32 vs 64 Max benchmark

[jccloutier]

I just did a benchmark with the same max file; first in 32-bit then in 64 with all the same settings.

 

the results:

 

32-bit = 4m16s

64-bit = 5m55s

 

No, there is no typo. I am speechless and think I'll treat myself to a nervous breakdown.

 

Anyone have benchmarks they can post?

[BrianKitts]

I'm curious...could you post a little more on your scene....

 

such as: rendering engine, number of lights, was it modeled in max, modeled in something else and imported, number of polygons in scene....etc......

 

I know there was a dramatic speedup on working with editing polys in 9, is that also translatting into faster rendering or is it just a working speedup.

[jccloutier]

If comparing the same file wouldn't it not matter? anyway...

 

Modelled in ADT 2006

 

Scene Totals:

Objects: 3306

Shapes: 1

Cameras: 6

Lights: 65

Helpers: 100

Space Warps: 0

Total: 3478

 

Mesh Totals:

 

Verts: 642141

Faces: 1150294

 

32-bit render:

 

XP Pro 32-bit

2 GB RAM

Max 7

v-ray 1.47

 

64-bit render:

 

Xp Pro 64-bit

2 GB Ram

Max 9 64-bit

v-ray 1.5 RC3 64-bit

[Chad Warner]

It could very easily be the different versions of Vray.

[Brian Smith]

Jeremy,

Out of curiosity...why do you have over 3000 objects in your scene? How is your scene structured? I just ask because a scene with that many objects can often take much, much longer to render than the same scene with objects with similar materials collapsed and attached together. The steps Max has to go through to prepare the scene for rendering is greatly increased with more objects.

[vizwhiz]

Intel or AMD?

 

from what i remember reading is That AMD can run 32 bit natively

while Intel (might) have To run The 32 bit under an emulation mode

 

if you are running a 64 bit OS

 

running a 32 bit emulation mode is SLOWER Than just running it natively

 

**

 

dont quote me on any of This

i am just Thinking out loud, so

 

it might be something completely different

 

**

 

Thanks

 

Randy

[BrianKitts]

It could very easily be the different versions of Vray.

 

I agree.... to truely test a speedup, I would eliminate the variable of using different release candidates of vray. I realize you have to use RC3 to get the 64bit benefit. But because of the improvements and changes jumping from 1.47 to 1.5 there really is no way to tell without seperate testing of how much the improvement is coming from the new releases of vray verses improvements to max.

 

Don't get me wrong, its awesome to see the benefits, you have a valid mark, but it would also be interesting to see a breakdown of who's giving that stronger enhancement.

[jccloutier]

Jeremy,

Out of curiosity...why do you have over 3000 objects in your scene? How is your scene structured? I just ask because a scene with that many objects can often take much, much longer to render than the same scene with objects with similar materials collapsed and attached together. The steps Max has to go through to prepare the scene for rendering is greatly increased with more objects.

 

I guess the only answer I have for you is I'm not that proficient in Max and wouldn't know how to begin gravitating towards that method. However, if you ever get bored and want to explain, you have an audience!

 

All of my models are built in ADT as solids and imported to Max. I've been using AutoCAD fo so long now that what takes 1 hour in ADT takes me 6 hours in Max.

[AJLynn]

It's not a Max 64-bit vs. 32-bit issue. It's changes in the Vray sampler and the required settings for it. (With 1.5RC versions there are lower settings you can use to get the same quality faster.) Try installing Max9 32 and 64 on the same machine, and Vray RC3 32 and 64, and render with the same settings.

 

Other people have had similar experiences, if you look at the thread on the Cornell benchmark scene on the chaosgroup forums. You can't compare across different versions of Vray.

[jccloutier]

It's not a Max 64-bit vs. 32-bit issue. It's changes in the Vray sampler and the required settings for it. (With 1.5RC versions there are lower settings you can use to get the same quality faster.) Try installing Max9 32 and 64 on the same machine, and Vray RC3 32 and 64, and render with the same settings.

 

Other people have had similar experiences, if you look at the thread on the Cornell benchmark scene on the chaosgroup forums. You can't compare across different versions of Vray.

 

Actually I remember seeing a popup when I first rendered in 64 regarding the noise threshold not being the same as 32 but I guess I didn't take it seriously yet it makes sense. Thanks, I'll check it out.

[kippu]

how about using the max 9 + 1.5 on the 32 bit and posting that benchmark?

[Brian Smith]

Jeremy,

If you would like an in-depth discussion on this subject of scene assembly, please make a topic suggestion in this thread.

http://www.cgarchitect.com/vb/20082-visualization-insider-what-topics-do-you-want-see.html