Has anybody ever tried compiling POV to run on, say, a Cray or
other really high-end supercomputer?  If so, how complex of a
scene could be rendered in a space of 24 hours, or a week, or
similar times used for complex scenes on PCs?

-- 
Tim Cook
http://empyrean.scifi-fantasy.com
mirror: http://personal.lig.bellsouth.net/lig/z/9/z993126

-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GFA dpu- s: a?-- C++(++++) U P? L E--- W++(+++)>$
N++ o? K- w(+) O? M-(--) V? PS+(+++) PE(--) Y(--)
PGP-(--) t* 5++>+++++ X+ R* tv+ b++(+++) DI
D++(---) G(++) e*>++ h+ !r--- !y--
------END GEEK CODE BLOCK------

Post a reply to this message

In article <3d9effab$1@news.povray.org> , "Timothy R. Cook" 
<tim### [at] scifi-fantasycom> wrote:

> Has anybody ever tried compiling POV to run on, say, a Cray or

A Cray, which it over ten years outdated, will not perform much better than
your old 486.  It is designed to process vectors* in order to process
matrices and similar structures which appear i.e. in weather forecasting.
If the data is not linear it will not perform well at all (in a sense it is
not an efficient general purpose computer).

> other really high-end supercomputer?  If so, how complex of a
> scene could be rendered in a space of 24 hours, or a week, or
> similar times used for complex scenes on PCs?

No, all systems you consider "supercomputer" are just massive parallel
systems and most (but not all) are build out of regular high-end processors
(DEC Alpha, SGI MIPS, IBM Power).  The other systems around are those from
NEC and Hitachi, which are based on vector processors.  Oh, and then there
is of course the Sun Sparc, which is frequently found in database servers of
large companies.  But no matter what processor these systems are based on,
they are all massive parallel systems, usually with distributed (and locally
shared) memory subsystems.  A good, but as usual not current (because data
in the list is on average six month old) overview can be found at
<http://www.top500.org/list/2002/06/>

The "problems" remain the same as for building your own local cluster.  The
distribution of global data (i.e. radiosity).  Of course, you can run
POV-Ray on such systems, but they won't change the limitations.


    Thorsten


* A "vector" in this context means one with about a few hundred to a few
thousand components.


____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde

Visit POV-Ray on the web: http://mac.povray.org

Post a reply to this message

"Timothy R. Cook" <tim### [at] scifi-fantasycom> wrote in message
news:3d9effab$1@news.povray.org...
> Has anybody ever tried compiling POV to run on, say, a Cray or
> other really high-end supercomputer?  If so, how complex of a
> scene could be rendered in a space of 24 hours, or a week, or
> similar times used for complex scenes on PCs?

It depends on what you think of as a supercomputer. Classical
vector-processing supercomputers will be of little use... however
clusters, such as Beowulf, etc, could be most useful. Imagine
dispatching each of your 1024 lines of a 1280x1024 pictures to
a distinct moderately powerful computer. 1024 p3 1 GHz do
pack a good punch :)


Years agos, on a series of 4 small machines ( p2, 333 Mhz
with 128 M memory), the gain was nearly four-fold. A script
called the linux povs with a region of the screen, got them to
do the renders, then piped 4 differents images in another
program that reassembled them to form only one. I supposed
that generalized to a great many computer, rendering would
be very nice indeed.'


Best,


    S.


>
> --
> Tim Cook
> http://empyrean.scifi-fantasy.com
> mirror: http://personal.lig.bellsouth.net/lig/z/9/z993126
>
> -----BEGIN GEEK CODE BLOCK-----
> Version: 3.12
> GFA dpu- s: a?-- C++(++++) U P? L E--- W++(+++)>$
> N++ o? K- w(+) O? M-(--) V? PS+(+++) PE(--) Y(--)
> PGP-(--) t* 5++>+++++ X+ R* tv+ b++(+++) DI
> D++(---) G(++) e*>++ h+ !r--- !y--
> ------END GEEK CODE BLOCK------
>

Post a reply to this message

I seem to recall one of the "ancient ruins" IRTC stills was claimed to be
done on a Cray 3, and the artist claimed the render time was 1 second.

Post a reply to this message

There are raytracing programs for the Crays, but they're not POV...
One of my former employers moved into an office that was being
vacated by Cray, and I got a short stack of 'Cray Times'
(an internal monthly newsletter) out of the throw-away pile.
On the back of each was a picture done on a cray and many were
raytraced.  One of them I recall was basically all shiny reflecting
metal, and it had rendered in I think 17 sec.  Fastest thing
I had access to at the time was a 32 MIP VAX 6340 (4 processor
VAX) or a Compaq 386-20e.  My homebrew Pascal raytracer
(before I'd found POV) ran one scene of comperable complexity
in 80 HOURS on the 386...
  I do have an DEC Alpha 500 at my disposal - I've been wondering
how that would stack up to a P4 1.7GHz...

Emory

Justin Smith wrote:
> 
> I seem to recall one of the "ancient ruins" IRTC stills was claimed to be
> done on a Cray 3, and the artist claimed the render time was 1 second.

Post a reply to this message

"Timothy R. Cook" <tim### [at] scifi-fantasycom> wrote in 
news:3d9effab$1@news.povray.org

> Has anybody ever tried compiling POV to run on, say, a Cray or
> other really high-end supercomputer?  If so, how complex of a
> scene could be rendered in a space of 24 hours, or a week, or
> similar times used for complex scenes on PCs?

btw - my dream (impossible now, hmm well maybe Bill G can affort it;) - to 
have cluster of i.e 1024 x 100 Pentiums 2 GHz 2 GB ram connected with very 
fast ethernet, so each will rander just ~7 pixels per frame. Then it would 
be possible to play Quake 3 with real soft shadows, real glass, radiosity, 
media :))) maybe even a focal blur - in real time :)

-- 
#macro g(U,V)(.4*abs(sin(9*sqrt(pow(x-U,2)+pow(y-V,2))))*pow(1-min(1,(sqrt(
pow(x-U,2)+pow(y-V,2))*.3)),2)+.9)#end#macro p(c)#if(c>1)#local l=mod(c,100
);g(2*div(l,10)-8,2*mod(l,10)-8)*p(div(c,100))#else 1#end#end light_source{
y 2}sphere{z*20 9pigment{function{p(26252423)*p(36455644)*p(66656463)}}}//M

Post a reply to this message

Rafal 'Raf256' Maj wrote:
> be possible to play Quake 3 with real soft shadows, real glass, radiosity, 
> media :))) maybe even a focal blur - in real time :)

"Say, what's that blur over there?  Where's my glasses?  OW!"

-- 
Tim Cook
http://empyrean.scifi-fantasy.com
mirror: http://personal.lig.bellsouth.net/lig/z/9/z993126

-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GFA dpu- s: a?-- C++(++++) U P? L E--- W++(+++)>$
N++ o? K- w(+) O? M-(--) V? PS+(+++) PE(--) Y(--)
PGP-(--) t* 5++>+++++ X+ R* tv+ b++(+++) DI
D++(---) G(++) e*>++ h+ !r--- !y--
------END GEEK CODE BLOCK------

Post a reply to this message

Rafal 'Raf256' Maj wrote:
> btw - my dream (impossible now, hmm well maybe Bill G can affort it;)
> - to have cluster of i.e 1024 x 100 Pentiums 2 GHz 2 GB ram connected
> with very fast ethernet, so each will rander just ~7 pixels per
> frame. Then it would be possible to play Quake 3 with real soft
> shadows, real glass, radiosity, media :))) maybe even a focal blur -
> in real time :)

And it generate enough heat to provide hot water on tap to the whole
street - perhaps a method of funding the running costs :)

--
Rick

Kitty5 NewMedia http://Kitty5.co.uk
POV-Ray News & Resources http://Povray.co.uk
TEL : +44 (01270) 501101 - FAX : +44 (01270) 251105 - ICQ : 15776037

PGP Public Key
http://pgpkeys.mit.edu:11371/pks/lookup?op=get&search=0x231E1CEA



---

Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.394 / Virus Database: 224 - Release Date: 03/10/2002

Post a reply to this message

Rick [Kitty5] wrote:
> And it generate enough heat to provide hot water on tap to the whole
> street - perhaps a method of funding the running costs :)

Or generates enough heat to power itself...perpetual motion!

-- 
Tim Cook
http://empyrean.scifi-fantasy.com
mirror: http://personal.lig.bellsouth.net/lig/z/9/z993126

-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GFA dpu- s: a?-- C++(++++) U P? L E--- W++(+++)>$
N++ o? K- w(+) O? M-(--) V? PS+(+++) PE(--) Y(--)
PGP-(--) t* 5++>+++++ X+ R* tv+ b++(+++) DI
D++(---) G(++) e*>++ h+ !r--- !y--
------END GEEK CODE BLOCK------

Post a reply to this message

On an only slightly less far fetched note:

I was looking at new computer parts for what I hope will be a new computer
by X-mas.  I realized that I ArsTechnics's recomdation for a budget
motherboard, the Asus A7N266-VM,  has built in nic and video for ~$75
assuming this nic could be convenced to remote boot, $150 worth of processor
and $300 worth of ram give a perfectly good render "box" for under $500.
With the distributed computing patch I've read about, and some left over
cases (we all have lots of left over cases right?) one could create a
POV-Ray rendering farm for less then the cost of a compact car.

JFMILLER


"Rafal 'Raf256' Maj" <raf### [at] raf256com> wrote in message
news:Xns### [at] 204213191226...
> "Timothy R. Cook" <tim### [at] scifi-fantasycom> wrote in
> news:3d9effab$1@news.povray.org
>
> > Has anybody ever tried compiling POV to run on, say, a Cray or
> > other really high-end supercomputer?  If so, how complex of a
> > scene could be rendered in a space of 24 hours, or a week, or
> > similar times used for complex scenes on PCs?
>
> btw - my dream (impossible now, hmm well maybe Bill G can affort it;) - to
> have cluster of i.e 1024 x 100 Pentiums 2 GHz 2 GB ram connected with very
> fast ethernet, so each will rander just ~7 pixels per frame. Then it would
> be possible to play Quake 3 with real soft shadows, real glass, radiosity,
> media :))) maybe even a focal blur - in real time :)
>
> --
> #macro
g(U,V)(.4*abs(sin(9*sqrt(pow(x-U,2)+pow(y-V,2))))*pow(1-min(1,(sqrt(
> pow(x-U,2)+pow(y-V,2))*.3)),2)+.9)#end#macro p(c)#if(c>1)#local
l=mod(c,100
> );g(2*div(l,10)-8,2*mod(l,10)-8)*p(div(c,100))#else 1#end#end
light_source{
> y 2}sphere{z*20
9pigment{function{p(26252423)*p(36455644)*p(66656463)}}}//M

Post a reply to this message