Yeah, I wouldn’t want to write a soft UART if I could avoid it. :)

Thanks for the info!

On Sat, Apr 25, 2020 at 7:47 PM Stephen Adolph <twospru...@gmail.com> wrote:

> Model 100 UART is able to clock at 9600 19200 38400 and in fact also 76800.
> Model 100 is very hard pressed to use the existing rom routnes to utilize
> 38400.  There's just a lot to do to process input bytes using the circular
> buffer etc.  I myself have passed characters between two m100 at 76800 by
> manually coding, but It was flakey because of how you have to set up the
> UART.
>
> Combined with the fact that the fastest common rates on a PC is 19200 then
> 38400 then 57600, the internal UART cannot run 57600.  So for PC file
> transfers, the M100 is limited to 38400 at best (unless using BCR at 57600).
>
> Also, it is not possible to take over the serial port interrupt. It is
> hard coded with no hook.  So, to use the UART fast you need to poll it.  I
> suspect that makes it more or less like what I have done here....
> dedicating the CPU to the read in process.  But it would be at best slower
> than the BCR port.
>
> If anyone is interested I can post the code on the wiki.  It was fairly
> difficult to get it to work.  I really had to manipulate it.
>
> On Sat, Apr 25, 2020 at 10:27 PM Tom Wilson <wilso...@gmail.com> wrote:
>
>> I’m curious... why is this faster than the UART feeding the real serial
>> port? Bit banged I/O should always be 1/10 the speed of a UART, unless the
>> UART is broken or something.
>>
>> On Sat, Apr 25, 2020 at 7:25 PM Stephen Adolph <twospru...@gmail.com>
>> wrote:
>>
>>> Still playing around with the BCR TTL serial port.
>>> ...yes it has it's limitations... half duplex, no flow control,
>>> bit-banged
>>>
>>> ..but it is fast!  Today I was able to get 32kB transferred in 7
>>> seconds.  That's pretty close to 100% utilization on the 57600 baud line.
>>>
>>> I think that is a new M100 record.  It might need independent
>>> verification!
>>>
>>> (why?  to reduce time for CP/M disk backups.  4MB will take a 1/2 hour
>>> with this approach, or more than an hour at 19200 baud.)
>>>
>>> Steve
>>>
>> --
>> Tom Wilson
>> wilso...@gmail.com
>> (619)940-6311
>> K6ABZ
>>
>

Reply via email to