MIPS vs. patents (was Re: Farewell)

Werner Almesberger werner at almesberger.net
Tue Jun 14 19:33:22 EDT 2011


Jane Andreas wrote:
> [RMS] tells me that the video playback [on his MIPS] sucks!

Hah ! Meanwhile, the Ben can do full-screen playback, thanks to
the nice work of David Kuehling :)

> Imagine! So not only is MIPS technically inferior, it also is no less
> good when it comes to patents as Sebastien pointed out!

I think what Sebastien wrote didn't came across quite correctly.
Here's how I see the situation:

The MIPS architecture per se is old enough that all the original
patents should have expired.

However, Chinese companies producing MIPS clones based on the
expectation of avoiding royalties this way have eventually still
signed licensing deals with MIPS. As usual, the content of these
contracts and the motivation for signing them is kept secret.

So we don't know if MIPS found some loophole to still make some of
their patents apply, whether this is part of a package including
more recent and very valuable technology for a good price, or
whether MIPS is actually paying them money just to be able to
intimidate other prospective MIPS cloners with the statement that
everybody who makes anything MIPS-ish is a MIPS licensee.

So if you were to make your own MIPS clone, and you were careful to
stick with their pre-1991 architecture, it's not clear whether
you'd have to expect serious trouble from MIPS or not. The question
is: can you afford the risk ?

Now, none of this matters all too much in any practical sense if
you don't plan to make your own chip anyway, or at least know
someone who will. You may still want to be careful about whose
architecture you promote and support, but other concerns will have
a far more direct impact on the freedoms you experience in real
life.

I would view the Ben as a starting point. Parents normally don't
abandon their newborns because they're wrinkled, too small, and
can't even speak a coherent sentence. Nor should the Ben's numerous
shortcomings be considered strategic failures.

It would be a strategic failure to become entrenched around some
proprietary feature set you can have no hope to ever free or
replace. We're not tryiong to do this on the Ben. There is very
little code what would need more than recompiling to run on some
other platform. Even most of the hardware hacks could be adapted to
some other SoC within a few hours or days.

I you look at my freedom achievements chart [1], you'll see that
the Ben has many red dots. The CPU isn't the only problem area. You
won't have a device that you can consider fully Free until all of
these red and yellow dots have been turned green.

[1] http://downloads.qi-hardware.com/people/werner/tmp/achieve.pdf

Things are actually moving on the CPU front, namely in the form of
the Milkymist. But it will be some time until we can even tell
whether the work there will be able to result in a competitive CPU
for mobile devices.

There are other areas where the Ben falls short. One particularly
troubling area is the mechanical design. Given sufficient funding
today, I have little doubt that we could design and manufacture a
nice new NanoNote circuit that improves substantially beyond the
Ben, in terms of freedom and in terms of technology, within many
half a year from now. But I would be far less optimistic when it
comes to making the shell that goes around it.

Speaking of funding, let's not forget the elephant right in the
middle of the room - how will all our great plans get financed ?

- Werner




More information about the discussion mailing list


interactive