From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp97.iad3a.emailsrvr.com (smtp97.iad3a.emailsrvr.com [173.203.187.97]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id 3DBF93CB39 for ; Sun, 15 Oct 2023 16:29:53 -0400 (EDT) Received: from app68.wa-webapps.iad3a (relay-webapps.rsapps.net [172.27.255.140]) by smtp13.relay.iad3a.emailsrvr.com (SMTP Server) with ESMTP id 6B4941928; Sun, 15 Oct 2023 16:29:52 -0400 (EDT) Received: from deepplum.com (localhost.localdomain [127.0.0.1]) by app68.wa-webapps.iad3a (Postfix) with ESMTP id 50B61E0154; Sun, 15 Oct 2023 16:29:52 -0400 (EDT) Received: by apps.rackspace.com (Authenticated sender: dpreed@deepplum.com, from: dpreed@deepplum.com) with HTTP; Sun, 15 Oct 2023 16:29:52 -0400 (EDT) X-Auth-ID: dpreed@deepplum.com Date: Sun, 15 Oct 2023 16:29:52 -0400 (EDT) From: "David P. Reed" To: "dave seddon" Cc: "Sebastian Moeller" , "dave seddon via Cake" MIME-Version: 1.0 Content-Type: text/plain;charset=UTF-8 Content-Transfer-Encoding: quoted-printable Importance: Normal X-Priority: 3 (Normal) X-Type: plain In-Reply-To: References: <1695068690.78066946@apps.rackspace.com> <6D098714-69DC-41D9-A7DC-E94FD9C77625@gmail.com> <79EC8CB2-0912-4D66-9FA3-990E9C38C281@gmail.com> X-Client-IP: 209.6.168.128 Message-ID: <1697401792.32573388@apps.rackspace.com> X-Mailer: webmail/19.0.24-RC X-Classification-ID: e5ada36d-5612-4526-9a5d-01c07f380be5-1-1 Subject: Re: [Cake] some comprehensive arm64 w/cake results X-BeenThere: cake@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: Cake - FQ_codel the next generation List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 15 Oct 2023 20:29:53 -0000 60K pings per second? Well, that's probably fast enough for Cake work..., b= ut I'm sure you can do a LOT better... Try AF_XDP and/or DPDK. I think AF_X= DP works on ARM.=0A=0Ahttps://adarsh-kumar-phe15.medium.com/receiving-14-mi= llion-network-packets-per-second-a9d5cc1408b6=0A=0ANow admittedly, my datac= enter networking work is focused on single-digit microsecond RTT protocols = (and doing them in Linux). But we really need to get clear on what "fast" m= eans in modern computers. It does bug me that the folks working on network = performance are still focused as if networks have to be slow.=0A=0AThe same= problems can be found throughout the Linux kernel, where assumptions still= seem to be holding on to what worked in Linus's original PC (single core, = slow rotating disks, ...). I recently found a comment (in the disk page swa= p code) where the idea that disk cylinders are fast, but seeks are slow, is= implicit in the design. It's the world's most "modern" "legacy system" - w= ith lots of design decisions that make no sense anymore, but can't be undon= e.=0A=0AOf course, Internet congestion control, in general, is still stuck = in the original Van Jacobsen sawtooth era. My guess is it won't get fixed, = though I applaud Cake, and despair the hardware folks who keep adding buffe= rs.=0A=0A=0AOn Sunday, October 15, 2023 12:24pm, "dave seddon via Cake" said:=0A=0A> ____________________________________= ___________=0A> Cake mailing list=0A> Cake@lists.bufferbloat.net=0A> https:= //lists.bufferbloat.net/listinfo/cake=0A> Oh thanks Sebastian. I have irtt= installed, but it looks like I need to=0A> start the server. That's easy= . Doing it now.=0A> =0A> ( Incidentally, I did write a golang based icmp p= inger. It can ping at=0A> very high rates: https://github.com/edgio/icmpen= gine. Really should write=0A> one with rust using io_uring... )=0A> =0A> = =0A> =0A> On Sun, Oct 15, 2023 at 8:53=E2=80=AFAM Sebastian Moeller wrote:=0A> =0A>> If I recall correctly, flent will use irtt for = its delay probes if=0A>> available on both ends. Sure fixing fping seems li= ke a good thing longer=0A>> term, but to get data in quickly, maybe try irt= t instead?=0A>>=0A>>=0A>> On 15 October 2023 17:11:23 CEST, dave seddon via= Cake <=0A>> cake@lists.bufferbloat.net> wrote:=0A>>=0A>>> G'day,=0A>>>=0A>= >> I've put more work into a test framework around the qdisc tests, but=0A>= >> unfortunately flent doesn't work easily with Ubuntu LTS (=0A>>> https://= github.com/tohojo/flent/issues/232, which I think is an issue=0A>>> with fl= ent parsing the fping output ).=0A>>>=0A>>> Results and graphs in this shee= t:=0A>>>=0A>>> https://docs.google.com/spreadsheets/d/1T59QwEdNwJFm4TgDFA_N= Y98gicOm8ABXKvDsSIMz9ag/edit#gid=3D1203641125=0A>>>=0A>>> Raw results of x2= test runs are here:=0A>>>=0A>>> https://github.com/randomizedcoder/qdisc_r= esults/blob/main/qdisc/report.csv=0A>>>=0A>>> Each run:=0A>>>=0A>>> https:/= /github.com/randomizedcoder/qdisc_results/blob/main/qdisc/2023-10-13T18%3A4= 5%3A45/report.csv=0A>>>=0A>>> https://github.com/randomizedcoder/qdisc_resu= lts/blob/main/qdisc/2023-10-14T14%3A22%3A53/report.csv=0A>>>=0A>>> Full ipe= rf outputs are available too, for example:=0A>>> https://github.com/randomi= zedcoder/qdisc_results/blob/main/qdisc/2023-10-13T18%3A45%3A45/nanopi-r2s/f= q_codel/iperf/test/16_iperf/stdout=0A>>>=0A>>>=0A>>> Logs for each run are = also available, for example:=0A>>> https://github.com/randomizedcoder/qdisc= _results/blob/main/qdisc/2023-10-13T18%3A45%3A45/log.json=0A>>>=0A>>> The c= ode repo updated here: https://github.com/randomizedcoder/cake ,=0A>>> with= thehttps://github.com/randomizedcoder/cake/blob/main/README.md=0A>>> which= explains how the test work.=0A>>> Updated google doc is started here:=0A>>= > https://docs.google.com/document/d/1fYKj3BS89aB9drg_DsSq289xSdVQhn1zUJYCj= 0WuCs0/edit?usp=3Dsharing=0A>>>=0A>>> Based on the questions on this list e= arlier, there is a folder with=0A>>> device information for each of the dev= ices=0A>>> https://github.com/randomizedcoder/cake/tree/main/device_info=0A= >>>=0A>>> For example, the Pi4 and the Lichee Pi (risc-v) hardware layout i= s here:=0A>>> -=0A>>> https://github.com/randomizedcoder/cake/blob/main/dev= ice_info/pi4/hwloc-ls-pi4.png=0A>>>=0A>>> -=0A>>> https://github.com/random= izedcoder/cake/blob/main/device_info/lpi4a/hwloc-ls-lpi4a.png=0A>>>=0A>>> T= he switch has also been upgraded to a Cisco 3750x, which I think based=0A>>= > on the "show interface" output has a max queue size of 40 frames. The te= st=0A>>> process clears the counters before each test and gathers the "show= =0A>>> interface" output at the end.=0A>>>=0A>>> The Lichee Pi 4A doesn't l= ook good (=0A>>> https://wiki.sipeed.com/hardware/en/lichee/th1520/lp4a.htm= l )=0A>>>=0A>>> [image: image.png]=0A>>> I really wish the flent was workin= g, so I'll probably see if I can work=0A>>> out the parsing.=0A>>>=0A>>> Th= anks,=0A>>> Dave Seddon=0A>>>=0A>>> On Fri, Oct 13, 2023 at 10:25=E2=80=AFA= M dave seddon =0A>>> wrote:=0A>>>=0A>>>> My bad. = There's a bug for this.... Looks like I have to downgrade fping=0A>>>>=0A>= >>> https://github.com/tohojo/flent/issues/232=0A>>>> https://github.com/sc= hweikert/fping/issues/203=0A>>>>=0A>>>> On Fri, Oct 13, 2023 at 8:59=E2=80= =AFAM dave seddon =0A>>>> wrote:=0A>>>>=0A>>>>> G= 'day,=0A>>>>>=0A>>>>> I've been working away on automation of the tests. P= retty close to=0A>>>>> having much nicer tests with a lot more details. I'= ve also got the risc-v=0A>>>>> device working.=0A>>>>>=0A>>>>> However, I'v= e run into something funny with flent. Flent is not happy=0A>>>>> with fpi= ng or ping.=0A>>>>>=0A>>>>> das@3rd:~/Downloads/cake/cmd/run_qdiscs_tests$ = /usr/bin/sudo=0A>>>>> /usr/sbin/ip netns exec network101 /usr/bin/flent rru= l --output=0A>>>>> /tmp/qdisc/2023-10-13T15:53:21/pi4/noqueue/flent/test/1= 5_flent/flent_pi4_noqueue.png=0A>>>>> --data-dir /tmp/qdisc/2023-10-13T15:5= 3:21/pi4/noqueue/flent/test/15_flent/=0A>>>>> --format summary --plot all_s= caled --title-extra=0A>>>>> 2023-10-13T15:53:21_pi4_noqueue --note 2023-10-= 13T15:53:21_pi4_noqueue=0A>>>>> --extended-metadata --host 172.17.51.10 --l= ength 60 --ipv4 --socket-stats=0A>>>>> Starting Flent 2.0.1 using Python 3.= 10.12.=0A>>>>> Starting rrul test. Expected run time: 70 seconds.=0A>>>>> W= ARNING: Found fping, but couldn't parse its output. Not=0A>>>>> using. = <---------------- ???=0A>>>>> ERROR: Runner Ping (ms) ICMP failed = check: Cannot parse output of the=0A>>>>> system ping binary (/usr/bin/ping= ). Please install fping v3.5+. <----- ??=0A>>>>>=0A>>>>> das@3rd:~/Downl= oads/cake/cmd/run_qdiscs_tests$ dpkg --list | grep ping=0A>>>>> ii fping = 5.1-1=0A>>>>> amd64 sends I= CMP ECHO_REQUEST packets to network hosts=0A>>>>> ii iputils-ping = 3:20211215-1=0A>>>>> amd64 Tools to test= the reachability of network hosts=0A>>>>> ii kpartx = 0.8.8-1ubuntu1.22.04.1=0A>>>>> amd64 create dev= ice mappings for partitions=0A>>>>> ii libharfbuzz0b:amd64 = 2.7.4-1ubuntu3.1=0A>>>>> amd64 OpenType text shaping e= ngine (shared library)=0A>>>>> das@3rd:~/Downloads/cake/cmd/run_qdiscs_test= s$ fping --version=0A>>>>> fping: Version 5.1=0A>>>>> das@3rd:~/Downloads/c= ake/cmd/run_qdiscs_tests$ ping -V=0A>>>>> ping from iputils 20211215=0A>>>>= >=0A>>>>> das@3rd:~/Downloads/cake/cmd/run_qdiscs_tests$ cat /etc/lsb-relea= se=0A>>>>> DISTRIB_ID=3DUbuntu=0A>>>>> DISTRIB_RELEASE=3D22.04=0A>>>>> DIST= RIB_CODENAME=3Djammy=0A>>>>> DISTRIB_DESCRIPTION=3D"Ubuntu 22.04.3 LTS"=0A>= >>>>=0A>>>>> I did install via "apt install fping"=0A>>>>>=0A>>>>> Any thou= ghts please?=0A>>>>>=0A>>>>> Kind regards,=0A>>>>> Dave=0A>>>>>=0A>>>>> On = Thu, Sep 28, 2023 at 6:27=E2=80=AFAM Sebastian Moeller via Cake <=0A>>>>> c= ake@lists.bufferbloat.net> wrote:=0A>>>>>=0A>>>>>>=0A>>>>>>=0A>>>>>> > On S= ep 28, 2023, at 15:19, David Lang wrote:=0A>>>>>> >=0A>>>>>= > > On Thu, 28 Sep 2023, Sebastian Moeller via Cake wrote:=0A>>>>>> >=0A>>>= >>> >> P.S.: I am tempted, but will likely wait until they are available=0A= >>>>>> in quantity and hope that the street price comes down a bit before g= etting=0A>>>>>> one ;)=0A>>>>>> >=0A>>>>>> > They aren't available at all y= et, and it's not clear when they will=0A>>>>>> be available.=0A>>>>>>=0A>>>= >>> The announcement was end of October, but I think I could=0A>>>>= >> pre-order right now if I was feeling an urge. You are right though,=0A>>= >>>> announced !=3D available or delivered.=0A>>>>>>=0A>>>>>> Regards=0A>>>= >>> Sebastian=0A>>>>>>=0A>>>>>> P.S.: I have a pi400 in use as "des= ktop" for my oldest kid, this is=0A>>>>>> close to be actually generally us= able, I would guess that changing a=0A>>>>>> potential p500 from the pi400'= s 4GB to 8 GB together with the other=0A>>>>>> imprivements the 5 brings mi= ght push it over the threshold into the truly=0A>>>>>> useful category. Whi= ch probably means that either a potential pi500 will=0A>>>>>> come late and= probably with only 4 GB, but let's see how this works out now=0A>>>>>> tha= t the supply situation is less problematic.=0A>>>>>> And I understand that = there are other capable ARM based SoCs for=0A>>>>>> homerouter/desktop duty= , I just happen ot have a soft spot for the=0A>>>>>> raspberry project ;)= =0A>>>>>>=0A>>>>>> >=0A>>>>>> > David Lang=0A>>>>>>=0A>>>>>> ______________= _________________________________=0A>>>>>> Cake mailing list=0A>>>>>> Cake@= lists.bufferbloat.net=0A>>>>>> https://lists.bufferbloat.net/listinfo/cake= =0A>>>>>>=0A>>>>>=0A>>>>>=0A>>>>> --=0A>>>>> Regards,=0A>>>>> Dave Seddon= =0A>>>>> +1 415 857 5102=0A>>>>>=0A>>>>=0A>>>>=0A>>>> --=0A>>>> Regards,=0A= >>>> Dave Seddon=0A>>>> +1 415 857 5102=0A>>>>=0A>>>=0A>>> --=0A>> Sent fro= m my Android device with K-9 Mail. Please excuse my brevity.=0A>>=0A> =0A> = =0A> --=0A> Regards,=0A> Dave Seddon=0A> +1 415 857 5102=0A> =0A