Joined: 18 Mar 2014 Posts: 12839 Location: Netherlands
Posted: Wed Mar 10, 2021 15:55 Post subject:
The latency spikes were due to the MIB counter I think and are resolved, some performance problems were due to the low netdev_max_backlog settings which is also resolved.
I use both OpenWRT and DDWRT for my R7800 and there is not much difference (also ran stock in the past which was marginally faster probably due to the help of the nss cores for my R7800)
But for the R9000 things could be different although there are a lot of similarities like the same QCA9984 wifi .
Looks like this has been a known issue for this router. This is sounding damn similar.
I'm about to try this older DD-wrt from
/betas/2019/01-29-2019-r38507/netgear-r9000/
Let's see what that does.
I know there is a lot of talk about BS (Brainslayers) builds.. but also know he left a while ago.. and all builds were taken down. Did he go elsewhere with the builds.. or just stopped?
Well, that's actually nonsense.
The thread is 2 years old and it was also a question of settings. (I'm reading something about a wrong MTU setting)
The R9000 has the QCA9984 as WLAN chipset, which is the same chipset I use and with which I test the builds...
so this is certainly a question of the setting and of course a question of the WLAN client used.
Just tell us with what kind of client you are testing...
As said 250Mbit over 2.4Ghz manage the least clients (most have only 1-2 antennas).
Quote:
my machine is a XPS 15
The WLAN card would actually be more important
Killer 1650 / Intel AX200 ?
Edit: best you also post a screenshot of the tab "Status > WLAN"
there you can see the connected clients too
I would also test directly with iperf3 between the router and the client, 150Mbit over 5Ghz are already a bit meager at short distance.
I have with my smartphone ~300Mbit in the bathroom through 3 walls
Quote:
while next to the R9000
you don't really need to test at a distance of less than 4m... that's all too close.
The devices just scream at each other
At 1m distance you can easily go down to 13dB (but I have not yet tested whether then the transmission power is still too high)
In any case, this is not an advantage when you test like this - there are definitely problems with too high transmitting power.
Just tell us with what kind of client you are testing...
As said 250Mbit over 2.4Ghz manage the least clients (most have only 1-2 antennas).
The WLAN card would actually be more important
Killer 1650 / Intel AX200 ?
Its a Killer 1650 in the XPS and Intel Wireless AC 9260 in one of the Precision ones
ho1Aetoo wrote:
I would also test directly with iperf3 between the router and the client, 150Mbit over 5Ghz are already a bit meager at short distance.
I have with my smartphone ~300Mbit in the bathroom through 3 walls
Yep.. I get decent speed on my Oneplus on stock too.
ho1Aetoo wrote:
you don't really need to test at a distance of less than 4m... that's all too close.
The devices just scream at each other
At 1m distance you can easily go down to 13dB (but I have not yet tested whether then the transmission power is still too high)
In any case, this is not an advantage when you test like this - there are definitely problems with too high transmitting power.
5g - clear channels no interference. I believe I'm on vht80 channel 100+6 and nearest neighbor is on 87. 100% signal quality. 3 meters from router.
ack900
my tablet gets about 280mb/sec
my phone gets about 105mb/sec
Joined: 21 Jan 2017 Posts: 1782 Location: Illinois Moderator
Posted: Wed Mar 10, 2021 20:59 Post subject:
I still can't believe how people miss the stickied "QCA BEST WIFI SETTINGS" topic at the top of the Atheros forum.
I own a R9000 and all the settings in that sticky are the ones I use (I made that sticky); with the recent exception of the new drop down for vanilla QCA drivers or DD-WRT drivers to be used for the specific radios.
I am in the process of making new pictures for the sheeple to follow; but, in the mean time- USE THE LINK IN MY SIGNATURE AND FOLLOW ALL THE SETTINGS IN THE PICTURE EXACTLY....
The exceptions right now are:
1.U-ASPD is garbage on the router and doesn't seem to work at all so keep that feature off for both radios.
2.ACK Timing =900 for both radios
3.For Beacon interval, use 300 and DTIM =1 for both radios.
4.For the 2.4ghz radio use Vanilla only and NG-Mixed (5ghz I'm using DD-wrt, but Vanilla s/b ok too)
*All other settings are mission critical to assure your clients connect at the highest MCS rate. Don't deviate from what's in the pictures. _________________ FORUM RULES
Joined: 03 Jan 2010 Posts: 7568 Location: YWG, Canada
Posted: Wed Mar 10, 2021 23:21 Post subject:
Edited by moderator. The comments are being presently discussed off-line. Steps underway to improve the wiki and best settings sticky. "Slamming" without any evidence/data is not appropriate.
edit: nice censorship, there was no slamming or insults. typical forum mods i guess.. but whatever, figure out the settings urself since clearly they "know everything" _________________ LATEST FIRMWARE(S)
BrainSlayer wrote:
we just do it since we do not like any restrictions enforced by stupid cocaine snorting managers
I still can't believe how people miss the stickied "QCA BEST WIFI SETTINGS" topic at the top of the Atheros forum.
I own a R9000 and all the settings in that sticky are the ones I use (I made that sticky); with the recent exception of the new drop down for vanilla QCA drivers or DD-WRT drivers to be used for the specific radios.
I am in the process of making new pictures for the sheeple to follow; but, in the mean time- USE THE LINK IN MY SIGNATURE AND FOLLOW ALL THE SETTINGS IN THE PICTURE EXACTLY....
The exceptions right now are:
1.U-ASPD is garbage on the router and doesn't seem to work at all so keep that feature off for both radios.
2.ACK Timing =900 for both radios
3.For Beacon interval, use 300 and DTIM =1 for both radios.
4.For the 2.4ghz radio use Vanilla only and NG-Mixed (5ghz I'm using DD-wrt, but Vanilla s/b ok too)
*All other settings are mission critical to assure your clients connect at the highest MCS rate. Don't deviate from what's in the pictures.
1. is not enabled anywhere on any of the screenshots
2. has already been recommended in this way, apart from that other values also work
3. 100 million routers all work wonderfully with a beacon interval of 100
4. we also already had
Joined: 21 Jan 2017 Posts: 1782 Location: Illinois Moderator
Posted: Thu Mar 11, 2021 15:48 Post subject:
ho1Aetoo wrote:
3. 100 million routers all work wonderfully with a beacon interval of 100
Most routers in the world are indeed using 100 and DTIM of 2...
this means 200 time units between beacons....
That doesn't necessarily mean that it's the best or right value to use. Shorter multiples of time units yields less battery life because the client is always replying back to the beacon.... Additionally, VAP's using atheros require that you use at least 100 time units between beacons. anything shorter won't work and the VAP won't broadcast. BS has hard baked a special check to assure if a beacon interval shorter than 100 is set- it will be overridden.
I have not found any clients that don't like 300 TU's... I have exchanged PM's where prime number beacons cause issues, so that theory is going to be yanked from the wiki soon when I get around to editing it again.
The fact remains. boosting the beacon to 300 should have no detrimental affect on a client and should help boost the client's battery drain from the wifi.
While this Applies to Bluetooth the same principles apply for wifi:
But here is how it looks to the router when you physically move and have the various beacon intervals:
So, you can see that having a super short beacon interval is not really useful and setting it between 300 and 600 is a happy "split the difference" solution. This is what ultimately led me to use the values I am suggesting.
I am open to anyone challenging this stance with actual white papers, articles, studies, etc.
All I know is that using a longer beacon stopped noise ("pulse") created by "smart" appliances from taking down the wifi radios temporarily on a A7v5, R7800, and R9000 in my house when those appliances are turned on.
I am certainly willing to play around more...maybe a beacon of 100 and dtim=3 will yield stable results...can't test until late night, though. _________________ FORUM RULES