[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[linrad] Major revision. Linrad-01.18
Problems solved in Linrad-01.18:
After upgrading to Red Hat 9, PentiumIV and the latest
version of OSS there have been occasional overrun errors
on my system. The problem turned out to be the usleep(20000)
call which sometimes causes a delay above 0.1 second.
By placing a call to gettimeofday() immediately before
the usleep() call and by making sure that the A/D buffer
is nearly empty it has turned out to be possible to make
the releasing of the processor not last more than the
Another problem with my new "fast" computer is that the
graphics is very slow. The AFC graph can not be wider than
30% of the screen otherwise the time to redraw it becomes
longer than the time it takes to fill the A/D buffer.
(Moving one pixel vertically takes 30 microseconds!!!!)
The new computer has a nvidia gforce4 board, it is in a
new and faster AGP slot into which my good old board
does not fit. svgalib selects the VESA driver, maybe that
is the reason it is so very slow.
(or maybe the old sound driver had a much larger buffer
so I did not notice how long time it took to move
The actual dsp computations take very little time, the
graphics dominates. To make Linrad run smoothly on this
kind of computer I have changed the graphical routines
to only update a fraction of the graph on each call.
This means (of course) that many changes have been required
in the main receive loop that keeps track of the priorities
of different things.
The oscilloscope functions are still useless on my new
computer, but they are not needed to operate Linrad.
I will try to install svgalib-1.9.18, hopefully it
includes a properly working driver for the GeForce card.
Because of the large number of changes I have not put
Linrad-01.18 at the "Linrad Home Page", it is only
available at the development page and its mirrors:
Please report any bugs!
Reports about differences in behaviour (if any) are also welcome.
Leif / SM5BSZ