On Mon, 20 Aug 2012, Mike Belopuhov wrote:
Always using the tx interrupt decreases performance significantly. On my
test system (in MBytes/s):
Sending UDP from 75 to 55
Sending TCP from 34 to 25
Why?
Because the interrupt load increases by approx. 7000 per second. I think
the problem is that the backend sends packets faster than openbsd can
enqueue them. This means that the tx queue will be empty often, and each
time an tx interrupt will be triggered.
I dont' think we want to do this.
Perhaps you got your OACTIVE handling wrong and don't call a start
routing from the appropriate places. Have you considered that?
I don't see how this could help, but I will look at it.