In article <yf91y1s103i.fsf@mc.pp.se>
marcus@mc.pp.se wrote:

> > It looks the unrolling version is a bit slower.
> > (Maybe due to calling intr suspend/resume functions)
> > I didn't check difference in interrupt response, though.
> 
> If you measured "wall clock" time, then any actual interrupts that
> are serviced would add to the total time as well.

I measured it with pkgsrc/net/netperf for network and dd(1) for storage.

Anyway, I'll commit g2bus_bus_mem.c changes without
UNROLLed part for now.
---
Izumi Tsutsui
tsutsui@ceres.dti.ne.jp