|
This weekend my faithful "grabber-PC", a nine year
old Pentium-M notebook refused to boot, which turned out to be due to a faulty
SO DIMM RAM extension board. Trying to run four or five SpecLab
instances with the original 512 MB ram led to very sluggish
behavior, and redrawing a window or even daring to start a
browser sent the harddisk into nigh endless swapping nirwana.
The cause was that I had configured SpecLab
with very large ram buffers, containing enough spectrogram lines
to refresh the whole width of each window. After startup, these buffers are
gradually growing larger and larger. Looking
again into the spectrum buffer configuration, I went for the option
to set up a large file buffer on disk. This turned out to be much more
efficient than ram plus swapping, and is really a key to running SpecLab
with limited memory. In addition there is the very welcome benefit
that the contents of slow spectrograms are preserved when exiting and restarting
the program.
This is probably not all new information but
perhaps useful to some, so I thought I would post it here as a
hint. Please note that the buffer size
shown in the screenshot (16 MB for a 4096 FFT) is
peculiar to my colour-DF display, a normal magnitude-only
spectrogram should probably require only half of that.
Thanks once more to Wolf DL4YHF for the excellent and continuous work!
I believe that the buffer layout had actually been straightened out in one
of the recent versions - in any case with the latest V2.78 b23 it
seems to work very well for me.
Best 73,
Markus (DF6NM)
|
