OK we had a followup discussion on the #milkymist channel
on Freenode. Here are my 2 cents:

- deliver as many as possible software updates to existing
and future Milkymist One customers. Too many to list here,
I think we all agree that M1 today just scratches on the surface
of what it could do.

- make rc4 the best possible product, hopefully we can switch
to dvi-i for example. Improve yield, improve sourcing (boom?).
Yield on rc3 with 3 months of testing and fixing is currently
standing at 71 out of 90. Improving the yield will give us much
better results than changing components costing a few USD.
Here's what we learnt from rc3 so far, work is ongoing every day
http://en.qi-hardware.com/wiki/Milkymist_One_RC3_Known_Issues
And for the record, the RC3 bom is here
http://en.qi-hardware.com/wiki/Milkymist_One_RC3_BOM
It seems people don't know this URL, although you just need to
type "Milkymist One bom" in google and it's right at the top...

- for pricing strategy, I believe we should keep the top
Milkymist One product at 499 USD, for the time being. At least
for 12 months. The more aggressively we are able to improve the
hardware, the easier it will be for us to offer an older hardware
version, or even just pcba spinoff, at a much reduced price,
without destroying the value and pricepoint of our top product.
Milkymist One today delivers unbelievable value at 499 USD, it's
almost too cheap. You won't find something like this anywhere
else in the world. I do not believe blindly reducing the price
helps, we need to communicate the value of the product better,
improve documentation, improve ways for our users to share their
experiences, etc.

- I am interested in Artix-7 for a while, since this is an improved
version of a key chip on Milkymist One of course we should evaluate
it. My #1 question for Artix right now is what boot options it offers,
and whether there are ways for us to remove the NOR chip.

- If we make a cheap entry-level Milkymist board, we should try to
make PCBA-only spinoffs of the regular Milkymist One, even leaving
chips or connectors off if possible. I want to focus my sales on
regular users, so we need someone who would want to market and sell
such a board. Maybe Bearstech could be interested, they indicated
that in the past.

- I am hoping that we can put more thought into the M1 expansion
header and expansion boards. The Spartan-6 is a powerful chip, we
should provide a reliable expansion interface for daughterboards,
and use this as a staging area for Milkymist development. People can
hack on all sorts of things there, theoretically, ideas can be tested...

- Milkymist One was really just launched as a product a few weeks ago,
and we already sold over 35 of 70 currently available units. We need
to continue down exactly this same path. Make the product better, find
new customers, find quality distributors, explain the story better,
provide better ways to give feedback and share experiences, make people
understand and love the box. I'm curious about Simon's feedback next.

I understand that one problem Sebastien sees is that he wants to design
something around Artix-7, but found it risky and difficult to do so on
Milkymist One. That's indeed a problem because two likable goals seem
to collide. Maybe Sebastien is off to do an Artix-7 evaluation board
for a while? Maybe we find ways to integrate this into Milkymist One
later, maybe not? I don't know. That's how it is with highly innovative
(and even open!) projects like Milkymist One, sometimes you just go in
a certain direction without fully understanding the environment :)

At least it gives us some time to think...
Cheers,
Wolfgang

P.S.: And in the meantime, if you don't have a Milkymist One yet, really,
get one. It will make you think too :-)
https://sharism.cc/milkymist
_______________________________________________
http://lists.milkymist.org/listinfo.cgi/devel-milkymist.org
IRC: #milkymist@Freenode

Reply via email to