Voxel Custom firmware build for R7800 v. 1.0.2.81SF & v. 1.0.2.81.1SF

[kokishin]

@Voxel

First of all thanks again.

I uploaded 81.1SF to my 7800. After the upload completed, I was expecting to click on the "Yes" button as usual to install and reboot. However, my 7800 uploaded, installed, and rebooted without any further intervention from me. I did not even see the "Yes" button.

Just an observation. Not an issue.

 

[Giudi001]

I just upgraded to 81.1SF and as far as I can tell, 5GHz performance on channel 36(+40,44,48) is unchanged with respect to 81SF. To my knowledge, there are no radars in my vicinity that could trigger DFS.

Dear,
but have u fiber gigabit internet connection to test speed?

 

[Voxel]

i've just upgrade to 81.1 but i've very slow wifi performance with this version with same settings!!!!
With this 81.1, speed test in download is 300Mb and 200mb in upload. With old firmware, like .79, speed test in download 910Mb and 200mb in upload with 5GHZ wifi. I've just tested this with my laptop.

Thanks for your report. Questions to clarify:

(1) Did you check 81SF (not 79SF) vs 81.1SF?
(2) Could you test your speed with cable connection (not Wi-Fi)?

I mean that AP mode fix touches WAN<->LAN driver(s). Wi-Fi itself is not changed. But Wi-Fi is also LAN<-WAN. I had to use QCA/NG/DNI codes "as-is" from the stock 1.0.2.74. I cannot confirm or deny that their changes are absolutely correct:

Fix AP mode issue ('qca-nss-ecm' package) · SVoxel/R7800@1109553

Custom firmware build for R7800. Contribute to SVoxel/R7800 development by creating an account on GitHub.

github.com

it is not my own changes of codes. And part of QCA specific drivers are enclosed, so it is like apart of "black box" for me. Moreover, GPL sources from NG/DNI do not 100% corresponds to binary version of firmware (e.g. missing kernel module in GPL, just is not compiled and I see this module in binary version of their firmware...).

So suspicious part is WAN<->LAN driver changed in 81.1SFtofix AP mode issue with WAN port. Wi-Fi itself i.e. your Client<->R7800 should have the same speed. My tests with iphone8 and laptop (sorry not so fast Wi-Fi card in laptop):

iperf3 from iphone8

laptop (Wi-Fi connection):

There is no degradation to 200/300 mbit. The same I had before 81.1SF.

Voxel.

 

[Voxel]

@Voxel

First of all thanks again.

I uploaded 81.1SF to my 7800. After the upload completed, I was expecting to click on the "Yes" button as usual to install and reboot. However, my 7800 uploaded, installed, and rebooted without any further intervention from me. I did not even see the "Yes" button.

Just an observation. Not an issue.

It is just strange logic of NG in versions of firmware. It recognize 1.0.2.81.1SF as more fresh vs 1.0.2.81SF. So no questions from WebGUI but just flashing. Just ignore this.

Voxel.

 

[Giudi001]

Thanks for your report. Questions to clarify:

(1) Did you check 81SF (not 79SF) vs 81.1SF?
(2) Could you test your speed with cable connection (not Wi-Fi)?

I mean that AP mode fix touches WAN<->LAN driver(s). Wi-Fi itself is not changed. But Wi-Fi is also LAN<-WAN. I had to use QCA/NG/DNI codes "as-is" from the stock 1.0.2.74. I cannot confirm or deny that their changes are absolutely correct:

Fix AP mode issue ('qca-nss-ecm' package) · SVoxel/R7800@1109553

Custom firmware build for R7800. Contribute to SVoxel/R7800 development by creating an account on GitHub.

github.com

it is not my own changes of codes. And part of QCA specific drivers are enclosed, so it is like apart of "black box" for me. Moreover, GPL sources from NG/DNI do not 100% corresponds to binary version of firmware (e.g. missing kernel module in GPL, just is not compiled and I see this module in binary version of their firmware...).

So suspicious part is WAN<->LAN driver changed in 81.1SFtofix AP mode issue with WAN port. Wi-Fi itself i.e. your Client<->R7800 should have the same speed. My tests with iphone8 and laptop (sorry not so fast Wi-Fi card in laptop):

View attachment 28589

iperf3 from iphone8

View attachment 28590

laptop (Wi-Fi connection):

View attachment 28591


There is no degradation to 200/300 mbit. The same I had before 81.1SF.

Voxel.

Dear Voxel,
I had checked with the 81 but I did not experience any performance anomalies, only the frequent deactivation of the 5GHZ network for a few seconds.
I had brought the fw back to version 79 to avoid disconnections, (strangely though they were also repeated with this fw after 1-2 days of use and in any case less frequent than the 80 version).

Last night, therefore, I switched from version 79 to 81.1, I checked the performance only in wifi (not via ethernet) and I found what was reported through speed test ookla.
I immediately rethroned at .79, performed the same speed test on the same server and had the usual best performances I've always had.

My setting is router and not AP,

 

[Voxel]

Dear Giudi001, I am sorry but such test

Last night, therefore, I switched from version 79 to 81.1, I checked the performance only in wifi (not via ethernet) and I found what was reported through speed test ookla.
I immediately rethroned at .79

reports just nothing to the developers of firmware. I.e. flashing new firmware, 10-15 min and back to previous version (your phrase "I immediately rethroned at .79" assumes very short time period).

After flashing new version of firmware your R7800 performs some additional actions/downloads what could influence to the speed measurement, for example: it downloads ReadyCLOUD, it downloads KWILT/Hipplay, started by default Transmission could initially eat some traffic etc. If you did not disable all of this...

I understand your wishes as end-user: to flash firmware and immediately use it. According to my knowledge (and I hope I really know that as a developer) all internal changes 79SF->80.XSF->81SF->81.1SF should NOT influence to Wi-Fi speed. I.e. when you use your iphone there is the sequence:

iphone <-> Wi-Fi LAN port <-> [R7800] <-> WAN port <-> OOKLA

And the only specific changes are in 81.1SF, i.e. changes in LAN<->WAN interface. This could be potentially problematic.

Also, OOKLA is good resource. But results of OOKLA depends on remote server randomly selected by OOKLA, current load of this server, etc. Precise tool is iperf3. So I have to use it in my testing stand. Currently this stand is connected to R7800 WAN port by cable (1Gb/sec).

And according to my tests

Client<->Wi-Fi LAN port<->[R7800]: the same speed as with 79SF
(i.e. iperf3 is running directly on R7800 and on Wi-Fi Client).

[R7800]<->WAN port<->[IPERF3 testing server]: the same speed as with 79SF
(i.e. iperf3 is running directly on R7800 and on testing server).

Client<------->[IPERF3 testing server]: the same speed as with 79SF
(i.e. iperf3 is running on Client and testing server).

I.e. I do not face any degradation of LAN<->WAN where LAN could be either cable or Wi-Fi,

So I'd expect comments, confirmation or denial from others who are using it, say, more than 1 day or so.

Thank you anyway,
Voxel.

 

[Giudi001]

Thank you Voxel,
I will try to put the 81.1 back and do further tests.
Mine is not a complaint, I just want to contribute to this "project", thank you for the work you do for all of us!!!
Just to clarify, the 81.1 firmware was running for about 1 hour before carrying out OOKLA speed tests, not with random servers but always with my favorite and carried out at least 6-7 tests from iPhone and pc.
Just downgraded to the 79 (about 3minutes to put it back) on power on again I carried out the speed test in wifi from pc and iPhone on the same OOKLA server with the usual speed results that I had indicated to you.
I can tell you that I do not use ethernet because I do not have fixed seat of the pc, I can also try with a cable to see if, the problem of firmware 81.1 that I verified yesterday is due to wifi as I indicated, or in general of the wan port and therefore also with cable I should get about 300mb instead of the usual 990mb.

I update you

 

[Voxel]

I can tell you that I do not use ethernet because I do not have fixed seat of the pc, I can also try with a cable to see if, the problem of firmware 81.1 that I verified yesterday is due to wifi as I indicated, or in general of the wan port and therefore also with cable I should get about 300mb instead of the usual 990mb.

Also, you can run OOKLA speedtest from WebGUI of router.

Voxel.

 

[kokishin]

Hi @Voxel,

Just sent you my annual donation via PayPal.

Thank you for all the time and energy you put into your firmware releases and posts in snbforums.

Happy holidays and all the best!

 

[LeKeiser]

Hello,
81.1SF installed yesterday. As usual, everything is working very well.
Same thing than @kokishin happened during the installation : I didn't have to approve the upgrade. It started right after the upload, and then the router rebooted right away.
WiFi is working well and steady.
Again, I tip my hat to Mr Voxel for all his continuous work

 

[scooter2765]

Thanks Voxel ...

 

[Voxel]

Again, I tip my hat to Mr Voxel for all his continuous work

You can leave your hat on...

Thank you,
Voxel.

 

[LeKeiser]

Mmmh there is something that's been on my mind though.
I have a NAS where I store my movies. This NAS has a Gigabit network card, and the two disks I have inside are configured as a Raid 0 pair. I used to transfer movies over ethernet at around 100MB/sec. Now, I barely reach 50MB/sec. I thought this drop in bandwith was due to some Windows updates. Some people have noticed the same thing.
I'm just want to make sure that my R7800 isn't the culprid, something in the configuration that has changed and that I could have missed. Any way I could make sure it is Windows related?
Thanx for your thoughts on this little problem

 

[Murtaza12]

I had been away from home for 3 months and was therefore unable to keep my R7800s up to date with Voxel releases. Last version they were running was .79SF. Both of them had around 102 days of uptime, zero issues. Thank you Voxel for the great stability and performance with every release.

Just updated both to 81.1SF, ran some before and after iPerf tests, got pretty similar results after the FW upgrade.
The first screenshot is while on 79SF, and the second screenshot is on the latest firmware.

Tests were done on a Lenovo laptop with an Intel AX200 WiFi card. R7800s were configured with Ch149, 5 GHz, 80 MHz channel width and MU-MIMO enabled.

 

[Voxel]

Just updated both to 81.1SF, ran some before and after iPerf tests, got pretty similar results after the FW upgrade.
The first screenshot is while on 79SF, and the second screenshot is on the latest firmware.

Thank you for your tests.

Voxel.

 

[Voxel]

I have a NAS where I store my movies. This NAS has a Gigabit network card, and the two disks I have inside are configured as a Raid 0 pair. I used to transfer movies over ethernet at around 100MB/sec. Now, I barely reach 50MB/sec. I thought this drop in bandwith was due to some Windows updates. Some people have noticed the same thing.
I'm just want to make sure that my R7800 isn't the culprid, something in the configuration that has changed and that I could have missed. Any way I could make sure it is Windows related?
Thanx for your thoughts on this little problem

Well, I have something like this. Week or two. Maybe Windows update. Access to my Linux server with SMB/FTP file storage is twice slower vs usual immediately after I resume my laptop from hibernation mode. 50MB/sec instead of more than 100MB/sec. After some time speed is OK. Or if I reboot my server. I do not think it is R7800 because I do not reboot it... And no something bad in R7800 system messages. Still investigating.

Voxel.

 

[LeKeiser]

Well, I have something like this. Week or two. Maybe Windows update. Access to my Linux server with SMB/FTP file storage is twice slower vs usual immediately after I resume my laptop from hibernation mode. 50MB/sec instead of more than 100MB/sec. After some time speed is OK. Or if I reboot my server. I do not think it is R7800 because I do not reboot it... And no something bad in R7800 system messages. Still investigating.

Voxel.

Thank you Mr Voxel
That's what I thought. I don't think our R7800 has anything to do with this drop of bandwith.
Alas, I have only Windows devices here... No more Linux server. So I cannot investigate on my side. But all my devices report the same slowness. So it has to be Windows related. I updated my network card on my desktop, no change.

 

[SteevieWeevie]

Had a strange thing happen after upgrade to 81. Everything was fine for about 30 minutes. Then the 5G WiFi just stopped working. Unplugged the router and restarted it and it seems fine now. I've never seen that happen. Will keep an eye on that.

I have this happen routinely shortly after flashing new Voxel firmware versions on it. I have found that upgrading the firmware, allowing a few minutes to settle and then manually rebooting the router avoids this from occurring.

In my experience the 5Ghz drop will then not happen and the router will just work indefinitely until you come to upgrade the firmware again. Though that is fairly often for me because I always want to put the newest Voxel on it

 

[Squuiid]

I have this happen routinely shortly after flashing new Voxel firmware versions on it. I have found that upgrading the firmware, allowing a few minutes to settle and then manually rebooting the router avoids this from occurring.

In my experience the 5Ghz drop will then not happen and the router will just work indefinitely until you come to upgrade the firmware again. Though that is fairly often for me because I always want to put the newest Voxel on it

+1. I also unfortunately had the exact same issue @Sizzlechest had and I ended up going back to stock. I really wanted to like and use Voxel's firmware but this issue was very clear cut and unquestionably tied to the cfw. This was a few versions back though, not 81. I lost confidence in it when 5GHz just dropped, something it had never ever done before.
To give some context to my findings, I'm an IT professional with several decades of experience and have written some fairly technical posts on other forums, I'm certainly not new to these sorts of things. I write this because I do think it is an issue worth looking into and not to just be dismissed.
https://forums.macrumors.com/thread...nics-to-use-small-tree-macos-drivers.1968456/

Thanks again Voxel for what you do here, it is mighty impressive work. Just wish I could make it work for me too.

 

[Voxel]

Thanks again Voxel

OK, some belletristic. Well, it happens sometimes. That my “customers” are not satisfied by the result of my job. Pity of course. But life is life. You should know, as an IT professional that e.g. your support of end-users depends on your own and your Company resources and capabilities. Time and money. My builds are free for all of you. It is just my hobby. I do them (builds) in my free time, it is a some kind of a toy for me. There are no any profits from this dealing for me. So primary for me (to survive) is my main job (to get the money for a life). My family. And when I have a time after all of this: the rest is for my hobby. And I do these build alone. I am not NG with $$$ budget and thousand of employees.

I do not ignore all of your issue/alarm reports. But. I am limited in reproducing (and thus fixing) all of problems/issues reported by the users of my firmware. I have about 12 devices connected to three of my R7800 (5GHz only). Intensively used by the member of my family. 24/7. For a real job and just for a fun. Also my own AP/routers/extenders not visible for my family. And they (family) did not report me any problems similar you have (otherwise they would not let me drink my beer during weekends but I’d have the only brainwashing ). So I am asking you are a professional: how to fix what you are not able to reproduce? Usually I try to pick up the statistics: how many reports of this or this. In my brief estimation there are about (well, I am as a modest man, just according to the amount of downloads) 10 thousands of my users. And of course some of them could report some issues. Statistics. If there are thousands of people in the place, e.g. meeting, one or two could get infarct not because of the meeting itself, but because of other reasons. What is that: the real bug or just side effect – I do not know in real. So if, say, 10 reports: seems as a real issue. Alarm for me. One/two reports: I am waiting/watching. If I know that something could happen because I know that I’ve changed this part of fw : I ask user to clarify. But your case: sorry. I do not face. My family (very strict testers): they do not face. So… Sorry. Stock fw is good for you: it is fine. Good luck.

P.S.

I'm an IT professional with several decades of experience

I do think it is an issue worth looking into and not to just be dismissed.

And, just to clarify: please believe me that my experience in IT area (and in real commercial development and IT support ) is much more than your. The same e.g. for kamoj. For example: do you know how to fix the bug using sharp knife with a punch card We are not ignoring any of your reports.

Stay safe,

Voxel.