I asked Konstantin Belousov <kib at FreeBSD.org> how much time he spent
On Wed, Dec 19, 2012 at 6:12 AM, Martin <martin.kelly4000%gmail.com@localhost
> Can i ask out of curiosity the donation amount that would be needed to get
> project rolling.
for implementing GEM support for FreeBSD. AFAIK he worked full time on
In total he spent ~6 manmonths including VM improvements.
- 0.5 manmonths - RTFM
- 3 manmonths - GEM (+ improvements in VM)
- 1.5 manmonths - KMS(including infrastructural things like
EDID parser) + DRM + hacking i2c
- 1 manmonth - debugging
He also estimated time for implementing support TTM + radeon cards as
another 6 manmonths provided that support for Intel/GEM is already
Unfortunately there is no information for nouveau(4) (nvidia(4) driver
is available on FreeBSD).
All in all, we have aproximately 12 manmonths of full time job.
To me this looks very challenging but not that hard as
improvements for SMP NetBSD did several years ago.
Personally, I like this idea. I thought about the same many times.
> Just an idea... but maybe it would be worthwhile considering
> sub-categorizing the donations into say the top 5 major projects (like
> GEM+KSM) that the NetBSD community is really wanting support for.
> You could advertise the target amount for each major project and allow the
> NetBSD community to donate to that particular fund.
> The way i see it is that by providing the choice and what people are
> donating for you are more likely to get the support. Rather than people just
> donating without any idea of where there money is going.
> The other benefit it provides is feedback for what the NetBSD community
> considers important and a sense of contribution for non-developers and the
The problem here I think is that NetBSD is a community-driven project.
There is no way to force anyone for doing some kind of work if he/she
doesn't want to do so. A lot of developers think NetBSD should not
fight with Linux for desktops and target NetBSD more as a server platform
than a workstation. This is why NetBSD has more or less scalable SMP,
good support for Xen, mpls, scalable npf and so on.
This is why further SMP improvements are funded projects.
So, the problem #1 is to find a developer who is willing to work
on support of modern GPUs.
On the other hand, there is a permanent demand to improve NetBSD as
a workstation both from users and developers. Several years ago
Jeremy Reed proposed to improve lxde as a DE for NetBSD.
Being rather ascetic in GUI I'm not interested in any DE but
I think Jeremy's project is very important for NetBSD.
Obviously, his project just cannot succeed without support
for modern graphic cards.