* [Make-wifi-fast] in search of a plot
@ 2016-10-26 1:25 Dave Taht
2016-10-26 7:47 ` Michal Kazior
0 siblings, 1 reply; 5+ messages in thread
From: Dave Taht @ 2016-10-26 1:25 UTC (permalink / raw)
To: Michal Kazior, make-wifi-fast
At some point in the last 8 months, I had a plot (of 50? stations)
that showed before/after results with the then current ath10k driver,
and the new stuff, with this "ramping up" chaotic plot ultimately
showing 10s of seconds of latency, lots of jitter, flows stopping for
a long time, etc.
followed by another one, that did 50 stations "cleanly". I cannot for
the life of me find them, or remember who I got the data from, and I'd
like to use those for the linuxplumbers talk. Any memories? I did not
apparently blog those, so I figure it's in email somewhere...
--
Dave Täht
Let's go make home routers and wifi faster! With better software!
http://blog.cerowrt.org
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Make-wifi-fast] in search of a plot
2016-10-26 1:25 [Make-wifi-fast] in search of a plot Dave Taht
@ 2016-10-26 7:47 ` Michal Kazior
2016-10-26 16:05 ` Dave Taht
0 siblings, 1 reply; 5+ messages in thread
From: Michal Kazior @ 2016-10-26 7:47 UTC (permalink / raw)
To: Dave Taht; +Cc: make-wifi-fast
Hi Dave,
On 26 October 2016 at 03:25, Dave Taht <dave.taht@gmail.com> wrote:
> At some point in the last 8 months, I had a plot (of 50? stations)
> that showed before/after results with the then current ath10k driver,
> and the new stuff, with this "ramping up" chaotic plot ultimately
> showing 10s of seconds of latency, lots of jitter, flows stopping for
> a long time, etc.
>
> followed by another one, that did 50 stations "cleanly". I cannot for
> the life of me find them, or remember who I got the data from, and I'd
> like to use those for the linuxplumbers talk. Any memories? I did not
> apparently blog those, so I figure it's in email somewhere...
It was me. 100 stations. Latency sky-rocketing. Flows starved.
The "clean" case included my WIP ATF implementation for ath10k.
Unfortunately I never got to finish it up.
Funnily enough I dug the plot up yesterday for a different reason:
http://imgur.com/a/sSn8V
Michał
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Make-wifi-fast] in search of a plot
2016-10-26 7:47 ` Michal Kazior
@ 2016-10-26 16:05 ` Dave Taht
2016-10-26 18:20 ` Michal Kazior
0 siblings, 1 reply; 5+ messages in thread
From: Dave Taht @ 2016-10-26 16:05 UTC (permalink / raw)
To: Michal Kazior; +Cc: make-wifi-fast
On Wed, Oct 26, 2016 at 12:47 AM, Michal Kazior <michal.kazior@tieto.com> wrote:
> Hi Dave,
>
> On 26 October 2016 at 03:25, Dave Taht <dave.taht@gmail.com> wrote:
>> At some point in the last 8 months, I had a plot (of 50? stations)
>> that showed before/after results with the then current ath10k driver,
>> and the new stuff, with this "ramping up" chaotic plot ultimately
>> showing 10s of seconds of latency, lots of jitter, flows stopping for
>> a long time, etc.
>>
>> followed by another one, that did 50 stations "cleanly". I cannot for
>> the life of me find them, or remember who I got the data from, and I'd
>> like to use those for the linuxplumbers talk. Any memories? I did not
>> apparently blog those, so I figure it's in email somewhere...
>
> It was me. 100 stations. Latency sky-rocketing. Flows starved.
Thanks! boy that first plot was ugly!
> The "clean" case included my WIP ATF implementation for ath10k.
> Unfortunately I never got to finish it up.
Will you be at linuxplumbers?
> Funnily enough I dug the plot up yesterday for a different reason:
>
> http://imgur.com/a/sSn8V
>
>
> Michał
--
Dave Täht
Let's go make home routers and wifi faster! With better software!
http://blog.cerowrt.org
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Make-wifi-fast] in search of a plot
2016-10-26 16:05 ` Dave Taht
@ 2016-10-26 18:20 ` Michal Kazior
2016-10-26 18:25 ` Dave Taht
0 siblings, 1 reply; 5+ messages in thread
From: Michal Kazior @ 2016-10-26 18:20 UTC (permalink / raw)
To: Dave Taht; +Cc: make-wifi-fast
On 26 October 2016 at 18:05, Dave Taht <dave.taht@gmail.com> wrote:
> On Wed, Oct 26, 2016 at 12:47 AM, Michal Kazior <michal.kazior@tieto.com> wrote:
>> Hi Dave,
>>
>> On 26 October 2016 at 03:25, Dave Taht <dave.taht@gmail.com> wrote:
>>> At some point in the last 8 months, I had a plot (of 50? stations)
>>> that showed before/after results with the then current ath10k driver,
>>> and the new stuff, with this "ramping up" chaotic plot ultimately
>>> showing 10s of seconds of latency, lots of jitter, flows stopping for
>>> a long time, etc.
>>>
>>> followed by another one, that did 50 stations "cleanly". I cannot for
>>> the life of me find them, or remember who I got the data from, and I'd
>>> like to use those for the linuxplumbers talk. Any memories? I did not
>>> apparently blog those, so I figure it's in email somewhere...
>>
>> It was me. 100 stations. Latency sky-rocketing. Flows starved.
>
> Thanks! boy that first plot was ugly!
>
>> The "clean" case included my WIP ATF implementation for ath10k.
>> Unfortunately I never got to finish it up.
>
> Will you be at linuxplumbers?
Yes.
Michał
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Make-wifi-fast] in search of a plot
2016-10-26 18:20 ` Michal Kazior
@ 2016-10-26 18:25 ` Dave Taht
0 siblings, 0 replies; 5+ messages in thread
From: Dave Taht @ 2016-10-26 18:25 UTC (permalink / raw)
To: Michal Kazior; +Cc: make-wifi-fast
Awesome!
Will anyone else here be at plumbers?
I confess that I started seriously writing the make-wifi-fast preso a
few days ago, and, as usual, am struggling - my ability to draw
diagrams in particular is not the best, and wanted something that
would adaquately explain the flow through the completion handler,
among other things.
If anyone here would like to collaborate in making that preso good
between now and next week, please drop me an email.
I'd planned on giving a few bits of the preso (mostly the ones I
intend to cut) at this friday's vuc.me conference. I had some ranting
to get out of my system, first.
On Wed, Oct 26, 2016 at 11:20 AM, Michal Kazior <michal.kazior@tieto.com> wrote:
> On 26 October 2016 at 18:05, Dave Taht <dave.taht@gmail.com> wrote:
>> On Wed, Oct 26, 2016 at 12:47 AM, Michal Kazior <michal.kazior@tieto.com> wrote:
>>> Hi Dave,
>>>
>>> On 26 October 2016 at 03:25, Dave Taht <dave.taht@gmail.com> wrote:
>>>> At some point in the last 8 months, I had a plot (of 50? stations)
>>>> that showed before/after results with the then current ath10k driver,
>>>> and the new stuff, with this "ramping up" chaotic plot ultimately
>>>> showing 10s of seconds of latency, lots of jitter, flows stopping for
>>>> a long time, etc.
>>>>
>>>> followed by another one, that did 50 stations "cleanly". I cannot for
>>>> the life of me find them, or remember who I got the data from, and I'd
>>>> like to use those for the linuxplumbers talk. Any memories? I did not
>>>> apparently blog those, so I figure it's in email somewhere...
>>>
>>> It was me. 100 stations. Latency sky-rocketing. Flows starved.
>>
>> Thanks! boy that first plot was ugly!
>>
>>> The "clean" case included my WIP ATF implementation for ath10k.
>>> Unfortunately I never got to finish it up.
>>
>> Will you be at linuxplumbers?
>
> Yes.
>
>
> Michał
--
Dave Täht
Let's go make home routers and wifi faster! With better software!
http://blog.cerowrt.org
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2016-10-26 18:25 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-10-26 1:25 [Make-wifi-fast] in search of a plot Dave Taht
2016-10-26 7:47 ` Michal Kazior
2016-10-26 16:05 ` Dave Taht
2016-10-26 18:20 ` Michal Kazior
2016-10-26 18:25 ` Dave Taht
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox