>i can't find a 1984 intel microcontroller book (got volume 1, not
>volume 2 ;-(, but the zilog book for the 8530 (which the intel chip
>seems to be a clone of) indicates that it's possible to set the clock
>mode to x1, and i'm wondering if anyone has tried this and if it had 
>any effect ?

Yes, I tried it on a 5000/240. 

Output works.  Input loses. 

The clock multiplier affects how many samples the Intel
(certainly, possibly Zilog)   chip does on the input stream.

If it only samples once per bit, it doens't reliably find the start
bit, and you end up with crud, that's someimes but not always shifted
up one bit.

I don't recall what the hardware specs for the 5000/1xx or Personal
DECstations say, but they may actually specify 38400 as the top speed.

Even the 5000/240 cant run reliably above 57600. That's why I took the
code for x1 PCLK support out of the driver.  (well, actually, it does
work if you put the chip in to sync mode, say to use it for Localtalk;
but nobody has tried that ethier).


>alternatively, might it be possible to restructure the driver to
.improve throughput ?  perhaps i should just give up and buy an intel
>pc (<-- joke, for the humour impaired) !

-current has some througput fixes for the scc driver.
I'd recommend trying that and seeing if it helps.