I have the 3 lines in my diffconfig at the top.
When i want ath10k i change those 3 lines to:
# Use "make defconfig" to expand this to a full .config
CONFIG_TARGET_ipq806x=y
CONFIG_TARGET_ipq806x_generic=y
CONFIG_TARGET_MULTI_PROFILE=y
CONFIG_TARGET_DEVICE_ipq806x_generic_DEVICE_netgear_r7800=y
CONFIG_TARGET_DEVICE_PACKAGES_ipq806x_generic_DEVICE_netgear_r7800="-kmod-ath10k-ct -ath10k-firmware-qca9984-ct kmod-ath10k ath10k-firmware-qca9984"
CONFIG_TARGET_PER_DEVICE_ROOTFS=y
# CONFIG_PACKAGE_kmod-ath10k-ct=m
CONFIG_PACKAGE_kmod-ath10k=y
# CONFIG_PACKAGE_ath10k-firmware-qca988x-ct=m
# CONFIG_PACKAGE_ath10k-firmware-qca99x0-ct=m
# CONFIG_PACKAGE_ath10k-firmware-qca9984-ct=m
# CONFIG_PACKAGE_ath10k-firmware-qca988x=m
# CONFIG_PACKAGE_ath10k-firmware-qca99x0=m
CONFIG_PACKAGE_ath10k-firmware-qca9984=y
Hey, @asvio,
I saw all the graphs, and to me, something is weird. So let me summing @dtaht too. Can you explain the topology of your network to us?
I'll try.
The r7800 is connected to the isp's router which is independently serving another network.
the r7800 supports several wireless printers under 2.4ghz, 3 wired computers and two notebooks connected to 5ghz. Normally I create an independent 5ghz wireless network for the mobiles, now they are hooked to a 2.4 network to be able to do tests on the 5ghz network.
I've sorted it out using my old diffconfig for ath10. Currently compiling. Let me see the final result.
i will change to 80 mhz and test.
So is something like this?
ISP router <--eth--> r7800 (flent server/irtt server) <--WiFi HE--> flent test machine
Yes, it's right
Okay, if you have time, try to compile my patches. Unfortunately, your results are horrible in that kind of test.
Or, if you have time, try to run:
for ac in 0 1 2 3; do echo $ac 2000 2000 > /sys/kernel/debug/ieee80211/phy0/aql_txq_limit; done
for ac in 0 1 2 3; do echo $ac 2000 2000 > /sys/kernel/debug/ieee80211/phy1/aql_txq_limit; done
Sorry, I don't know which one is your 5 GHz phy interface.
ISP router <---eth---> r7800 <---eth---> flent server/ irtt server
^
|
|--------WiFi----> flent test machine.
This is more real...
ieee80211/phy0
Yeah, that's fine, that's similar to my testing scenario.
Okay, why don't you try compiling with the two following patches in? Click me to download.
Just add them to your quilt series under <openwrt-git-home-folder>/build_dir/target-*/linux-*/linux-*.
@asvio, then this is the one
for ac in 0 1 2 3; do echo $ac 2000 2000 > /sys/kernel/debug/ieee80211/phy0/aql_txq_limit; done
Over 200 ms latency, that's crazy. Please, tell me your VM is running virtio on the network interface, otherwise, this latency might be induced by the VM.
No. It an easy made vm. Nothing special. Just for fun...
It will be ideal if you can keep running tests on the Kubuntu machine. However, try to apply the patches to see if they make any difference. 150 ms latency is terrible; 30-ish should be your landing spot. And, on AX, even lower.
Graph with this setting.
for ac in 0 1 2 3; do echo $ac 2000 2000 > /sys/kernel/debug/ieee80211/phy0/aql_txq_limit; done
I'm sorry I have to go now. Tomorrow I can do more tests. I'm at CEST.
The ax210 is supposed to fq_codel also, and probably? also has AQL in it. An even narrower channel (HT20?) would be revealing. And probably depressing.
I had NOTHING to do with the ax210 port to fq_codel, I just made the assumption that the 300 member engineering team at intel would get it right.