not seeing anything 802.11v but hope theres some less stripping in this one..
Interesting. Perhaps more testing and fiddling with a 4.9 build in my future _________________ Netgear R7500v2, DD-WRT v3.0-r37845M kongat
Netgear R7000, DD-WRT v3.0-r37715M kongac
Thanks Kong for this great build..QoS it’s working great and in the latest builds I has problems with WiFi 5ghz disconnect with R9000 and EX8000 extender...this build it’s very stable, WiFi continuous to disconnect but after 1 or 2 days, that it’s more than acceptable (in the latest builds WiFi disconnects after 1 or 2 hours).
What are vht160 valid channels for upper than 128?
Joined: 03 Jan 2010 Posts: 7568 Location: YWG, Canada
Posted: Thu Nov 29, 2018 9:35 Post subject:
seems the qos in k4.9 is a fair amount.. different.. than 3.18..
websocket performance is utterly down the drain with htb/hfsc+pie, pie was best for me with htb, not so much anymore.. it still is best on delivering full what i enter, speeds, with a FLAT LINE, unlike fq_codel.. which loves to weave up and down by as much as 12mbps.
actually, all are bad with websocket connections unless its fq_codel, totally wasnt true in k33.18, hfsc+codel (not fq) is the best mixed bag of stable near flat line throughput of pie, with (somewhat) acceptable latency when using websockets, still quite elevated, but usable. but this is on ethernet only, over wifi, hfsc (with anything) remains to be abnormally slower even when no core is saturated.
this is for UPLOAD by the way.. download websocket connections seem to be fine. the one thats been free of issues all this time (except when limiting under 1mbps, thats still there)
*shrug*
edit: apparently its only with mobile devices like phones.. even laptop is fine. safari on ios 12.1 has poor uplink latency with websocket connections, only that so far, but on k3.18 pie that doesnt happen, everything is all over the place.. hfsc on smartphones has always given weaker throughput though than htb, on a laptop on wifi its same as htb and ethernet... _________________ LATEST FIRMWARE(S)
BrainSlayer wrote:
we just do it since we do not like any restrictions enforced by stupid cocaine snorting managers
seems the qos in k4.9 is a fair amount.. different.. than 3.18..
websocket performance is utterly down the drain with htb/hfsc+pie, pie was best for me with htb, not so much anymore.. it still is best on delivering full what i enter, speeds, with a FLAT LINE, unlike fq_codel.. which loves to weave up and down by as much as 12mbps.
I found performance of USB3 is higher on 4.9
Reading USB by SMB via wireless 5GHz gave me 30 MB/s on 3.18 while on 4.9 it is 40 (on voxel it was also 40)
And using FTP it's almost 50 MB/s
im close to giving up on https and turning it off, to silence that crap that no one wants to fix. plus https when viewed remotely makes webif slow as molasses, like when viewing ath0 wifi status, change it to ath1, takes an eternity to load..
I tried disabling HTTPS and it didn't fix the problem. Made it A LOT better for sure but didn't completely stop the errors. _________________ Routing:.......Asus RT-AX88U (Asuswrt-Merlin 384.14) Switching:....Netgear GS608_V3 & GS605_V4, TrendNet TEG-S82G & TEG-S50G
im close to giving up on https and turning it off, to silence that crap that no one wants to fix. plus https when viewed remotely makes webif slow as molasses, like when viewing ath0 wifi status, change it to ath1, takes an eternity to load..
I tried disabling HTTPS and it didn't fix the problem. Made it A LOT better for sure but didn't completely stop the errors.
FYI ... In my previous post where showed syslog erros.
Note: I have never had https enabled
Router Model: R9000
Status: Running
Reset: Yes
Errors: CPU0 High usage by KSOFTIRQD/0
Enabling QoS cause high utilization on CPU 0 by apparent process ksoftirqd/0 provoking slow/poor transmissions speed on WAN, WLAN and LAN. This is occurring since the last 2 or so previous versions.