Paul G. Allen wrote:
I've written a rather large Java application for my company. It uses
MySQL and serial communications. Part of the application loads firmware
from a file on the PC into an embedded device over the serial ports. It
can support as many serial ports as the computer can support (currently
I've used it with 10). It runs a separate thread for each port. It runs
on Linux (specifically, I'm using FC4). The MySQL database is on a server.
The portion of the code used for doing the serial port communication was
taken from another application that performs a similar task. This other
application only supports a single serial port and only runs on Windows.
It seems the other application can load the firmware in a fraction of
the time that mine can, but the code is the same. I haven't yet been
able to try my code on a Windows box that works to see if it'll run
faster. I'm wondering if a Java on Linux might be slower than Java on
Windows? I'm also wondering if there is a JIT compiler in existence that
I can use to compile the Java app. into native code for Linux (there are
many for Windows). The JIT would have to support Sun Java 1.5. (Note:
Seems I can't find HotSpot for Java 1.5)
It also seems that my application takes far longer than it should to
load firmware depending upon the number of devices connected: loading
203KB at 115,200 on a single device takes 15 minutes, 10 devices takes
20 minutes).
I think you need to break up the problem a bit. First off would be check
what the serial port speed is set to when you start and when you finish.
The default serial port settings might be 9600-8-N-1 instead of the
115200 you think it is. I'm not even sure Java is allowed to change the
serial port settings without a tweak to the security manager.
Have you tried checking/setting the port speeds using something like
setserial before running your program?
Why not write a small test program that just reads the firmware file
from disk and dumps to null first, make sure that works, then write to
the serial port. That way you can check that that is really the problem
to begin with. It might turn out that you have too many threads running
and a synchronize problem instead.
Gus
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
