>>>>> "Rick" == Rick Kelly <rmk@rmkhome.rmkhome.com> writes:

Rick> David Gilbert said:
>> The problem is that the chips have a 3 byte FIFO.  Those three
>> bytes are easily exhausted.  Moreover, when I'm full bore receiving
>> with ppp, the system is hit hard --- close to unusable.
>> 
>> While I can't 'know' yet --- I don't have a Sun3 to play with yet,
>> I suspect that 19200 is good fishing for a Sun3.

Rick> When Sun switched from 68k to sparc, the decided that ring
Rick> buffers were a good thing as compared to clists.  Thus, a Sun
Rick> 3/60 had better serial throughput than a sparc 1+.

Rick> To this day, the serial driver in Solaris sucks.

	Does this mean that it's all a software issue?  I know that
the NetBSD uses ring buffers.  Would it be possible to rewrite it in
terms of a faster or more efficient structure?

	Serial data at 19200 looks like about 2 characters per
microsecond (at 10 bits per character, that's 1920 characters per
second) as a maximum rate.  Is it really too much to expect a one
milisecod reaction time to that interupt.

	While I'd like to finish support for the multi-port serial
card that I have for my 4/260, I'd also like to see the stock serial
ports working better.

	It seems from this discussion, that when I get my 3/260 (in a
little while), it will make a better uucp host that my 4/260, as it
currently stands.

Dave.

-- 
----------------------------------------------------------------------------
|David Gilbert, PCI, Richmond Hill, Ontario.  | Two things can only be     |
|Mail:      dgilbert@jaywon.pci.on.ca         |  equal if and only if they |
|http://www.pci.on.ca/~dgilbert               |   are precisely opposite.  |
---------------------------------------------------------GLO----------------