Page 1 of 1

serial throughput

Posted: Sun Oct 28, 2012 7:37 pm
by YahooArchive
> Another question; how much data can I transfer to ARMmite through
USB port?
> 1,5 Mbyte? 1 ? 500KByte? 250?

I have run the ARMmite over USB at 921.6Kb. The FTDI driver seems to
limit that to common serial baud rates, though I have not investigated
it much.

While the ARMexpress is limited to 38.4K maybe 56K when running over
RS-232, I have someone experimenting with running it faster when using
a short (to the FT232 chip) connection. It may be possible to allow
that to run up to 115Kb.

If your question truly is how much data can you send, there is no
limit if the PC is actually reading the data, there is probably a
limit in the size of the buffer before it starts overflowing on the PC
side if you are not reading the data. I don't know how big that is,
but from experience I'd say its reasonably large (like many KB,
probably not MB)

Re: serial throughput

Posted: Sun Oct 28, 2012 7:38 pm
by YahooArchive
--- In ARMexpress@yahoogroups.com, "basicnode" <bruce@...> wrote:
> ... I have someone experimenting with running it faster when using
> a short (to the FT232 chip) connection. It may be possible to allow
> that to run up to 115Kb.

I have successfully ran the ARMexpress at those rates (115,200
[117,600 ] baud), with Bruce's assistance, in a manner as described
(FTDI USB UART direct to ARMexpress via 7404 hex inverter (3/6's used)).

Theoretically, the speeds could be pushed even further, but I am
skeptical as to how much faster it will run, until I test it further.

The math indicates that I can hit the following speeds provided that
the silicon can keep up. I will attempt to do so over the weekend:
235,200, 470,400, 940,800, and 1,881,600 baud.

I'll chime back in here to advise what the results of my testing are.

Until then...

-t

Re: serial throughput

Posted: Sun Oct 28, 2012 7:38 pm
by YahooArchive
> Theoretically, the speeds could be pushed even further, but I am
> skeptical as to how much faster it will run, until I test it
further.
>
> The math indicates that I can hit the following speeds provided that
> the silicon can keep up. I will attempt to do so over the weekend:
> 235,200, 470,400, 940,800, and 1,881,600 baud.

Ok, I couldn't wait 'til tomorrow, so I messed around with it a bit
this evening. I got reliable comms up to 235200 working with
BASICtools. I had to edit TCLterm.tcl a bit and add the lines for
the bauds that I wanted (117600, 235200, 470400, etc.

I then uploaded patched firmware and viola, I was able to get 235200
running. I tried to patch further for 470400, but it didn't work. I
see garbage when the ARMexpress is talking, so I am thinking that I
may be approaching the threshold of how quick the target can keep up
(with the RS-232 level translators in the mix).

At any rate, I wanted to pass along the news of success to this
point. I am able to flash a 35K user app in about 8 seconds. Pretty
darn quick considering how long it takes at 19200 to flash the same
file (much too long IMHO [yes, I am impatient :-)]...).

Take care, folks. Have a great weekend.

-t