From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from sonic307-1.consmr.mail.bf2.yahoo.com (sonic307-1.consmr.mail.bf2.yahoo.com [74.6.134.40]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id 6FAF83CB35 for ; Fri, 12 Jul 2019 04:37:30 -0400 (EDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=yahoo.com; s=s2048; t=1562920650; bh=xRrY/dgL1IZMcsJ4pk959aBa0Z0W951X71NSkTlHJwA=; h=Date:From:To:In-Reply-To:References:Subject:From:Subject; b=PUtFAuxhKzkZXgb2WRSqjNG9ahRp9qmKOETkLCgC+cNVWqzQb00oYAqo0N95T4P0HFXmE8A0DTBtyn/z+/KALgfRWgXSaAiHdMSP+e87eLKPVIcBGMnGdnk9knZY0eziVoL6rgddRTiCfFkS+tg2wJhSvkMGzBKdncupXwmBYmynYUrqEJlu0rjfF0unOa2Vjcx917BrrJ8NUbEHCHe5+voxcpje1zsF7bnSKTY3/DCnsQFTcYrMgDDUdLAZgRqyLC09CVB2uZQj8puCWYp1t6MzV6ru1xk/0O2/GOqM7yZYR8Ha+Q4yNem1CQ252Za/wAb6/YRmuLvFlqeEO5iS2A== X-YMail-OSG: Pt2eW44VM1kGAc3EnrZc.XCrhtZjsaH1RBXunY8Molh3qYJ9o4Da68bFpV2ZNVm w6.ejExioHu9fiiAx3Jev22wzjgNJAzwlsQg7aCz1SlCuAyttzTJoCJ2zQ3EtHchMiR1.EHrxygv 6f_kVB8UXZM8Fu1AR3sVAJ6KiA_CSJY45VDSFNP1NjaijFTT29WtXkMgXvj5qdt4fWTFqgWTPTMU MRlzxpMnulXAR2pogvr_GzTzl6kBs1W.w8bcywSxoa1dymzjf5k2khip40TdG8pt61n67lcMglQa l.hWnrIaxe5hGruPmjCKGCmi1ZghKk9g9ySwvkjlRrHmgrpTDWH2TgTCEMjDXxHEyO4cj797stoT TFOSAM4_WanKhLMX9SmP78gdG6ErMk4vh6U4uOEvQQy48x.nBiWqYGW4rZWxPn3FuAoYj1EumheT sv1hQPlfFkb2NAPApzjEIjV.PPG9a1rIj4jl8C.ySd3Ts0EnFBS0OJMVIe18yJoiQAnrFqYakWNf Gi5t_F.3EtzhuA1BLTpUE3gxzsCVN8moq6tYJ8wT_tvcQUugysXVBPePfOaR3joLtvfYB7Io_vbf IAO6ynV0LPLntY1C0Z0wCJAQiVkqVn4L9iZRb8VswbrSlj6SQdSkizbmqtUCTqWk8i5BToeyhqS6 jROZm27ObODXGLRkBGSkG33t_RsCzMPwBT74F13TGkzxDrb1Bfov9f4qxD.aVeUizOLQDS9KEIZo fNAEe46.08T8ml94YQvUfSTQ6cng1sryLHWwNTGLBvLjKvDFOSkxl7yy5oiQTFbJa3WltEbr1l2z 7XKW.eZwhdoNYqRFnoWDwDTwdVTj9tTbSwUYFWfvfq_e3poyF4TVE5dFpPIA8KGcgk34XksovZib gvtBUIJ60bvvSJLYh8PoXHQ1Wd715YjPOs3UUYBY293lPLylXxHrR6W_JAN03mTr1eSQjuUGQpIW biMsF5rLqRR44CGajiGlEzONbQq4XMzl23PvftVs5TStdCYq1yeUYkSBOAsVbgmurD_2XXyOIYF_ mGRStmaqSBSyiw0b.Hwrrj0B9WvHAV2mEKkMwXzEF3Xk_qq4HwlS_lpIcTcukYSaZvHEDJh3QjFI YvJAErGBoa7dqlxrQv_iyYHPhVyavF_ehhV3lyLOj62B4fcl6T3y8tXKVVWP1hxF_YHI4lyvwgbW 6oGgf4SrM6VvNjgbdApyGAproqYBFWpsoa.9SLg4SKZm969EXJotRxIgGrM9H_uy_fNg6tF4NhP0 EeGajZC8OC__q8elx8s8- Received: from sonic.gate.mail.ne1.yahoo.com by sonic307.consmr.mail.bf2.yahoo.com with HTTP; Fri, 12 Jul 2019 08:37:30 +0000 Date: Fri, 12 Jul 2019 08:37:24 +0000 (UTC) From: E Diable Si To: "cake@lists.bufferbloat.net" Message-ID: <1067804730.155701.1562920644963@mail.yahoo.com> In-Reply-To: References: <1406985555.48468.1562879701286.ref@mail.yahoo.com> <1406985555.48468.1562879701286@mail.yahoo.com> <560938987.28604.1562881745799@mail.yahoo.com> Subject: Fw: Split-gso vs No-Split-gso MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_155700_627641536.1562920644961" X-Mailer: WebService/1.1.13991 YMailNorrin Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/75.0.3770.100 Safari/537.36 X-Mailman-Approved-At: Fri, 12 Jul 2019 05:07:21 -0400 X-List-Received-Date: Fri, 12 Jul 2019 08:37:30 -0000 ------=_Part_155700_627641536.1562920644961 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Hi, i think Cake has some issues.=C2=A0 I've read multiple reports on CAKE SQM failing to do its job and massive bu= fferbloat.=C2=A0 They still can't find out whats the issue.=C2=A0 I used to have a WNDR3700v4 back in may 2019 and it worked fine, my brother= then wanted me to revert back to oem settings and after a month and half i= got a new router R7800 and for some reason it's failing to do the job with= 2 cores rather than 1 core and more RAM & higher CPU frequency.=C2=A0 Can you please take a look at it. 1. Also, i'm not sure whether its openwrt or cake thats responsible for lis= ting the link layer adaption but you guys say its 44.https://openwrt.org/do= cs/guide-user/network/traffic-shaping/sqm https://openwrt.org/docs/guide-user/network/traffic-shaping/sqm-details https://forum.openwrt.org/t/your-post-in-sqm-high-latency/40018/35 While on this site, it says to choose NONE if you have fiber.https://www.bu= fferbloat.net/projects/cerowrt/wiki/Setting_up_SQM_for_CeroWrt_310/ https://forum.openwrt.org/t/your-post-in-sqm-high-latency/40018/35 Articles on issues with Cake SQM:https://forum.openwrt.org/t/r7800-sqm-sett= ings-keep-causing-bufferbloat/39720=C2=A0<- This is my thread on my issue w= ith the R7800. I also have a private message with mindwolf but can't send i= t to anyone as no option for sharing. https://forum.openwrt.org/t/startup-initscripts-enabled-or-disabled-by-defa= ult-also-sqm-question/37364/7=C2=A0<- Older article with my Wndr3700v4 rout= er that was working great. https://forum.openwrt.org/t/sqm-high-latency/1087/93 https://forum.openwrt.org/t/sqm-has-no-effect/39825 https://forum.openwrt.org/t/sqm-no-change-in-dslreports-test/37037 3. Below is a email i sent to Dave Taht because i believed maybe he could h= elp me out with the split-gso setting, with it on by default in SQM even th= ough you can't change it yourself in the GUI for OpenWrt, it caused me hour= s of headaches trying to figure out which setting is causing me massive buf= ferbloat.With no-split-gso i was getting less ping/ms bufferbloat in dslrep= orts.=C2=A0I noticed with split-gso that with it enabled it felt a bit more= less latency but dslreports indicated my pipelines are being hammered with= traffic at 200-500ms.With no-split-gso i was getting maximum was 25ms. Ple= ase take a look at my thread. ----- Forwarded Message ----- From: Dave Taht To: E= Diable Si Sent: Thursday, July 11, 2019, 11:41:57 P= M PDTSubject: Re: Split-gso vs No-Split-gso cake mailing list On Thu, Jul 11, 2019 at 2:49 PM E Diable Si wrote: > > Didn't quite follow, if you meant mess with Cake settings, I've done that= . I went with stock settings for CAKE, layer_cake.qos or piece_of_cake.qos = and it did not perform well.=C2=A0Ihad issues with my bufferbloat tests rea= ch from around 100ms and stayed around 250-450ms and peaks to 1100ms.=C2=A0 I did everything imo i could, my tests were inconclusive until mindwolf gav= e me some tips and advice by going at it by CLI rather than SQM in GUI and = realized my issues has much to do with split-gso test.=C2=A0 If only i can send you a private message SHARED of what i sent to mindwolf,= i would love to see if you can detect something i've missed, split-gso sho= uld be better for latency. > Have you wondered that with some latency settings like split-gso, you act= ually can feel the less latency in your mouse movements?=C2=A0 It doesn't feel like your mouse skates are binding/sticking to your mouse p= ad.=C2=A0 I might redo my tests once more time with SQM in GUI and delete all command= s from local startup.=C2=A0 Some said my issues had to do with my R7800 router as it has problems with = the CPU freq dropping back to idle maybe causing the higher ping so i went = with performance governor and been testing for 2 weeks. > > On Thursday, July 11, 2019, 02:26:19 PM PDT, Dave Taht wrote: > > > sure, but interact with the cake list, not me? > > On Thu, Jul 11, 2019 at 2:15 PM E Diable Si wrote= : > > > > I've read your tests but i got something totally different. Could you t= ake a look? > > > > Is there any way i can send you the private messages i had with another= user on openwrt forums to take a look at the test i performed because i go= t tons of bufferbloat right after i enabled split-gso. With SQM in LuCI GUI= , it doesn't perform right for me. I keep getting massive ping spiking to 2= 00ms-1000ms. > > > > https://lists.bufferbloat.net/pipermail/cake/2018-July/004182.html > > > > > -- > > Dave T=C3=A4ht > CTO, TekLibre, LLC > http://www.teklibre.com > Tel: 1-831-205-9740 --=20 Dave T=C3=A4ht CTO, TekLibre, LLC http://www.teklibre.com Tel: 1-831-205-9740 =20 ------=_Part_155700_627641536.1562920644961 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable
Hi, i think Cake has some is= sues. 

I've read multiple reports on CAKE SQM faili= ng to do its job and massive bufferbloat. 

They stil= l can't find out whats the issue. 

I used to have = a WNDR3700v4 back in may 2019 and it worked fine, my brother then wanted me= to revert back to oem settings and after a month and half i got a new rout= er R7800 and for some reason it's failing to do the job with 2 cores rather= than 1 core and more RAM & higher CPU frequency. 

Can you please take a look at it.

1. Also, i'm not s= ure whether its openwrt or cake thats responsible for listing the link laye= r adaption but you guys say its 44.


While on this site, it says to choose NONE if yo= u have fiber.


Articles on issues wi= th Cake SQM:

<= br>
3. Below is a email i sent = to Dave Taht because i believed maybe he could help me out with the split-g= so setting, with it on by default in SQM even though you can't change it yo= urself in the GUI for OpenWrt, it caused me hours of headaches trying to fi= gure out which setting is causing me massive bufferbloat.
With no-split-gso i was getting less ping/ms buf= ferbloat in dslreports. 
I= noticed with split-gso that with it enabled it felt a bit more less latenc= y but dslreports indicated my pipelines are being hammered with traffic at = 200-500ms.
With no-split-gso i = was getting maximum was 25ms. Please take a look at my thread.



=20
----- Forwarded Message -----
From: Dave Taht <dave.taht@gmail.com>
To: E Diable Si <tehrslegend@yahoo.com>
Se= nt: Thursday, July 11, 2019, 11:41:57 PM PDT
Subject: = Re: Split-gso vs No-Split-gso

cake mailing list
<= br clear=3D"none">On Thu, Jul 11, 2019 at 2:49 PM E Diable Si <tehrslegend@yahoo.com> wrote:
>
> Didn't quite follow, if you meant mess with Cake settings, = I've done that. I went with stock settings for CAKE, layer_cake.qos or piec= e_of_cake.qos and it did not perform well. 
I
had issues = with my bufferbloat tests reach from around 100ms and stayed around 250-450= ms and peaks to 1100ms. 
<= br>
I did everything imo i coul= d, my tests were inconclusive until mindwolf gave me some tips and advice b= y going at it by CLI rather than SQM in GUI and realized my issues has much= to do with split-gso test. 

If only i can send you = a private message SHARED of what i sent to mindwolf, i would love to see if= you can detect something i've missed, split-gso should be better for laten= cy.

> Hav= e you wondered that with some latency settings like split-gso, you actually= can feel the less latency in your mouse movements? 

It doesn't feel like your mouse skates are binding/sticking to your mouse = pad. 

I might redo my tests once more time with SQ= M in GUI and delete all commands from local startup. 

Some said my issues had to do with my R7800 router as it has problems wit= h the CPU freq dropping back to idle maybe causing the higher ping so i wen= t with performance governor and been testing for 2 weeks.


>
> On Thursday, July 11, 2019,= 02:26:19 PM PDT, Dave Taht <dave.taht@gmail.com> w= rote:
>
>
> = sure, but interact with the cake list, not me?
>
> On Thu, Jul 11, 2019 at 2:15 PM E Diable Si <tehrslegend@yahoo.com> wrote:
> >> > I've read your tests but i got something totally = different. Could you take a look?
> >
> > Is there any way i can send you the private messages i had w= ith another user on openwrt forums to take a look at the test i performed b= ecause i got tons of bufferbloat right after i enabled split-gso. With SQM = in LuCI GUI, it doesn't perform right for me. I keep getting massive ping s= piking to 200ms-1000ms.
> >
> = > https://lists.bu= fferbloat.net/pipermail/cake/2018-July/004182.html
&g= t;
>
>
>
> --
>
> Dave T= =C3=A4ht
> CTO, TekLibre, LLC
> <= a shape=3D"rect" href=3D"http://www.teklibre.com" rel=3D"nofollow" target= =3D"_blank">http://www.teklibre.com
> Tel: 1-831-2= 05-9740




--

Dave T=C3=A4ht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-831-205-9740
------=_Part_155700_627641536.1562920644961--