I was sitting here trying to get motivate to get this done, and I was
wondering if anyone out there would really be interested in it?

I'm doing the initial client and a rough server in Java. I have
prototype tests working on Windows and Linux at the moment. So, are
there any people who could actually use this once it's in shape for a
public beta? I'm going to be putting it to use for the IMP, but I would
like to help others if possible.

The system is designed to work in WAN/Internet type settings first of
all, but should work fine on a LAN for a local render farm as well.

--
"My new computer's got the clocks, it rocks
But it was obsolete before I opened the box" - W.A.Y.

Post a reply to this message

"Jon A. Cruz" wrote:
> 
> I was sitting here trying to get motivate to get this done, and I was
> wondering if anyone out there would really be interested in it?
> 
> I'm doing the initial client and a rough server in Java. I have
> prototype tests working on Windows and Linux at the moment. So, are
> there any people who could actually use this once it's in shape for a
> public beta? I'm going to be putting it to use for the IMP, but I would
> like to help others if possible.
> 
> The system is designed to work in WAN/Internet type settings first of
> all, but should work fine on a LAN for a local render farm as well.
> 
> --
> "My new computer's got the clocks, it rocks
> But it was obsolete before I opened the box" - W.A.Y.

How would it work exactly. I don't mean from a programmers veiw point
but rather as a group of people hooked up to the system from remote
locations. Would the combined power from all of the computers online
be used to render one scene or could multiple sessions be running at
one time and the task shared evenly for all those connected or ?

  Pardon my ignorance of such things but I am curious to know and I am
sure other lurkers out there might be interested in the details as well.

-- 
Ken Tyler

See my 700+ Povray and 3D Rendering and Raytracing Links at:
http://home.pacbell.net/tylereng/index.html

Post a reply to this message

Jon A. Cruz wrote:
> 
> I was sitting here trying to get motivate to get this done, and I was
> wondering if anyone out there would really be interested in it?
> 
> I'm doing the initial client and a rough server in Java. I have
> prototype tests working on Windows and Linux at the moment. So, are
> there any people who could actually use this once it's in shape for a
> public beta? I'm going to be putting it to use for the IMP, but I would
> like to help others if possible.
> 
> The system is designed to work in WAN/Internet type settings first of
> all, but should work fine on a LAN for a local render farm as well.
> 
> --
> "My new computer's got the clocks, it rocks
> But it was obsolete before I opened the box" - W.A.Y.

I would set up a LAN for it. 
(come to think of it: if it isn't too expensive I'd like to have some chickens
and a pig or cow for my render farm too...)

Regards,

Remco

Post a reply to this message

Ken wrote:

> How would it work exactly. I don't mean from a programmers veiw point
> but rather as a group of people hooked up to the system from remote
> locations. Would the combined power from all of the computers online
> be used to render one scene or could multiple sessions be running at
> one time and the task shared evenly for all those connected or ?
>
>   Pardon my ignorance of such things but I am curious to know and I am
> sure other lurkers out there might be interested in the details as well.
>

I have a slightly out of date early design doc that gives a high level
picture at http://www.geocities.com/SiliconValley/Network/4453/rfarm/

In general, there will be a set of renders in the queue. Clients will ask
for the next render, and will get them in turn. Needed assets (includes,
textures, etc.) will only be sent to the client if they do not already have
it (reduces network traffic). Once the image(s) are done, they are packaged
up and sent back to the specified location. So it is one render per client
computer. However, it is not hard to get each render to be only a strip in
an image as long as no radiosity is used.

At the moment I have a client that works OK with the user getting needed
assets and placing them, then running POV-Ray once all needed assets are in
place. The clients use "transports" that are 'plugins' in a sense. I have
the "user transport", and had done a little on a "mail transport", but now I
am going to do a simple "TCP/IP or HTTP transport" so that things will work
better for a LAN situation. I'm starting with a very dumb server, but this
can be tweaked as needed. (my original server was me, so it was a _very_
dumb server at first)

Part of this is that I'd first be working on areas that I wanted for my
work. But if other people need other areas polished first, I can get to
those instead. But I'd need people who have more than one computer with
POV-Ray, can run Java on those computers, and who would like their own
render farm. As far as the Java goes, I'm keeping to 1.1 for now, so it
should run on many platforms.

Also, my intention is that once I have the code a little cleaned up and out
of prototype stage, I'll be releasing it under GPL.

--
"My new computer's got the clocks, it rocks
But it was obsolete before I opened the box" - W.A.Y.

Post a reply to this message

Jon A. Cruz <jon### [at] geocitiescom> wrote in message
news:37B25B55.9C47EE4B@geocities.com...
> I'm doing the initial client and a rough server in Java. I have
> prototype tests working on Windows and Linux at the moment. So, are
> there any people who could actually use this once it's in shape for a
> public beta? I'm going to be putting it to use for the IMP, but I would
> like to help others if possible.
>
> The system is designed to work in WAN/Internet type settings first of
> all, but should work fine on a LAN for a local render farm as well.

I would participate in this.  My interests are in personal LAN rendering,
but it sounds like any support we can give to you project will also be
beneficial for personal LAN users.  Hey, it wouldn't hurt to have a global
rendering network either, much like other famous distributed efforts.

I can offer my humble network of 9722 PPS (SkyVase) in the effort.  Although
I'm a coder, I don't do Java, so sorry, but no help there.

Post a reply to this message

On Thu, 12 Aug 1999 07:58:04 -0700, "Jon A. Cruz"
<jon### [at] geocitiescom> wrote:
[snip]
>In general, there will be a set of renders in the queue. Clients will ask
>for the next render, and will get them in turn. Needed assets (includes,
>textures, etc.) will only be sent to the client if they do not already have
>it (reduces network traffic). Once the image(s) are done, they are packaged
>up and sent back to the specified location. So it is one render per client
>computer. 

Certainly sounds interesting - and I've a small render farm to test it
on. Actually more like a tortoise cage in the back garden than a farm,
but you get my drift I'm sure.

One thing that'd be interesting (or not) would be a LAN farm with
occasional WAN clients - maybe with HTTP/FTP communication on the LAN
but requests and results via e-mail to "the outside world" (it's
prohibitively expensive in the UK to remain online for any large
amount of time).

It's be nice to know how much work is required at the server end
(setting up dependancies and the like) but... well, I guess I'll have
to go take a peek at your early documentation!


Cheers,

Cliff Bowman
Why not pay my 3D Dr Who site a visit at http://www.who3d.cwc.net/
PS change ".duffcom" to ".net" if replying via e-mail

Post a reply to this message

Cliff Bowman wrote:

> Certainly sounds interesting - and I've a small render farm to test it
> on. Actually more like a tortoise cage in the back garden than a farm,
> but you get my drift I'm sure.
>
> One thing that'd be interesting (or not) would be a LAN farm with
> occasional WAN clients - maybe with HTTP/FTP communication on the LAN
> but requests and results via e-mail to "the outside world" (it's
> prohibitively expensive in the UK to remain online for any large
> amount of time).

Ummmm. Gee. Guess what. That's one of the main design issues. In fact my first
little tests have a "User" transport that has you pick a text file on your
system with the job description. Then if you are missing assets, it gives you a
text snippet to send as a request to the server. I also had a basic mail one so
that it would automatically mail the requests via SMTP.

For the IMP, the use case is that Internet traffic time is in short bursts, and
render times take quite a while. Disconnected rendering is a must.

At the moment I'm working on the HTTP stuff. Another test version should be
viable any day now. In theory you could even use a simple static page HTTP
server to feed a local turtle ranch.

> It's be nice to know how much work is required at the server end
> (setting up dependancies and the like) but... well, I guess I'll have
> to go take a peek at your early documentation!

--
"My new computer's got the clocks, it rocks
But it was obsolete before I opened the box" - W.A.Y.

Post a reply to this message

On Tue, 17 Aug 1999 18:37:56 -0700, "Jon A. Cruz"
<jon### [at] geocitiescom> wrote:

[snip]
>At the moment I'm working on the HTTP stuff. Another test version should be
>viable any day now. In theory you could even use a simple static page HTTP
>server to feed a local turtle ranch.
>
Should I try and catch up with my newsgroup reading so I notice an
"announce" post in here, or try and check your documentation web site
regularly? I plan to check out the URL posted (days ago) when I
connect to send this :)

I'll try and check out NetRay at the same time (we could really do
with some of these distributed rendering *users* comments!).


Cheers,

Cliff Bowman
Why not pay my 3D Dr Who site a visit at http://www.who3d.cwc.net/

Post a reply to this message

Cliff Bowman wrote:

> On Tue, 17 Aug 1999 18:37:56 -0700, "Jon A. Cruz"
> <jon### [at] geocitiescom> wrote:
>
> [snip]
> >At the moment I'm working on the HTTP stuff. Another test version should be
> >viable any day now. In theory you could even use a simple static page HTTP
> >server to feed a local turtle ranch.
> >
> Should I try and catch up with my newsgroup reading so I notice an
> "announce" post in here, or try and check your documentation web site
> regularly? I plan to check out the URL posted (days ago) when I
> connect to send this :)
>
> I'll try and check out NetRay at the same time (we could really do
> with some of these distributed rendering *users* comments!).

Well, keep an eye out on the group for an announce. If you want, you can also
bug me off and on with email.

Or just try to catch me on #IMP on irc.new.net or #povray (Been hanging out a
lot more since getting DSL)


--
"My new computer's got the clocks, it rocks
But it was obsolete before I opened the box" - W.A.Y.

Post a reply to this message