Greetings all!

  As many of you already know I am very interested in starting a render farm service for Bryce. Recently I have been compiling numbers such as utility costs and hardware costs, getting things together for Return of Investment calculations. There's a few technological questions I have left and I'm in the process of researching those things. But I would like to ask you all your thoughts on costs. How much would someone be willing to pay for a rendering a static Bryce scene or Bryce animation? What would be some questions I should look into? What would be some ideas for servicing?

  I'm currently considering buying four quad cores to start a dedicated farm this year, more next year. There's much more involved, but for now I have these questions to ask and would like to feel the waters, in a manner of speaking, to what sort of business opportunity is out there and what sort of demand.

  Would anyone be willing to use a render farm? If so, how often?

Before you invest your money, please check out my test results if you havent already.

http://www.renderosity.com/mod/forumpro/showthread.php?thread_id=2692302

If you are planning on using Bryce Lightning, you may want to contact DAZ and have a chat with them about optimizing the program.

Four Quad Cores will be very fast in rendering you're jobs, but you might not be getting as much out of that processing power due to shoddy network rendering engine.

Here are a couple of links that will give you an idea what the pros charge:
www.rendercore.com/rendercoreweb/priceInfoView.do

http://www.rebusfarm.com/costs.htm

I think it is great that you want to provide this service, the Bryce community needs it.  However I think something should be done in the Network Render Department first the render engine is seriously flawed and needs work done, at this point of time I think that part of Bryce was an afterthought.

You might find that For Each Quad Core processor you will only save like 4-8 Minutes of render time.. and you have to think if its worth the investing for the small amount of savings.

Thanks Analog-X!  I have been following your threads about your render farm closely. I'm a graph sort of person, like to see a collection of numbers put into a graph. To help visualize I would need to graph the data I've collected. Still researching everything.

I'm getting involved with clustering, and a Bryce render farm would just be an extension of my efforts with clustering. There's other reasons why I want to get into a cluster, all business related. But in short, there's other uses for all the power that gets into a render farm. But in terms of Bryce, there's a lot of work that needs to get done, and I'm interested not only doing the service but also the programming and stuff that goes into making such a service work.

The idea to contact DAZ is a good one I'll probably do within the week or two. Those other farms are interesting, especially that they support more than one software package.

Need more research, need more testing, need more benchmarks.

Phew!! :) thank you for understanding my post, I was a little worried that you might take offense to it :).

Ok... speaking of graphs. While doing the test I did a quick little excel spreadsheet and here are the results.  I did not include all this info in my other Render Farm related post.

PC Number and Processor Speeds.
PC#1 (2Ghz), PC#2 (1Ghz), PC#3 (800Mhz), PC#4 (800Mhz), PC#5 (700Mhz), PC#6 (600Mhz)

Initial Test to see render time on my fastest PC using Omac2's Bryce 6 Benchmark Scene.
This was done without Bryce Lightning.
Render Time:         PC:
18:56                      PC#1                  

All Tests Below Were done with Bryce Lightning.

Render Time
20:00                                  PC#1  (2 Ghz Total) <--- Interesting how its 1:04 Longer through BL vs Standalone.

14:55    4:01    21.21%    PC#1+PC#2 (3 Ghz Total)             

12:17    6:39    35.12%    PC#1+PC#2+PC#3 (3.8 Ghz Total)         

10:32    8:24    44.37%    PC#1+PC#2+PC#3+PC#4  (4.6 Ghz Total)    

9:21       9:35    50.62%    PC#1+PC#2+PC#3+PC#4+PC#5 (5.3 Ghz Total) 

8:37    10:19    54.49%    PC#1+PC#2+PC#3+PC#4+PC#5+PC#6 (5.9 Ghz Total)

First Column is the Render Time.
Second Column is Time Savings compared to original Stand Alone Render time of 18:56
Third Column is the Percentage in Speed Gained again compared to Original Render time.

I would like to see some features in Bryce Lightning to help increase render times such as.

• Multiple Processor/Core Support
• Automatic Detection of Processor Speed
• Assign Render Frame Size according to Processor Speed.  i.e.  Machines with faster processor speed would get larger portions to render.
• Better Utilization of CPU i.e. You can Set Higher Priority levels for Render Dedicated PC's.

Looking at the above chart....It would probably safe to say that I can provide some assistance to someone who may have a PC with a Single 2Ghz Processor, they can potentially reduce their render times in half.  Someone with a Dual Core or Dual 2 core would not benefit from my render farm.

Since there is no Multi Core Support with the current Bryce Lightning, you would have to Run an instance of Bryce Lightning for each CORE and even than, you would have to find a solution for providing a separate IP Address for each instance of Bryce Lightning.

I hope this bit of info helps.

A quick little Update:

I got to thinking what if I set the Priority for each Bryce Lightning to "Real Time" and do a test render.

The Result was a savings of 0:04....:) Yup 4 Seconds.  So total render time was 8:33 instead of the 8:37 with each Bryce Lightning running at Normal Priority.

one other thing I forgot to mention.  All the Pentium III Systems run Windows 2000.  After doing some research and looking at benchmarks, its was clear on Slower PC's or Older PC's, Windows 2000 outperforms Windows XP.

That is good data to know. I'll have to put that into Excel. That is also good info about Win2K vs XP.

To keep costs down, I intend to buy just the minimum to make a running system. Off the top of my head without being precise, I'm looking at buying the motherboards, chip, ram, and power supply. No case, no fancy extras. The case will be something custom made (it's being designed in Solidworks at the moment.) But I'm curious about other things... is it possible to have 1 OS installed and use the other motherboards as a diskless system? Meaning that there's 1 OS installed, 1 hard drive, and have the motherboards work from that? According to some web sites Windows Terminal Server could be used for a diskless system.  I'm leaning towards a diskless system.

What are the Windows booting options is another quick question. Boot from a CD.. over a network? If all these motherboards are going to be networked, it would be nice to use one OS or at least a minimum of OS. Licensing issues aside.

what you are describing is almost something like a Thin Client situation.  Somethings that come at the top of my head.

• BartPE bootable live Windows CD/DVD this is something that so far I've been using to Clean Virus/Torjan/Spyware infected systems that people bring to me without having to boot off the HD.  You or I would have to look into writing drivers to get Bryce Lightning or any other Clients running with this setup.   You could potentially Boot off of CD and use a RAM DISK to store everything.

• I havent looked into Windows terminal Server, but that is an interesting idea.

• LitePC  is another interesting approach, this program allows you to customize and run a very lite version of windows 2000 or XP.   So potentially a dedicated render system could run windows without any annoying extras getting in the way.  I had purchase 98Lite a few years ago so I was able to purchase XPlite / 2000Lite at 40% off.  So I can do some tests with it as well.

• VirutalPC - Microsoft offers a Virtual PC Solution for free.  There is a 32 Bit and 64 Bit version.  You can run multiple PC's on a Single Machine.

Just a few ideas from the top of my head :)

virtualpc is not really going to help boost performance though.. ur still simulating those multiple pc's through 1 pc. That and VirtualPC is rather slow, ánd virtual machines dont do hardware accelerated graphics as far as i know, so ur opengl view in bryce might have problems as well.

This is all some pretty good info. I've been getting some pretty interesting feedback both here and elsewhere. My main goal is to build a high performance cluster, and as I was thinking about how Bryce Lightning does not take advantage of multi-core's, I felt terrible until I realized the grander scheme of things. Still reading up though. Other issues that must be considered seriously is the heat such a system will generate. I'm also looking into a/c units, for instance. It's a lot more than just buy a bunch of computers and go.

Those prices from those professional render farms seem kinda high to me. As a Bryce user, I don't like high prices and I think to be attractive for other Bryce users, the price for using a render farm will have to be something low enough someone won't think twice about. Anyway, more later ;-)   Have to think about the next challenge.

Also having some type of UPS system to handle the load incase of  a blip in the grid.

Oh yes, definately. Probably an expensive one for the load of all those systems. cringes at that thought  I have a friend who just had to spend $130 on a new UPS battery. All the UPS's I personally use are the small $40 ones with a $6 battery replacement cost, but the render farm will probably have to have something bigger.

--Death

I have about 6  1000KVM units all bought on sale around $120-$130.  You can get good ones on bargain.