 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Now go on and tell me that I could save my work cause exactly that is
already avaiable :-) ?
If so, I'd be intrested to know where and how they work and if they are
free.
I did not touch the POV 3.5 EXE, cause I want my system to be compatible
with any future new POV-version.
Originally I hoped something like this could be
inside the POV-Shell. But it wasn't and in the Net I found only Patches
(mainly for older POV-Versions).
Thats not what I want. Only V. 3.5 and full throttle.
--Theo
PS: Its only intresting for those who have a) multiple CPU's or b) more than
one PC.
"Christoph Hormann" <chr### [at] gmx de> schrieb im Newsbeitrag
news:3D55121B.1C467686@gmx.de...
> Yes, so it's an automatism for distributing partial renders. No special
> handling of radiosity, photons, etc.
>
> Christoph
>
> --
> POV-Ray tutorials, IsoWood include,
> TransSkin and more: http://www.tu-bs.de/~y0013390/
> Last updated 03 Aug. 2002 _____./\/^>_*_<^\/\.______
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
The test rendering proofs that it was a good idea to distribute things.
I could render optics.pov at 1600x1200 distributed on 5 CPU's in :
14:51:03 - 15:29:49 = ~ 30 minutes
now I tried it "the normal way" starting at 15:32:28 its now 16:06 and we
are still in Line: 277 from 1200.
I think I can stop that now cause I want to render other things ...
--Theo
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Theo Gottwald *" <The### [at] t-onlinede> wrote in message
news:3d551e3e@news.povray.org...
> I think I can stop that now cause I want to render other things ...
>
You might want to start by rendering some Radiosity samples - As
Christoph hinted, there are problems with your approach to distributed
rendering and some of POVs features, such as Radiosity - try it out, I think
you'll soon see why distributed-POV solutions are generally patches...
--
Pandora/Scott Hill/[::O:M:C::]Scorpion
Software Engineer.
http://www.pandora-software.com
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I hope I understand what you want to say.
Several scenes have a long "precalculation-time".
In some scenes the real rendering part is rather small.
However that is theoretical.
Try my programm and feel the diffrence when you render your picture on three
computers.
Feel the brute force.
If I render things in pieces I have other advantages:
- better caching inside the CPU/L2
- geometries outside that area get not touched.
- Just try it: render a small piece of your actual picture. Now think that
in the same time other PC's have just
rendered the rest of the pic. Thats it.
However, if you have two or more PC's you should try my program
and see the advantage.
In most cases there is a real advantage of (I guess) 25% -45% (2 CPU's).
We know that this numbers heavily depends on the resolution and the
precalculation-time.
It may not be intresting to make small pictures with a long
precalculation-time.
In every case its intresting in resolutions higher then 640x480.
If you compile it with a faster C-Compiler you get lousy 7%. If you render
it distributed you can get it in a quarter of the original time.
See it and feel the diffrence.You don't need to have a network with 8 PC's
in your bussiness.
But if you have ... thgen you're lucky :-) cause you have three CPU's more
then me :-))).
On each of these PC's there must be a installation of POV 3.5.
It should be in W2k or up. (tested with).
Under these circustances the Programms should not even need an installation.
Just copy the folder somewhere and share it if you want to render
distributed (not necessary for SMP-rendering).
Thats all you need to get your "1 week picture" in less at the same day !
Example: Try to edit glasses.pov (one number must be changed to enable the
photons).
Render it "the normal way" at 1600x1200 and wait 3-4 days for the result.
Then use my system and distribute it on several PC's. Get your pic at the
same day !
It works fine for me. I'll work on the surface a bit and then publish it on
my web-site.
However anyone who is intrested can qualify for beta-testing.
Thanks
--Theo
"Pandora" <pan### [at] pandora-softwarecom> schrieb im Newsbeitrag
news:3d553519@news.povray.org...
> "Theo Gottwald *" <The### [at] t-onlinede> wrote in message
> news:3d551e3e@news.povray.org...
> > I think I can stop that now cause I want to render other things ...
> >
>
>
> You might want to start by rendering some Radiosity samples - As
> Christoph hinted, there are problems with your approach to distributed
> rendering and some of POVs features, such as Radiosity - try it out, I
think
> you'll soon see why distributed-POV solutions are generally patches...
>
> --
> Pandora/Scott Hill/[::O:M:C::]Scorpion
> Software Engineer.
> http://www.pandora-software.com
>
>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Its not www.fa2.de as said in the first posting,
but www.theogott.de where you can see my three PC's.
They have together 6400 MHz and if there is anyone out there who thinks he
can render any picture at 1600x1200 faster, then lets try it. :-)
--Theo
"Theo Gottwald *" <The### [at] t-onlinede> schrieb im Newsbeitrag
news:3d554b13@news.povray.org...
> I hope I understand what you want to say.
>
> Several scenes have a long "precalculation-time".
> In some scenes the real rendering part is rather small.
>
> However that is theoretical.
> Try my programm and feel the diffrence when you render your picture on
three
> computers.
> Feel the brute force.
>
> If I render things in pieces I have other advantages:
> - better caching inside the CPU/L2
> - geometries outside that area get not touched.
> - Just try it: render a small piece of your actual picture. Now think that
> in the same time other PC's have just
> rendered the rest of the pic. Thats it.
>
> However, if you have two or more PC's you should try my program
> and see the advantage.
>
> In most cases there is a real advantage of (I guess) 25% -45% (2 CPU's).
> We know that this numbers heavily depends on the resolution and the
> precalculation-time.
> It may not be intresting to make small pictures with a long
> precalculation-time.
>
> In every case its intresting in resolutions higher then 640x480.
>
> If you compile it with a faster C-Compiler you get lousy 7%. If you render
> it distributed you can get it in a quarter of the original time.
>
> See it and feel the diffrence.You don't need to have a network with 8 PC's
> in your bussiness.
> But if you have ... thgen you're lucky :-) cause you have three CPU's more
> then me :-))).
>
> On each of these PC's there must be a installation of POV 3.5.
> It should be in W2k or up. (tested with).
>
> Under these circustances the Programms should not even need an
installation.
> Just copy the folder somewhere and share it if you want to render
> distributed (not necessary for SMP-rendering).
>
> Thats all you need to get your "1 week picture" in less at the same day !
>
> Example: Try to edit glasses.pov (one number must be changed to enable the
> photons).
> Render it "the normal way" at 1600x1200 and wait 3-4 days for the result.
>
> Then use my system and distribute it on several PC's. Get your pic at the
> same day !
>
> It works fine for me. I'll work on the surface a bit and then publish it
on
> my web-site.
> However anyone who is intrested can qualify for beta-testing.
>
> Thanks
> --Theo
>
> "Pandora" <pan### [at] pandora-softwarecom> schrieb im Newsbeitrag
> news:3d553519@news.povray.org...
> > "Theo Gottwald *" <The### [at] t-onlinede> wrote in message
> > news:3d551e3e@news.povray.org...
> > > I think I can stop that now cause I want to render other things ...
> > >
> >
> >
> > You might want to start by rendering some Radiosity samples - As
> > Christoph hinted, there are problems with your approach to distributed
> > rendering and some of POVs features, such as Radiosity - try it out, I
> think
> > you'll soon see why distributed-POV solutions are generally patches...
> >
> > --
> > Pandora/Scott Hill/[::O:M:C::]Scorpion
> > Software Engineer.
> > http://www.pandora-software.com
> >
> >
>
>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Theo Gottwald * wrote:
Looks like an intersting stop gap till povray gets its head round
distributed rendering, and I ca see situations where it will be excedingly
useful
> but www.theogott.de where you can see my three PC's.
hmmm, perhaps you should concider vmware. hooking those 3 vids to the same
boxen and reclaiming some of your deskspace (keeping the other machines
running and using vnc)
--
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
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I've tested two original Scenes in diffrent ways. Here is the original
Log-file:
patio-radio rendered in 1024x768 in 1 Tiles (-) from 08-11-2002 / 07:03:09
to 08-11-2002 / 07:21:14 (1084).
cornell rendered in 1024x768 in 1 Tiles (-) from 08-11-2002 / 07:21:16 to
08-11-2002 / 07:29:47 (511).
patio-radio rendered in 1024x768 in 2 Tiles (V) from 08-11-2002 / 08:01:28
to 08-11-2002 / 08:15:17 (829).
cornell rendered in 1024x768 in 2 Tiles (V) from 08-11-2002 / 08:15:18 to
08-11-2002 / 08:20:01 (284).
cornell rendered in 1024x768 in 8 Tiles (H) from 08-11-2002 / 08:49:56 to
08-11-2002 / 08:53:47 (230).
patio-radio rendered in 1024x768 in 8 Tiles (H) from 08-11-2002 / 08:49:53
to 08-11-2002 / 09:00:06 (614).
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Theo Gottwald * wrote:
>
> I've tested two original Scenes in diffrent ways. Here is the original
> Log-file:
>
> patio-radio rendered in 1024x768 in 1 Tiles (-) from 08-11-2002 / 07:03:09
> to 08-11-2002 / 07:21:14 (1084).
>
> cornell rendered in 1024x768 in 1 Tiles (-) from 08-11-2002 / 07:21:16 to
> 08-11-2002 / 07:29:47 (511).
>
> patio-radio rendered in 1024x768 in 2 Tiles (V) from 08-11-2002 / 08:01:28
> to 08-11-2002 / 08:15:17 (829).
>
> cornell rendered in 1024x768 in 2 Tiles (V) from 08-11-2002 / 08:15:18 to
> 08-11-2002 / 08:20:01 (284).
>
> cornell rendered in 1024x768 in 8 Tiles (H) from 08-11-2002 / 08:49:56 to
> 08-11-2002 / 08:53:47 (230).
>
> patio-radio rendered in 1024x768 in 8 Tiles (H) from 08-11-2002 / 08:49:53
> to 08-11-2002 / 09:00:06 (614).
The question with radiosity and photons is not whether your method is
faster but whether the image *quality* is the same. If you carefully
compare the images you are quite likely to find there are distinct
differences between what a single engine render produces and what
your tiles method produces. There is no doubt that for most of
POV-Ray's features your program has substantial possibilities for
speed improvements but for a small subset of POV-Ray's features your
program cannot resolve the problems inherent with those features in
muti tile renderings. It is true they will still render faster but
if you did an analytical comparison of the two methods you will find
that the images your program produces are not identical to the images
a single instance of POV-Ray will produce.
--
Ken Tyler
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hi Rick,
(VNC)
1. IF I do so, how can I play next month Unreal Tournament 2 with my friends
:-) ?
(I REALLY HOPE the man from Intel has been there and coded it for
SMP/Hyperthreading) :-)
2. The system works so I can proof that it works .... you can download a
BETA that will change in a few days here:
http://www.it-berater.org/smpov.htm .
3. If you have questions or suggestions mail me.
--Theo Gottwald
PS: I've never understood why so simple things are not already built in the
POVRAY Shell.
The person who did this Shell was really a perfect programmer, he did
not forget ANYTHING ...
just that ....
"Rick [Kitty5]" <ric### [at] kitty5com> schrieb im Newsbeitrag
news:3d558e7a@news.povray.org...
> Theo Gottwald * wrote:
>
> Looks like an intersting stop gap till povray gets its head round
> distributed rendering, and I ca see situations where it will be excedingly
> useful
>
> > but www.theogott.de where you can see my three PC's.
>
> hmmm, perhaps you should concider vmware. hooking those 3 vids to the same
> boxen and reclaiming some of your deskspace (keeping the other machines
> running and using vnc)
>
> --
> 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
>
>
>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Sat, 10 Aug 2002 15:16:11 +0200, Christoph Hormann <chr### [at] gmxde>
wrote:
> Yes, so it's an automatism for distributing partial renders.
BTW: I have readed some old pov related pages from my own archives and found
link to www.distributed.net. Let's dream about distributed rendering :-)
ABX
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
