Hi Rob!

Thus quoth Rob Basler to Jaap Van.Veen:
JV>>As I have recalled earlier, my system (486DX2 66) is much faster
JV>> with FP40 installed. This has to do with the faster handling of the
JV>> harddisk.

JV>>However during the regular status checks the COM1 port is not
JV>> serviced in time causing overruns. Does somebody know how I can
JV>> avoid overrun of COM1.

 RB> Most likely, if you add /A:0 /U:0 /!BM to your IBM1S506 line your
 RB> problem (and the added speed) will go away.  This disables
 RB> busmastering which is on by default in the newer HD drivers and
 RB> which can cause serial problems. You might also try adding /!SMS to
 RB> the end of the line as well to make things even slower.  You can
 RB> use the /V switch to see what the driver is seeing for your HD's
 RB> and how they are being set up.  Note that there is at least one
 RB> version of IBM1S506 that ignores the !BM switch, if you try the
 RB> DANIS IDE driver, this problem is rectified.

This sounds like the infamous SMS issue. Older versions of IBM1S506 honoured
the /!SMS switch and sacrificed performance for COM stability. The later ones
suffered from a bug which lead to disregarding /!SMS. In my driver /!SMS
works, and even better: there is a /SMS:n option which gives you optimum
control on this parameter. People have reported failures with later IBM1S506
versions, but stable COM operation with my driver at SMS values of 4 or 8
(nearly optimum performance).

bye, Dani

--- Sqed/32 1.14/r01354
114/477
143/1
5100/8