CUDA under FreeBSD

[Valeri Galtsev]

On Tue, December 5, 2017 4:04 am, Carmel NY wrote:
> On Tuesday, December 5, 2017 2:53 AM, Shane Ambler stated:
>> On 04/12/2017 21:19, galtsev at kicp.uchicago.edu wrote:
>> > On Mon, December 4, 2017 4:24 am, Carmel NY wrote:
>> >> Out of morbid curiosity, I was wondering if anyone could tell me the
real reason that Nvidia does not support CUDA under
>> >
>> > Arrogance would be my guess.
>> The morbid part is that they give us the linux libcuda, so we should be
able to
>> run linux binaries that use cuda, just not native apps.
>> >> FreeBSD? Also, what are the realistic expectations for it getting
supported shortly?
>> >
>> > Zero is my estimate. The way to let one's steam about them is just
not
>> > to buy ther hardware. Their attitude to open sourse and unwillingness
to disclose details of their hardware was always much worse than that
of their competitors (ATI/AMD, matrox...).
>> >
>> > This is just my opinion based on my subjective observations.
>
> Personally, I have always considered Nvidia products to be superior.

Well, this is just your subjective opinion opposing my subjective opinion.
With all due respect.

I have seen nvidia chips giving artifacts (probably after mild overheat,
really just mild). I can not compare some product (to consider it superior
to another) if I can not use that product fully under variety of systems
_I_ use. And the last is true about nvidia video chips. As opposed to
variety of their competitors.

At some point Apple agreed with me (well, of course independent on me made
up their opinion ;-). There were infamous MacBook Pro 15 inch made by
Apple somewhere around 2012. These contained discrete video chip by NVIDIA
(in addition to integrated on intel CPU substrate... I'm lying, it was
inside CPU case, but etched on different substrate...). Anyway, there was
some crap about that NVIDIA chip, so Apple didn't manage to make later
releases of MacOS work with later hardware and with that 2012 MacBook Pro,
kernel just crashed inside NVIDIA kernel module. Apple even had (really
short lived) program of replacing that hardware, realizing that this is
just crap. Program closed very quickly, so only small portion of bad
hardware was actually replaced. My guess is: nvidia decided not to carry
their side of financial losses. After which Apple made good IMHO decision,
and switched over to AMD (which are actually bought out by AMD well known
ATI). Incidentally, way back someone made excellent argument when
comparing ATI with NVIDIA. Here it is: NVIDIA releases new drivers (or
driver updates) almost monthly. ATI takes about half a year to release
driver. From which the conclusion can be made (which I fully agree with)
that ATI thoroughly tests and debugs drivers before releasing them (and
doesn't need to fix crap in the driver soon after release). Not true about
NVIDIA, whose drivers quite likely are much buggier.

Anyway, just my observations, potentially a bit biased by the fact that
NVIDIA discloses much less about chip internals than, say, ATI (hence the
ability or lack of such by open source driver developers to write decent
open source drivers.

>
>> I'm sure on an episode of bsdnow, they mentioned asking an nvidia dev
at
>> one of the conferences and they said there shouldn't be any technical
reason, it just isn't enabled in the build and they would look into it.
Still hasn't helped any.
>

Which just confirms that individual developer is likely less arrogant than
a company as a whole.

Well, all I said is just my subjective opinion based on my (by no means
thorough) observations.

Valeri

> Interesting. I was not aware of that. It would seem to me that there
should be more of an
> concerted effort to get this issue resolved.
>
> --
> Carmel
>


++++++++++++++++++++++++++++++++++++++++
Valeri Galtsev
Sr System Administrator
Department of Astronomy and Astrophysics
Kavli Institute for Cosmological Physics
University of Chicago
Phone: 773-702-4247
++++++++++++++++++++++++++++++++++++++++