Moin Timothy!
On Thu, 6 Dec 2012 10:23:28 -0500
Timothy Normand Miller <[email protected]> wrote:
> Besides OGD1, another example I like to point out is the lack of Free logic
> synthesis software, competing with the likes of Mentor Graphics, Cadence,
> and Synopsys. The main reason it doesn't exist is that the skills required
> are very specialized, and none of the people with the skills are interested
> in volunteering to write a free version. We could GET a free version if we
> hired these people and paid them to code it, but where would we get that
> money? The usual route is to recoup the investment by making the software
> proprietary and/or patenting some of the technology. A socialized
> alternative like Kickstarter wouldn't work because too few people would be
> interested in the product, so it would never get funded. There's a good
> reason why a Synopsys seat costs somewhere around $200K.
I actually think it is possible to get such a software to be written
as OSS. But i am not sure when it can happen. The problem here is that
you need very specific knowledge and skills (as you said) and there are
very few people who have that. As only a very small percentage of people
do work on OSS in their free time, the probability of getting enough
people with the right skills is low. And you need a damn lot of people
to get something like this of the ground.
But something interesting is going on out there: Those who have been
watching the internets in the last couple of years have for sure noticed
that there are more and more hardware related projects out there.
Not only does the whole 3D printer "hype" get a lot of attention, also
a lot of electronics projects (like whole arduino movement) are going on.
And as a spill over projects that have been in existance since the dawn
of time, but never got much exposure because they are way to specialized
(like opencores) are getting more and more contributions.
If this continues (and i'm quite sure it will) more an more people will
get enough skills and interest to actually work on something like a
synthesiser. And IMHO the slowly increasing number of such projects is a
proof of that. I dont know how long it will take until one of those is
getting to the critical mass to be actually successfull, but it should
not take forever anymore.
>
> BTW, if you have no idea what goes into logic synthesis software, don't
> bother arguing with me. For one thing, the problem is NP-hard. That alone
> make is extremely difficult.
Do i qualify? :-)
Attila Kinali
PS: NP-hard doesn't mean a lot to CS people in this context, as compilers
are NP-hard as well, but are fairly simple to build when compared to
logic synthesisers. What you have to mention here is that a compiler can
choose one of all working solutions, optimize it a bit and will be
considered as working correctly, if a bit slow. A synthesiser has not
only a larger number of possible solutions, these solutions can often
not be transformed into each other (eg different adder structures).
And not meeting the timing requirements will constitute in a failure of
the circuit, even if it would work with a slower clock.
--
There is no secret ingredient
-- Po, Kung Fu Panda
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
