pc rigs for C4D + ePMV

#1

Hi, I’m new to C4D and ePMV. My current desktop is Win 7 64bit, dual core CPU i5-2500K at 3.3 and 3.6 GHz, with 16GB RAM. I was playing around with some viral particles (2plv): when I load the whole viral particle, 60 subunits, with “Atoms” representation, it took forever; even if it finally got loaded, it froze when I tried to change the atoms’ scale. So, I’m thinking about upgrading my system. Can anyone recommend some rigs for a budget under $2500 (main consideration is for CPU + motherboard + video card)? Largely for scientific modeling and rendering of macromolecules/viruses, and some animation as well (short ones under 5 min maybe)

This one looks really sexy: http://blog.digitaltutors.com/building-a-home-render-farm-without-breaking-the-bank/, but a bit over my budget.

Here are some nice ones too: http://www.tomshardware.com/answers/id-2061283/rendering-computer-buy.html

I’m kinda debating on dual Xeon cpu vs single i7 CPU. Any suggestions? Do I need a killer graphic card? (BTW, when my pc froze, I found that C4D ate >95% of my 16GB RAM, while the total CPU usage was only ~33%).

A million thanks!

#2

Don’t.

First lets get one thing out the way, you need to find out where the slowdown is if you want to tackle it. First of all, explain a bit more about the models and atoms. Are these simple static meshes? Are you using any tool or plugin to generate meshes? Hypernurbs/SDS, Atom array objects? Some models you’re not going to be able to manipulate in real time on any machine.

RAM, if youre filling your entire ram then this is the first place you should start. Either plain old get some more, or better still, look how you can make your scene more efficient, 16 gigs is quite a bit to use. Also have a look at your undo buffers, if youre editing a massive heavy multi million poly mesh, then each action creates an undo buffer so you can go back, this could be killing your memory so set your undo buffer down to 1-2 and see if this helps.

cpu, yours is still OK, it was never the most amazing one to begin with being dual core, but it isnt bad, and you need to realise that editing the mesh is a single threaded task, i7’s, xeons and opterons wont help you. There are two things worth doing. First consider upgrading the cpu, you can double your render speed by switching in an i7 chip, this will cost about £200. Secondly go hit the overclock button, youd be mad not to. Your 2.4GHz machine will easily run at 4-4.5GHz with very little effort. If youre happy entering the bios and changing a couple of numbers you can give your machine a massive speed boost for free on the i5 or i7 chip.

gfx card, what is it? you didnt list one.

Either way, that first link with the opteron machine would be a terrible idea, it will be slow and expensive.

#3

It sounds like the workload doesn’t use all of the processor cores so the clock speed is more important than the number of cores, so the processor you already have is a decent one for the task. If that’s not the case and it does use all the processor cores please let me know.

It also sounds like all of the memory in the system is being used and you need more memory to work with that data. That processor is able to handle up to 32GB of memroy but you’ll need either Windows 7 Professional or any edition of Windows 8 because the Home Premium edition of Windows 7 has a 16GB limit on memory.

The graphics card matters if you’ll be viewing the results in a viewport rendered with OpenGL. If you are just rendering to image files from a renderer the graphics card won’t have any impact on that (unless the renderer is using OpenCL or CUDA, unlikely though). If you find 32GB isn’t enough memory for your needs then it might be time to upgrade to a dual processor socket workstation with many more memory slots.

#4

Hi Mat and Luke, many thanks for the input! Perhaps it’s really unrealistic to handle this thing on my desktop: the whole viral particle contains 60 subunits, and each subunit contains 6683 atoms (spheres in the model I was trying to play around). The attached screenshot shows 1 subunit. The structure was built upon protein database data, by an open-source plugin called ePMV.

Actually I should say I haven’t yet touched the rendering part for this viral particle. It’s the loading of the whole particle, i.e. 60 subunits, under the atom representation mode that kills my RAM. Also, even when I was trying to manipulate this single subunit - changing the atom scale, it froze. I don’t know how much it’ll help if I add 16GB more RAM (mine is Win 7 Ultimate); the thing is I’m hesitating to spend more on an old build while only get little improvement.

Upgrading my cpu would require a new mobo, I guess. My current video card is a low end AMD Radeon HD 6670. That’s why I’m thinking about building a new system: dual Xeon cpu vs single i7 CPU, and what video card is sufficient? How do I tell if I’m viewing the results in a viewport rendered with OpenGL? Thanks again!