time if you don't mind,
everyone on the list will be appreciated that.

Thanks,
Joonwoo

On Wed, Feb 11, 2009 at 11:08 PM, Kyoungho An <an2315 at gmail.com> wrote:
> Hi,
> I'd like to ask about how to measure the bps performance.
> Did you get a value from bps_rate of count handler which is in /proc?
> If so, how did you get the accurate value from fluctuated bit_rate?
> Thanks!
>
> On Sat, Feb 7, 2009 at 3:01 PM, Joonwoo Park <joonwpark81 at gmail.com> wrote:
>>
>> Hello,
>>
>> Theses patches are for click NAPI improvement for polling mechanism.
>> The main goal of this patch is doing polling with minimal processor
>> usage with using linux NAPI.
>> To use NAPI feature for PollDevice element, please use INTERRUPT=1 as
>> argument.
>> I am also attaching brief performance test result by linux pkggen
>> module and it seems pretty nice.
>>
>> Please consider to apply this patch Eddie.
>>
>> Thanks,
>> Joonwoo
>>
>> _______________________________________________
>> click mailing list
>> click at amsterdam.lcs.mit.edu
>> https://amsterdam.lcs.mit.edu/mailman/listinfo/click
>>
>
>
>
> --
> Kyoungho An
> School of Computer Science and Electronic Engineering
> Handong Global University
> Tel: +82)10-2048-9686
>
>
>