Hi everyone, I am writing to you for a technical opinion regarding the “maximum” system requirements of Rhino 6 for Windows, because the latest information on the English forum dates back to June 2018 and are not clear:
Does Rhino 6 work with single CPU based systems with more than 64 cores? Therefore exceeding the maximum number of 63 cores indicated in the system requirements.
If I mounted a CPU with 64 cores and 128 threads, would Rhino crash?
If it crashes, would it be possible to set the use of a defined number of cores in Rhino, so as not to exceed the maximum limit of 63?
I ask you this because I would only use Rhino for 3D modeling, as I would use V-Ray for rendering (which is why I would like to buy the 64-core processor).
Thank you so much to those who want to help me.
P.S. Do you have any news regarding Rhino 7 system requirements? And when will it be on the market?
Hello - as far as I know 63 cores remains the limit.
I believe the 63 core limit is imposed by the Microsoft developer tools we use for Rhino.
I would be surprised if Rhino crashes with >63 cores, it’s just that you won’t use the full computing power of your CPU when you are running Rhino. Or am I missing something?
I agree Menno; I wouldn’t expect Rhino to crash
I kinda half remember this limitation was added to the System Requirements by @dalelear , specifically because Rhino did crash. As I recall, it came up in a development meeting a long time ago. That must have been early V5 time frame.
Looking at old posts, could be that it was 2015 - 2016… Win10 was released in 2015.
Has McNeel the chance to check if today Rhino still crash? (I mean, after all the upgrades of win10)
At the moment processors with 64 cores are at the top of the market but I assume that in one or two years the prices will decrease and will be a good choice for a lot of users…
Would be better to be ready for that.
Dimitar on the italian Discourse has tested Rhino on a 64 cores machine
I’d be interested in hearing how the Raytraced mode works on that machine. The link fails to resolve for me, so can’t read the discussion.
Un processore da 64 core non l’avevo ancora sentito: vedo però che questo Ryzen è piuttosto recente.
Come va? Un saluto da un compaesano.
I’ve cutted my username from the link; is it working now?
Dimitar is a member of https://3dws.net/
They build professional workstations so I think you can ask them to make the tests you need
Si, stiamo parlando di processori che sono sul mercato da qualche mese… roba da top di gamma (fa conto che solo il processore costa intorno ai 4000 euro… se vai sugli Epyc passi a cifre intorno agli 8000 euro…).
E ci fanno macchine multi processore…
Lucio I go on in english, so we don’t exclude other people. : )
I’ve checked that processor: it also sucks 280 W of power: I can’t imagine a multiprocessor machine with this kind of CPUs.
Anyway, back to the topic: at this point would be nice to understand if Rhino works even with a multiprocessor assembled with 64 cores processors.
Even if all this computing power is more useful for the Rhino renderer than for Rhino itself: I don’t see the point for using such a powerful machine for just modeling.
So, verified that Rhino opens and doesn’t crash on such a machine, now it would be nice to check how Cycles works (and IF it works) with such a processor!
Rhino prefers high frequency vs multi thread so multi processor isn’t the best choice but, in my opinion, it’s the SO that has to manage the hardware resources in a way that the software doesn’t crash.
It depend on how big your models are, how many instances of the same program you use at a time and if you use plug-ins and other stuff that require a lot of resources (obviously I assume that on a workstation not only Rhino is used…).
Cycles don’t use the processor, only GPU… so we have to look at how “expensive” is the graphic card
@lucio_zadra That is not true. Go to
Tools > Options > Cycles in v6 and Rhino WIP, and set render device to CPU.
If you want to max out all cores you can use the command
RhinoCycles_SetThreads to set the amount of threads you want.
There have been tests of Cycles running on 128 threads I believe
Out of curiosity:
Are those threads very different from the 2000+ treads(CUDA cores) on modern GPU’s?
Well, the idea is still the same - divide up the work over available hardware using threads.
On the CPU though it is easier to find hardware that has huge amount of RAM to load your scene into, say 32GB. Or if you can find the money a server with two Epycs in and 4TB of RAM…
And if you’re working with larger scenes the time to move the data between CPU and GPU can be long.
Both have their pros and cons. But it should be clear that 128 CPU threads is much better than 4 CPU threads :°
Yes, my fault, sorry.
I answered that cycles doesn’t use cpu because, usually, gpu cuda is faster than cpu but with this new cpus… the “game” change.
Btw, what about a workstation with multiple gpu? I mean, cycles can go faster if we have two or more graphic cards installed (like in SLI mode)?
If yes, I think that gpu would be better at a lower price…
That is fine as long as you can fit the rendering in the GPU RAM. If you work with scenes that eat more than 16GB, then generally you’re toast, as GPUs with more than 16GB RAM are very, very expensive, and not easy to find. CPU+16GB on the other hand is cheap, even CPU+32GB or even 64GB.
As an example, the movie Next Gen is rendered with Cycles on CPU precisely because on the GPU it wasn’t possible.
Last week, in a nice talk with a render specialist, it came out that GPU render is not the solution, especially in professional behaviours, due to other important factors (other than RAM and others already mentioned).
First of all it seems that GPU’s are less “reliable” than CPUs: they can get stuck in the rendering of certain scenes with particular features, causing the process to hang or crash, much often than CPUs (at the end of the day, CPUs are made expressly for calculation).
Moreover, GPUs too intense activity can lead to very fast GPU electronics wear and necessity to substitute the hardware.
I am not so much into it, so I can’t really be sure about the reliability of this information, but the explaination they gave to me was making sense, I reckon.
This was my sensation, at least: it makes sense.