Hello,
I started to make a rt-linux application a few month ago,
and it is running quite well (Sampling time of 25kHz for
DAQ of a single channel on a 486/33 using a ISR routine,
and later on display data with a GTK+ user interface).
It is astounding what a 486 is still good for under Linux.
A competitive OS refused to be installed on this machine,
thus: Ceterum censeo MIRCOSOFTem esse delendam!
However, I am not entirely satisfied with my architecture:
a kernel module having registered an RTirq puts captured
data into shared memory at a given frequency as unsigned
short int. From user space, these values are converted
into Volts and later on in other units. What bothers me
is that the conversion depends on DAQ board settings. I
would prefer a clearer interface between the DAQ board
specific functions and the rt-module such that they
already deliver floats representing Volts into shared
memory.
And now the questions: Since I do not explicitely use a
rt-task but moreless a module doing DAQ in a ISR, how do
I tell rt-linux to use_fp? In init_module? And do I have
to insmod the rtl_fpsched even if there are no rt-tasks?
Does the use of fp slow down the rt_linux performance
significantly?
Thanks,
peterw
--
Dr. Peter Wurmsdobler
CTM - Centre de Transfert des Microtechniques
39, av. de l'Observatoire, BP-1445, 25007 Besancon CEDEX 3
TELEPHONE: +33 3 81 47 70 20 TELECOPIE: +33 3 81 47 70 21
E-mail: [EMAIL PROTECTED]
Ceterum censeo MIRCOSOFTem esse delendam.
--- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
----
For more information on Real-Time Linux see:
http://www.rtlinux.org/~rtlinux/
