Subj : Sio vs Sio2k
To   : Mike Luther
From : Torsten Bamberg
Date : Sat Apr 16 2005 02:00 am

Hello Mike!

on Friday, 15. April 2005 14:21, Mike Luther wrote to me:

 ML> I think I see what you are saying.
Well, sorry for my english. My german is much better. ;-)

 ML> That by specifying that the IRQ is
 ML> hard coded in the setup line for IRQ3 and IRQ4 like that, you suggest that
 ML> is where the conflict arises?  As I understood his post, he had only ONE
 ML> actual hard cable connected modem in this setup.  That was the COM1
 ML> device.  It normally is on IRQ4 and has to be known as that way to SIO or
 ML> SIO2K per what I think I understand.
Me either.

 ML> I, in fact, use a four port hard coded serial board on this BBS box here.
 ML> While it is an ISA-type card, and it has the ability to define and set up
 ML> each COMM port with a different IRQ specified and that is the way I used
 ML> it, it also *CAN*, share the same IRQ for all the ports, even in an ISA
 ML> slot.  That because of special design and driver work. However, I don't do
 ML> that common IRQ driver operation.  I use separate IRQ's for each port.  As
 ML> has been posted for a long time, OS/2 can share the same IRQ for different
 ML> driver purposes.  However that is dependent on the fact that the drivers
 ML> which do this stunt have all been coded very carefully to allow this.  You
 ML> saw this heavily on the old COMPAQ computer systems which were shipped
 ML> with OS/2 long ago and also were shipped with custome made Adaptec SCSI
 ML> controller cards, as well as hard disk drives that were carefully
 ML> specified to be compatible with these common single IRQ use techniques.
Well, I've tried to use all 4 buildin Com-Ports with modems, and on some 
mainboards it does work, on others not. Sometimes there are strange timing 
errors, sometimes I couldn't reach com3 and com4. Aehm, I usually have european 
Boards, witch are produced in china or taiwan (gigabyte, msi, intel)

 ML> I never caught that conflict IRQ use in his post.  But SIO and SIO2K both
 ML> don't normally care about merged IRQ assignments for multiple internet
 ML> ports, as I understand this.  Thus, here are the two different CONFIG.SYS
 ML> lines for my system here with four hardware ports and the additional
 ML> internet ports that work just fine here:

 ML>   DEVICE=C:\SIO\SIO.SYS (COM1:57600,03F8,4,-) (COM2:57600,02F8,3,-)
 ML>   (COM3:9600,03E8,15,-) (COM4,02E8,12,-) (COM5,INTERNET:02A8,11)
 ML>   (COM6,INTERNET:02B8,11)

 ML> Obviously that above is all on one line.  Or in the case of SIO2K here:

 ML>   device=C:\sio2k\uart.sys logfile=C:\sio2k\sio2k.log
 ML>   device=C:\sio2k\vsio2k.sys logfile=C:\sio2k\vsio2k.log
 ML>   vIrqList(1=4,2=3,3=15,4=12)
 ML>   device=C:\sio2k\vx00.sys
 ML>   device=C:\sio2k\vmodem.sys logfile=C:\sio2k\sio2k.log NoPause nPorts=1
 ML>   device=C:\sio2k\sio2k.sys logfile=C:\sio2k\sio2k.log

 ML> I don't use a PS/2 style mouse.  It is a serial port mouse that is fed to
 ML> my beloved ROSE rack mount KVM switch operation.  Thus in this case the
 ML> IRQ12 is available for that COM4 use above which otherwise would be in
 ML> conflict with the PS/2 type mouse.

 ML> Notice the two internet ports configured in that COM5 and COM6 SIO
 ML> operation above that use different ports but the same IRQ?  As far as I
 ML> know the software simulated ports in SIO and SIO2K can handle the internet
 ML> port creation stuff just fine with a common IRQ specified.  But I don't
 ML> assign that to an IRQ which is used elsewhere in hard coded ISA or
 ML> hardware board level COMM port work. And I do not use any of the normal
 ML> communications ports for an internet port with SIO either,
Ok, I see. As I understood, he has theese specific problems on com1, 3f8, irq4.
He mapped the virtual comports to com2/com3/com4.
ok.
Well, my expirence is, that sometimes the timing-options of the builtin fifo 
chips are very worse, or/and speed down the hole system. I found out, that 
especially via-chipsets having a very worse timing on comports.
Just, after I disabled all buildin comports, and using a multi-comport-card, 
these problems have been gone.


 ML> Mike @ 1:117/3001
By/2 Torsten

--- GoldED/2 3.0.1
 * Origin: DatenBahn BBS Wismar, Germany (2:240/5832)

.