Hi !
> ioctl(3, 0x400c0514, 0xbffffc3c) = 0
> ioctl(3, 0x400c0530, 0xbffffc3c) = -83886082
Weird. Seems the manpage is not correct here. I always only tested for !=0,
which should do it.
Actually the behaviour is good, because errno is inherently not thread-safe,
and the construct used to make it thread-safe is very ugly.
I pretty much suppose you just get the return value from kernel handed through.
> > > optimized 2D rutines, the performance increases using the Virge HW accel
> > > - lines: 20% - boxes: 74% - hlines: 64%
> > Only ? Usually one gains more. How were those measured ?
> * lines: 1.000.000 placed pseudo-randomly with pseudo-ramdom lengths
> -SW: Bresenham algorithm, all with integers, you know: 34.68 seconds
> -HW: 27.58 seconds
Ah - for performance increase I would count that in another fashion. I.e. you
got the same work done in less time, giving a performance gain of longer/
shorter time i.e. lines 29 %, boxes 280%, hline 180%, vline 15%.
That's always the problem with the speed calculations. You can either give
the time saving or the increase in throughput.
The latter gives higher values, so it is most commonly used. :-).
The values I computed above are more in accordance with my expectations
on a fast CPU and a slow accel.
CU, ANdy
--
Andreas Beck | Email : <[EMAIL PROTECTED]>
