Ahhh-
And you make my point, somewhat, for me, the monetary difference
between ~1000 Intel 32 bit boxen and an equivalent amount of SGI hardware
is daunting, to say the least. And I do agree with their motivations, but
my real sticking point here is this: if it is money, they should say so,
but no one wants to say, "our film was like any other, no one knows what'll
sell, so we went the cheapest route possible to get the best that we could
and crossed our fingers." Not many joints outside the gov. buy large scale
SGI hardware, and if they do, SGI won't tell you they did. I think in a
strict performance comparison, the same amount (~1000) of SGIs would
outstrip the Linux boxen, but that's theoretical for me, my allowance isn't
quite that high. But I will also say that the structure of the linux
cluster is also different than the SGI structure, SGI uses ccNUMA, this
allows for memory and cpu access across the entire cluster at any point in
time for any process, no shared memory limitations, etc. From what I can
tell from teh article, the clustering they are looking at resebles both
Mosix and something like SCore, out of Japan. the issue is that parallel
linux clusters have a limited scalability in comparison to SGI (I don't
remember numbers, something like 64 procs?), with Mosix and the sharing of
multiple CPUs you get out of that hole somewhat, but you dig yourself
another with the limitation on shared memory processes outside of the
originating node. This is well known and is a limitation I've hit, myself.
So, I say again, it's a monetary issue, no matter what "shading" is put on
it, and that's all I would like is the understanding that in comparison, in
some areas, pound for pound, cpu for cpu, proprietary things just work
better, no ifs ands or buts, they just do. And my reasoning is, I like
BSD, and the related areas, linux included, they are great technology, but
I see that there is an area where people become blinded by hopes and
beliefs and do not pay attention to the reality, it's a nice thing, because
sometimes that blindness affords them the will to overcome some things, but
sometimes it is unreasonable, and is like saying it is noon, when it is
midnight.
I understand the hope many hold for linux, I hold much hope as well,
for Open Source and for Free Software and such, but I will not step into
that sort of fight with anyone, not knowing the strengths and uses of what
some would term "the enemy". Sun Tzu had many a great quote for this, but
I am so used up, that I don't remember any of them, but I will say this,
please, I'm not putting linux down, I wouldn't do what I do without it, but
to properly bring it into the environment in which I work, I need to
understand what battles I can win and which are not for me to take. And
there are times, that I must say that linux will not work, and knowingly
wish taht I *could* use it, but know that either politically or
technically, that I cannot.
I hope I am understood, it's late, thank you,
mbutler
http://www.sgi.com/features/2001/july/fantasy/
http://www.arstechnica.com/wankerdesk/01q3/ff-interview/ff-interview-1.html
Also here, they used SGI hw for the workstations and servers, but the
rendering was done with Pixar tools in linux.
-- cut here --
"The 16-cpu Origin 2000s are primarily used for batch-processing MTOR
jobs, which is the Maya-To-Renderman conversion."
"The renderfarm consists primarily of ~1000 Linux machines (PIII,
custom-built, rack mounted), running Red Hat 6.2. These machines do all
the RenderMan renders, as well as a number of other tasks."
-- cut here --
I still dont mind that its a money move, any sort of use of linux helps
promote it in other uses.