Re: mt76x0u. Unusable performance while compiling stuff.

(cc linux-wireless & Lorenzo)

On Fri, Jan 03, 2020 at 12:19:20AM +0000, Daniel Similä wrote:

Bus 001 Device 005: ID 148f:761a Ralink Technology, Corp. MT7610U ("Archer T2U" 2.4G+5G WLAN Adapter

Driver is mt76x0u
Wifi is 5GHz ~260Mb

while doing:
nice make -j16 

Pinging my router goes from:
ping _gateway
PING _gateway (192.168.0.1) 56(84) bytes of data.
64 bytes from _gateway (192.168.0.1): icmp_seq=1 ttl=64 time=1.20 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=2 ttl=64 time=1.02 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=3 ttl=64 time=2.12 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=4 ttl=64 time=1.32 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=5 ttl=64 time=2.04 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=6 ttl=64 time=1.31 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=7 ttl=64 time=1.71 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=8 ttl=64 time=0.903 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=9 ttl=64 time=1.00 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=10 ttl=64 time=1.42 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=11 ttl=64 time=1.15 ms

to unusable:

ping _gateway
PING _gateway (192.168.0.1) 56(84) bytes of data.
64 bytes from _gateway (192.168.0.1): icmp_seq=1 ttl=64 time=167 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=3 ttl=64 time=306 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=4 ttl=64 time=255 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=6 ttl=64 time=253 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=8 ttl=64 time=84.7 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=9 ttl=64 time=522 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=10 ttl=64 time=211 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=11 ttl=64 time=304 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=15 ttl=64 time=2825 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=16 ttl=64 time=1812 ms
64 bytes from _gateway (192.168.0.1): icmp_seq=17 ttl=64 time=799 ms

Running kernel 5.4.6-arch3-1
CPU is Amd 2700X and ram is 16GB

I think this should be fixed by 5.5 commit:

commit 284efb473ef5f02a7f2c13fdf8d516ecc589bdf1
Author: Lorenzo Bianconi [off-list ref]
Date:   Mon Oct 28 17:38:05 2019 +0100

    mt76: mt76u: rely on a dedicated stats workqueue

and if it indeed fixes the problem, this commit should be requested
to stable.

Stanislaw