On Tue, Jan 3, 2012 at 1:55 AM, Joerg Wunsch <[email protected]> wrote:
> Xiaofan Chen <[email protected]> wrote:
>
>>> Does AVRDude have any facilities to do this? Does it already
>>> understand how to use D2XX or does it not include that
>>> functionality?
>
>> I do not think the mainline codes do not support d2xx. There are
>> patches though but none of them have been integrated.
>
> And frankly: they won't ever be.  One of the hosting requirements on
> savannah.nongnu.org is that only "free software" (in the FSF's
> terminology, and by their definition of "free software") must be
> provided there.  Linking against D2XX would not satisfy that
> condition.  (Linking against libftdi is, of course, OK, as it is free
> software by the FSF's definition.)

I can totally understand the decision against D2XX. OpenOCD
supports both libftdi and D2XX since D2XX used to offer much
better performance. A long discussion later came to the conclusion
that it is against GPL to distribute OpenOCD binaries linking
with D2XX.

And I did some benchmark, D2XX does not have any performance
advantages against libftdi for OpenOCD.
http://lists.berlios.de/pipermail/openocd-development/2011-July/020231.html
(more in the full thread).

> But sorry, nobody so far taught AVRDUDE about these details.
> "avrftdi" is a bit-bang programmer using the MPSSE's native bit-bang
> mode (on top of libftdi).

MPSSE is probably the best (for JTAG and SPI at least)
in terms of the bit-bang. That is why OpenOCD only officially
supports FTDI chips with MPSSE features (FT2232D,
FT2232H, 4232H, 232H) and not those without (FT232R).

There is a patch to use FT232R for OpenOCD but the
author says that MPSSE is 90 times faster.
http://forum.sparkfun.com/viewtopic.php?f=18&t=21411

-- 
Xiaofan

_______________________________________________
AVR-chat mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/avr-chat

Reply via email to