Hi people, people always ask about Carrara and processors performance, so a friend find a way to create a database that others can look at and do some comparisons and decide what system buy or use.
I recreated the scene that you will use to create the benchmark and that scene will use something from almost all great features from shader and render engine: SSS, TRansparency Absorption and In-Scattering, Radiosity, Reflection and Refraction.
Help us to create a good database for Carrara here: just download the file, render and take note of the time:

http://www.3dspeedmachine.com/?page=3

Happy to participate. It was through these 3D benchmarks, that I was able to select my system. I had to use other 3d applications as an estimate of performance, but now we can add Carrara to the list.

I have a feeling that a lot of these "tests" are pretty superflous-----there are just so many ifs/buts!

I have three tests of your file.(Carrar6 Pro)
A took 32.16 (*in my system)
B took 0.57
C took 9.46

Question
I only altered the render settings in each.( Is B a fault in the "lighting" sets originally?)

In you "quest" are the variations more in setting un-necessary "theoretical" aims rather than, practical, acceptable, end results?

Just adding some food for thought.

Brian

The idea is to have a point to compare machine to machine, so no change can be done before the render. The settings are for a good result with GI without lots of splat shadows artifacts.
If everyone use your system to render the scene file we can look at it and see how other system performe the same scene... so we can determine the real diference between a Core 2 Duo 2.4 and a Core 2 Duo 3.0 for example.

I am trying to counter that by suggesting

A---What is a machine?---the variations within any one are quite large (HUGE)--even  "have you the latest card update!?" for example.

B---I am sugesting that, also, settings are maybe made for rendering that are way, way in excess of the needs for the end result.

If a render takes longer than 60mins --for anything (MY spec!)---I think the problem is the settings and nothing to do with the end result render! (Obviously, this is subject to wanting a render better than one needs for A4 prints maybe?----another if/but consideration?

Just some thoughts

Brian

Brian, the point is to compare different systems handling the exact same information, so the quality of the end result are not important.
Also I think the videocard does not play a roll with rendering.

The processors and the operating system are what matters. And putting Carrara in that list is nice too :)

Right Hoofdcommissaris, the idea is to compare processors speedy and I used high quality setting because a simple scene can get a very low render time and maybe cant be good for  comparison. In the other hand my real ideia is to put Carrara sid by side with the big guys out there, so people look and ask: what software is this here?? I had talked with a lot of people that even dont know that Carrara is in the market! So yes we are there.

There are a couple of snags which your test doesn't take into account.

First off speed can vary quite a bit between different brand and model motherboards.  Asan example, both my wife and I have the same CPU a Phenom X4 9600.  I use a higher end MSI K9A4 Platinum motherboard, wheres she uses a lower cost board from Asus.  I ran the test on both machines and mine got a speed of 4:09.  Hert machine got a 7:24.  Both machines have 4GB of RAM which brings me to the next item.  Memory will also affect speeds.  Not just the amount, but the type and memory speed.  For instance, my wifes system uses DDR memory, but mine uses DDR2 Dual Channel RAM.  Major speed difference.  There are other variables as well.  One is the Hard Drive speed.  Carrara will always end up going to the hard drive no matter what you do.  A slower drive will slow things down considerably, whereas a faster drive helps a lot.

Just a few things to keep in mind when folks do these tests.

Quote - The idea is to have a point to compare machine to machine, so no change can be done before the render. The settings are for a good result with GI without lots of splat shadows artifacts.
If everyone use your system to render the scene file we can look at it and see how other system performe the same scene... so we can determine the real diference between a Core 2 Duo 2.4 and a Core 2 Duo 3.0 for example.

I did a rendering of the scene on my Dual Mac 2.3 Ghz (uhm... 32:45 minutes...), but I seem unable to upload the result to the site. I will try again later...

I was doing some work related stuff (even opening Photoshop), I think my machine can do better than that.

Great, I saw that some people are having trouble to add data to the site...I will talk to Joe to see what is happening. Thanks again people.

Sparrow its almost impossible to cover all components... this is only an aproximation. The good idea is that all this data is posted in the site, so we can see this problem and maybe Joe can do something in the future to update the data base with more information from system.

I only bring it up because it might be a good idea to provide a few more categories like MoBo Brand and Model, RAM type etc.  Then they could get a much broader look at what other systems are configured as, and how they perform.  Just make them text entry fields so people can just write in their details, or simply put a comment box people can use when they enter their system data for the database.  The thing is that if I were to put in my wife's system and mine, they would look exactly the same in the database, but they would have a very different speed value.  That could confuse people, buy putting something in, so specific info about the systems is available,  would lessen the confusion and give them an idea of how best to configure a system they may be planning to build.  In short, it would make it much more useful.  Without it, you could have a situation where the user can't get any info which would be of use.  For instance.  My system smoked most of the systems listed, although there were only a few.  However, my wife's system was smoked by most of those systems.  Since they are both AMD, using the same processor, and the same amount of RAM, how would a user, at seeing this, differentiate the two AMD machines when they look the same in the Database but have very different test scores?  If he could tell what is Different about them, then he could see why the scores differ so much.

Quote - Sparrow its almost impossible to cover all components... this is only an aproximation. The good idea is that all this data is posted in the site, so we can see this problem and maybe Joe can do something in the future to update the data base with more information from system.

When I was still only working with Poser (sssssssss..) and photoshop, I bought a new system to handle it. I wanted the right video card to get the best assembly room response and enough processor power to do complicated renders. I had no idea what to buy.
There are people out there who use benchmarks like these to test every computer component available. By comparing them, I found the right components for the least $.
A local computer shop assembled my list of components, for the same price that an in-store gaming computer would cost.
The $ I saved have gone to Renderosity and Daz. : ) 

Hehehe this is the idea SuePoser, I passed the infomration about other components parts to Joe and let see if he can add this soon to the data base too.

Gk, *muitos agradecimentos para fornecer esta informação.

*however, did they explain why the Max renders are so much better than the carrara renders?
e.g., did they explain why the carrara renders are missing the yellow glow fx that are seen
in the Max renders?  AFAIK carrara can do that, but these renders aren't showing the glow.

They probably just forgot to set up the Aura effect in the Carrara version.

I edited the file and just used simple shaders with things that is usefull for processor time like SSS in the strange red thing and the new transparency options in the water... Aura is great but it is a post process that take long to calculate, so I cut it out. The idea isnt to compare render quality but render time. I talked to Joe and it will work in a way so people can describe better your system, the problem is that only 1/3 of people that is registered sended your benchmark, and if he add more fields to the data base people will not take the time to full fill it... its a shame but happens.

I had some trouble sending mine in.  The link didn't seem to do anything when clocked.  I'll try it again later on then.