Comment 42 for bug 1480877

I just tried Tony's lp:~phablet-team/network-manager/lp1480877-wifi-rm-dup-scan-signals branch and did the same thing I've done before to trigger this bug: walk around my neighborhood with Terminal open, running top.

To be fair, I was testing rc-proposed tip + Tony's branch, whereas I previously tested with OTA6. So there's more than one change here. But I haven't heard of related changes in this area, so I will attribute all improvements to Tony's branch.

Before this patch, I was seeing NM spike to 30-50% CPU while scanning and causing a matching spike in dbus-daemon to 100% for a while (at least 3s, sometimes a fair bit longer).

With this patch, NM stayed at ~15% when scanning, and dbus-daemon never went above 7% (usually at 4%). A massive improvement.

From a UX perspective, I still saw UI stuttering. Seemingly without the CPU-load justification to warrant it. But I subjectively don't think it was as bad or long-lasting as before. And top doesn't lie. The CPU was way less loaded with this patch than without it.