[openib-general] [PATCH] mthca: move code from post send to post receive

[Michael S. Tsirkin]

>  > While unlikely to give a large gain, this makes sense to me.
> 
> Out of curiousity -- can you measure any difference at all with this?
> I would have guessed that the addition can be scheduled so that it
> costs nothing at all on any common CPU.

I didn't actually try to measure it.
But maybe it will all add up with time as small tuning adjustments are done.

> I guess it doesn't hurt though. Want to make a similar patch for libmthca?

Sure.

-- 
MST