Build for Netgear R7800


#1365

thx your reply
i see


#1366

How do you guys find the perfomance of openwrt on the R7800? Is there much noticeable difference from losing the stocks hardware acceleration?

Ive tried openwrt a while back but it was a little complicated for me at the time so i switched back to voxels firmware adaption of stock,,, but im now ready to give it another go. So ive downloaded the latest master build from @hnyman, ive picked the old ath10k version as the -ct version seems to get mixed reviews, (thanks for your hardwork with these builds btw)

Im currently reading up how to add some static ip addresses as some devices i need in specific places... and ive got to locate the method to block certain firewall ports to certain ip addresses.

Is there any simple way to see router wan sync speed? On stock netgear firmware theres a speedtest link that gives the speed from modem to router, i know its not something you need daily, but its nice to keep an eye on occasionally.


#1367

root@OpenWrt:~# hdparm -Tt /dev/sda1
/dev/sda1:
Timing cached reads: 1288 MB in 2.00 seconds = 644.67 MB/sec
Timing buffered disk reads: 172 MB in 3.09 seconds = 55.58 MB/sec
root@OpenWrt:~# hdparm -Tt /dev/sdb1
/dev/sdb1:
Timing cached reads: 1270 MB in 2.00 seconds = 635.50 MB/sec
Timing buffered disk reads: 664 MB in 3.01 seconds = 220.79 MB/sec
[/quote]

I have 60 MB / s writing on the cable, reading 40 MB / s. I would like to achieve higher values. It is possible?

Timing cached reads: 1324 MB in 2.00 seconds = 662.48 MB/sec
Timing buffered disk reads: 496 MB in 3.01 seconds = 165.01 MB/sec**


#1368

hi all
i just bought a r7800 and installed owrt1806-r7638-c5c20f510a-20190105 and i have problems with the download and upload speed

i tested the speed from router to my pc with iperf3 and obtained the following resullts:
iperf3 -4 -c 192.168.1.1 (pc to router)
[ ID] Interval Transfer Bandwidth Retr
[ 4] 0.00-10.00 sec 975 MBytes 818 Mbits/sec 0 sender
[ 4] 0.00-10.00 sec 975 MBytes 817 Mbits/sec receiver

iperf3 -4 -c 192.168.1.1 -R (router to pc)
[ ID] Interval Transfer Bandwidth Retr
[ 4] 0.00-10.00 sec 546 MBytes 458 Mbits/sec 0 sender
[ 4] 0.00-10.00 sec 546 MBytes 458 Mbits/sec receiver

the download speed from the r7800 to the pc is only half of the upload speed, with the settings below:

  • 1000Mb/s Full Duplex on both sides
  • SQM off
  • MTU default 1500
  • fine tuned ondemand thresholds
  • tried both ondemand and performance governors
  • tried both flow offloading on/off

any suggestion pls?


#1370

I have noticed a similar limitation with iperf on my own r7800. I am able to get full LAN to WAN speeds on my gigabit fiber (~920 down of Verizon's promised 960), however, when I run a speedtest.net test through the router. Those speeds required the tuned ondemand thresholds and minimum clock speeds discussed earlier in the thread.

However, no matter the the amount of parallel sessions or options I throw at iperf, I can't get higher than about 600 Mbps on the OpenWRT command line.


#1371

thanks but still the same with iperf3 -4 -c 192.168.1.1 -Z -R, a bit slower actually

[ ID] Interval Transfer Bandwidth Retr
[ 4] 0.00-10.00 sec 536 MBytes 449 Mbits/sec 0 sender
[ 4] 0.00-10.00 sec 536 MBytes 449 Mbits/sec receiver

btw i have checked the cpu gov, when sending from the router to the pc it's nowhere near its max. however when receiving from the pc it stays at max freq 1725000. tried to force it higher but the results were no different


#1372

yeah that's right, i believe there are some limitations with iperf perhaps since on my old router wdr3600 the speed are 2xx and 1xx respectively but actually i could get over 300 download when tested with browser

however i think this is a weak assumption unless someone come out and point out the actual problem with iperf or otherwise no one will use iperf to test the in out speed. so iperf must be reliable in most cases i believe but not in this case

i couldnt find anyone complained about this issue so far on the net regarding lede/openwrt/r7800/iperf so i asked if there is something wrong with my settings etc


#1373

sorry, a noob question, since i am not familiar with git

i am currently running OpenWrt 18.06-SNAPSHOT r7409-d5afaa4114 / LuCI openwrt-18.06 branch (git-18.340.83383-3dea6b5) as shown in the system status page and i want to try out something

i have downloaded the owrt1806-r7409-d5afaa4114-20181212.zip but after i executed the scripts, what i got was still updated to 4.14.91 instead of 4.14.82, which is the kernel version in r7409-d5afaa4114, even after following some suggestions found on the interenet. the filestamp in newbuild.sh is R7800-owrt1806-r7409-d5afaa4114-20181212-1927

so my question is, how can i get the exact 4.14.82 kernel and all the stuff with the build r7409-d5afaa4114-20181212-1927 i am running on my r7800?

thanks a lot


#1374

To rule out iperf, you can use something like dd & netcat to simulate a real-world transfer and do the math. You'd want to write to /dev/null on the router to rule out IO bottlenecks and avoid any encryption or compression.
i.e. Something along the lines of the following should tell you the time to send a 100MB file from the laptop to the router:
router $ nc -l -p 1234 > /dev/null
laptop $ dd if=/dev/zero of=100MBfile bs=1048576 count=100
laptop $ time nc -w 3 192.168.1.1 1234 < 100MBfile

You can then reverse laptop and router to see time in the other direction (router to laptop).


#1375

you r the man! i have totally forgot this command.

unfortunately the one comes with the busybox doesnt support other params so i will have to find one or to build one

thanks a lot

edit ok, just installed netcat


#1376

When testing router network performance, generally do NOT run benchmark program on router itself as it uses part of routers power that is normally available to handle traffic. This is particularly important with iperf(3).

Get two fast enough computers to run benchmark programs and insert router into traffic between of those two.


#1377

Kind of off topic to this thread, but for eg. https://192.168.1.1/cgi-bin/luci/admin/network/vlan


#1378

yes u r right. this was the 1st time i did a real test on the router and i didnt realize it's cpu bound. thanks a lot


#1379

regarding the git question...
anyone pls? or it's simply impossible (i dont think so though.. IMHO)?


#1380

finally have time to perform some tests. here are the results, in case anyone interested

** test1
#cat /sys/devices/system/cpu/cpufreq/policy?/scaling_governor
performance
performance
#cat /sys/devices/system/cpu/cpufreq/policy?/cpuinfo_cur_freq
1725000
1725000
#time dd if=/dev/zero  bs=1MB count=1000 | netcat -c 192.168.1.2 1234 (router to pc)
1000+0 records in
1000+0 records out
real    0m 39.55s
user    0m 0.04s
sys     0m 8.40s
#time dd if=/dev/zero  bs=1MB count=1000 | netcat  192.168.1.1 1234 (pc to router)
1000+0 records in
1000+0 records out
1000000000 bytes (1.0 GB) copied, 19.2954 s, 51.8 MB/s
real    0m19.314s
user    0m0.268s
sys     0m2.772s

** test2
#cat /sys/devices/system/cpu/cpufreq/policy?/scaling_governor
ondemand
ondemand
#cat /sys/devices/system/cpu/cpufreq/ondemand/up_threshold
95 (default)
#cat /sys/devices/system/cpu/cpufreq/ondemand/sampling_down_factor
1  (default)
#cat /sys/devices/system/cpu/cpufreq/policy?/cpuinfo_cur_freq
384000
384000
#time dd if=/dev/zero  bs=1MB count=1000 | netcat -c 192.168.1.2 1234 (router to pc)
1000+0 records in
1000+0 records out
real    1m 23.64s
user    0m 0.03s
sys     0m 12.98s
#time dd if=/dev/zero  bs=1MB count=1000 | netcat  192.168.1.1 1234 (pc to router)
1000+0 records in
1000+0 records out
1000000000 bytes (1.0 GB) copied, 33.5519 s, 29.8 MB/s
real    0m33.569s
user    0m0.268s
sys     0m3.332s

** test3
#cat /sys/devices/system/cpu/cpufreq/ondemand/up_threshold
35 (custom)
#cat /sys/devices/system/cpu/cpufreq/ondemand/sampling_down_factor
10 (custom)
#time dd if=/dev/zero  bs=1MB count=1000 | netcat -c 192.168.1.2 1234 (router to pc)
1000+0 records in
1000+0 records out
real    0m 38.51s
user    0m 0.04s
sys     0m 5.93s
#time dd if=/dev/zero  bs=1MB count=1000 | netcat  192.168.1.1 1234 (pc to router)
1000+0 records in
1000+0 records out
1000000000 bytes (1.0 GB) copied, 19.7107 s, 50.7 MB/s
real    0m19.720s
user    0m0.256s
sys     0m3.056s


** test 4 (more info)
#time dd if=/dev/zero  bs=1MB count=1000 of=/dev/null (on router)
1000+0 records in
1000+0 records out
real    0m 0.46s
user    0m 0.00s
sys     0m 0.46s

and still, the speed from pc to router is more than twice of the speed from router to pc
so it this normal and nothing to worry about?

pls suggest

thank you


#1381

Gentlemen, what is the best sqm setting for a bookmark: Queue Discipline ?

I try different settings but the differences are quite small


#1382

Dear all!
I am desperately trying to get openwrt onto my r7800 but every time I am ending up in a boot loop.
I am using the tftp method but no image is booting. Is there anything special I am not aware of?

I have the same issue with dd-wrt.
when tftping the stock firmware the router comes back to life, so the tftp process in general works ...

any help would be highly appreciated.


#1383

Did you try tftp, which is a necessity for switching between different a partitioning maps.


#1384

did you even read my question? :smiley:
The whole question is about tftp but boot loop...


#1385

You really use the reset button triggered TFTP method to successfully flash OEM stock firmware, but you fail to flash both OpenWrt and ddwrt ?

There is no difference in the TFTP flash process regarding OEM or OpenWrt firmware, so it should work ok for OpenWrt if you have generically figured out the TFTP triggering.

The bootloop reminds me of the year-ago trouble with Linksys WRT3200ACM, where the manufacturer had changed the flash chip for the newest devices, so that OpenWrt did not boot successfully as the new flash chip was unsupported. Linksys added support for that new chip to their newest firmware, but the older OEM firmwares were unusable, just like OpenWrt. Once that was figured out, the support was added to mtd in OpenWrt.

So, have you tried to flash an older version of OEM firmware? The current one seems to be 1.0.2.62. But what about a year-old firmware version, can you use that? If not, then it might be something similar. (That of course assumes that your device is new, so that this kind of hardware change would be recent.) I have not seen similar trouble or rumours about R7800 earlier, so a bit hard to believe, but still possible.