mixed AMD and Intel ??..

[r.saracini]

A render farm with Athlon's and xeon's with max could be work?..
or there are compatibility problems?..

thanks in advance!..

[percydaman]

in the past people have reported slight differences between renders from the intels and the amds. Not sure if its still the case but I wouldn't be surprised.

[richard_rosenman]

Hi;

When I used to use Softimage on SGI's and NT's there were certainly differences - slight, but enough to make difference passes useless. We did a test rendering the same frame on each with Mental Ray, and imported them into Photoshop - then lay one on top of the other and switch to difference comp mode. If you get true black, both frames are identical. If you don't, they are different.

We bought a large Intel Renderfarm at work, and we had 5-6 additional AMD boxes (I think Athlons), as well as P3's and P4's and I was concerned again about this so I ran a few tests. All frames rendered from each chip came out identical when we put them through the difference pass. Of course, there may be differences that arise in other cases but so far it seems that we have successfully mixed AMD's and Intels for rendering.

Whenever possible, I still prefer to assign a job to just one batch of processors.

-Richard

[Syclone1]

We have a mix of AMD and Intel here and no noticeable differences so far.

-dave

[mike.edel]

i remember some reports of different frames after discreet enabled some pentium 4 specific optimizations (those were more accurate and caused different renderings)
this was max 4.25/4.26

some expected differences are described in the readme:
ftp://ftp1.discreet.com/web/support/...smax_42sp1.htm

"
o Raytrace materials can exhibit differences in edge antialiasing along object edges. These differences are generally not noticeable in single images unless compared pixel-by-pixel in a zoomed format.

o Cellular maps render slightly differently. These differences are generally not noticeable in single images unless compared pixel-by-pixel in a zoomed format.

o Reflection and Refraction maps with Bump maps render slightly differently. Most noticeable of these is the Thin Wall Refraction map. These differences are generally not noticeable in single images unless compared pixel-by-pixel in a zoomed format.
"

those seem to be connected to the intel compiler used for that update

[r.saracini]

Ok, thank you !!

[mike_kennedy]

Back in the day particles could be a challenge (flickering, but we haven't seen any of that lately, but we only have one intel box in production to run our Velocity....