yes, r.4.3.1.
i used both method. tried to install .ipk directly with errors, and added your repo and tried to install from software section, the same result -incompatible kernel version
upd
in the evening ill try your link for repo, perhaps i made any mistake
For sure you are doing something that is wrong.
Sure. Are you interested in seeing 2.4 or 5ghz? Or both?
now it works, thank you o/
Just to confirm, @pesa1234, the script has been renamed from irq_affinity to smp_affinity -- is that correct?
Yes it have also:
a stop function in case you want to disable
It move usb irq to cpu2
I mostly care about 5ghz. And longer than a minute if possible - 5 minutes. Over here are some tests at 25 feet from last month showing things falling apart over time. https://github.com/Zoxc/crusader/discussions/55
any suggestions in case if i don't need wifi enabled. Is there a way of removing the wifi driver removed?
While I'm on the learning curve, I've not yet figured out how to do tests beyond cloudflare/bufferbloat. I hope others can help.
Set the transmit power to the lowest value (just in case they kick on briefly at startup or something — unlikely bordering on impossible but just in case™ ) and turn off the radios. This is what I do with old devices I am using as a switch. No reason to worry about removing anything.
Sorry for the dumb question, but by reading this I was wondering: do I have to do anything to set/run this script? Is everything done automatically?
Not sure if these settings are related. I didn't configure anything for irqbalance specifically.
irqbalance is one thing move the event (irq) from one core to an other
packet steering move the packets from cpu cores
the script in smp_affinity 4.3.1 fix ethernet event to cpu 1-3 usb on 2 and wifi on 0 (all cores to be onest, I leave to ff)
Packet steering can be setted as Enabled, and it works in this way:
use RPS to redirect packets to a single core only, then try to spread NAPI thread and RPS target CPUs across available CPUs
and try to ensure that the NAPI thread is on a different CPU than the
RPS target.
Or enabled (old style) and it will divide all packet to all available cores.
The enabled version teorically should be better, but you have to check on your configuration.
You can see differences with 1gbit/s connection, with less connection maybe you can not find any differences.
Dear Pesa1234, may you check whether the following packages may be added to your custom build? I can't install them manually: kmod-nft-queue, kmod-nf-conntrack, kmod-nfnetlink-queue
Dear Pesa1234 can you had "kmod-tcp-bbr"
Yes, in next release.
Maybe a new release at the end of this week
Dear Pesa1234, what about my question above? thanks.
Please read post #1.
Please give the error that you get
Pesa1234, sure, I did this, but packages kmod-nft-queue and kmod-nfnetlink-queue are unsearchable.
Ok, I'll add on next build, only these 2 or do you need others?
kmod-nft-queue and kmod-nfnetlink-queue + please add also kmod-netlink-diag, thank you!
