I think I’m seeing a reason for some of my slave issues. It seems that when I increase my dynamic memory above what my slaves like, they don’t contribute to the rendering. My main workstation has 24G of ram, but my 5 slaves have 12G. If I increase the dynamic memory to, let’s say 8000G, the slaves stay idle, but if I lower the dynamic memory to 4095G, they all chime in. When I say idle, I mean that they don’t contribute. They are not idle at all. When I look at the task manager, of the slaves, they are all using 100% of their resources.
is this typical? What is the rule of thumb when setting up your memory limits? Is it a percentage of the weakest machine?
I have assumed it was 80% of your main workstation. I am just wondering what technically happens when you set it to high. I am doing some tests, but it seems like the slaves don’t even try when it knows it doesn’t have enough resources.
For en example: when I copy files, and I don’t have enough space, it’ll tell me and not even try. It seems like the same thing is happening in vray, minus the warning. If this was the case, it would be nice to see something in the vray messages window.
I have the same problem, the main machine have 96GB ram and the dr-slaves have 16GB ram if I set it on dynamic 88GB ram the dr-slaves won’t help to render the scene.
Is there a fix to this or do every computer have the same amount of ram to work properly?
Sorry this is off topic but whoa, 96GB ram! I’m curious, can you still see benefits of all that ram? can you for instance render an entire planet full off trees without Vray crashing?
Actually, you can set the dynamic memory limit to a negative value from MaxScript; in that case, V-Ray will start unloading geometry when the free system RAM falls below that limit. F.e. if you set it to -1000, V-Ray will try start unloading stuff when the there is less than 1000 MB of free system RAM. This may work better across machines with varying specifications.
Note that the amount of free RAM that you want, is only approximate, so it’s not a good idea to set it close too close to 0.
So, entering the -1000 value directly in the field, in the system rollout, would not work? What’s the difference between entering the value from MaxScript and entering it manually, as we are looking at the same value?
No, it will be clamped to 0 (which means no limit).
What’s the difference between entering the value from MaxScript and entering it manually, as we are looking at the same value?If you do that from MaxScript, the negative value will be set correctly. If you do it from the text field, it will be clamped to 0 at the moment. This is because the feature is somewhat experimental and not completely well tested.
Testing, but I guess on a scene that does not need it enough.. not much speed gain ( not testing the original DR issue here, but just checking if I can have any performance boost..)
Actually slightly slower than imputing the values by hand..?
On a 48GB slave system ( which idle seems to run at 3GB..)
Setting 44000 manually = rendertime 8:39
Setting -4000 via MaxScript = rendertime 8:53 ( prerender script ) or 9:03 ( regular script)
I would love the use of a - negative memory value, as lots of our slaves have very different ram levels.
44000 won’t be the same as -4000 tho will it?
44000 will always be 44000 but -4000 will be 48000-4000-any RAM used by OS, other applications and non dynamic memory in max (?), so it will always be less.. correct?
I have tried this script in Vray 3.6 and it doesn’t work.
I think I have the same problem described before, most of the time, as project become bigger, my slave machines stop rendering.
I have 64 gb RAM in my main computer, another one 32 gb, and 4 more at 24 gb.
And it is the order they stop rendering, first the 4 weakest, than the 32gb and at the end only my main computer is rendering.
I red that the subject is pretty ancient, Is there any solution fixed since then?
Can’t we put directly a negative amount of RAM to save what the system needs to work?
