Virtual Machine performance

[im.thatoneguy]

I'm trying to benchmark virtual machine performance with V-Ray but running into a peculiar artifact that I was hoping maybe someone could cast some light on. Performance on modern CPUs seems to be nearly the same between a native system and a virtual machine with one exception: "Updating instances".

Any idea why there would be a slight but notable slowdown in the instance updating step of the rendering? Does Chaos Group also have any experience with testing in a VM environment vs a native environment and have any insights on best practices or drawbacks?

[vlado]

I've not run Windows in a VM, only Linux and V-Ray works more or less fine. But in any case, can you take a look at the V-Ray log file to see what is happening at that moment?

Best regards,
Vlado

[im.thatoneguy]

I'm running it through deadline to try to minimize as many environment variables as possible and the last output in the console is:

0: INFO: Updating instances

The next update when it progresses is:
0: INFO: Preparing lights

So I'm not sure if Vray prints "Updating instances" when it finishes or at the start. I presume the start although I guess it could theoretically be the "preparing lights" output.

[vlado]

I'm not sure what is the relation between the V-Ray log file and env. vars? In any case I don't think the Deadline output contains the same info.

Best regards,
Vlado

[im.thatoneguy]

Ahh, figured it out. There is a loop that it sits on here for about a minute on the VM machine. (There are many, many... many instances of this object).

[2014/Mar/12|17:36:12] Can't use fast shading for object "ChamferBox008": generate GI is not 1.0
[2014/Mar/12|17:36:12] Bounding box for "ChamferBox008" is [52.3531,7.56557,109.05]-[53.2516,7.85335,109.946]

[vlado]

Hm ok. How much faster is it on an actual machine?

Best regards,
Vlado