Lets talk about your render farm


We all know about the big studios, what about smaller studios?

How big is your team?

How many render nodes do you have?

What type of nodes?

  • Mini-towers?
  • 1 or 2 unit Rack servers?
  • Blades?
  • Mini-itx boxes?
  • Desktops stacked in a closet?
  • Specialized form factor?

How are the nodes arranged/installed?

  • Rackmounted?
  • Sit on a table?
  • on wire rack shelving?
  • on a cart?
  • something custom-made?

What type of file server?

  • High end NetApp?
  • standard name brand rack server?
  • basic desktop tower?
  • cluster?
  • something else?

What type of cooling?

  • Dedicated AC?
  • Nodes spread across multiple nodes?
  • None at all?

Any pictures?


I’ll go first

How big is your team
2-3 people working on animations

How many render nodes do you have

What type of nodes?
16 Blades
9 Mini-towers
5 Mini-itx boxes

How are the nodes arranged/installed?

  • Rackmounted blades
  • others on a cart in a closet, or sitting on a table

What type of file server?
standard name-brand rack server Windows server 2016

What type of cooling?
Dedicated AC


On my side gig (my main gig we have 13 3d generalists, but don’t have a need to render, everything is for interactive media);

1 animator that scales up with contractors/freelancers as needed.

I work with 2 mechanical engineers. We use the Amazon EC2 for rendering. Not sure all the specs, be we do change based on render speeds needed. I know our cost ranges from $.03 p/hr to $3.00 p/hr per CPU depending on what type of virtual machines we specify.

We manage it organically, meaning we ramp the render nodes up and down in quantity and specs to maximize cost efficiency.

I’ve had decent render farms in the past and prefer this much more. It feels cleaner, faster and less maintenance hassle.

We tend to average a run of about 3000-8000 frames about every 2-3 months, so our demand may not be as high as some others. Our costs are factored into our budgets for each job, so we do not carry the overhead cost of growing and maintaining a render farm


We networked 3 Commodore 64s and 2 Nintendo Gameboys using high quality copper telephone wire. Our rendering speed went up 100000000%.

That’s because before, we were rendering manually with pen and paper and Texas Instruments graphing calculators. There was 3 of us, each calculating one pixel in the image (we called that “distributed rendering”).

Damn these newfangled 8-bit computers render fast! I started a render job 2 hours ago, and we almost have 22 pixels fully rendered now.

Our setup is so effective that we are considering starting a remote rendering farm for people who need HUUUGE rendering firepower.

This is an idea nobody has really done before - You can send us models plotted onto paper by fax. We’ll enter the vertice coordinates into our powerful rendering farm and get the finished wireframe render back to you in less than 4 years.

We are geniuses at what we do. :smiley:


rim-shot (“crickets chirping in the background”)…


We use amazon EC2.
not gone higher than 100 nodes yet, really not found it necessary and it gets hard to keep track of shots when they are rendering faster than they can be downloaded.
We basically do some rough estimates as to how long it should take, add on a 20% management fee and bill all rendering back to the client. no need for a render farm ever again, it’s great.

DBOX has a total of about 70 people worldwide - the London team has their own ec2 account, and east coast US had their own too.


Yeah I know a lot of freelancers and commercial studios that use cloud rendering , then pass the bill to the client

For non-commercial groups that are on a fixed budget who can’t charge clients (say a university or academic institution who produce work for staff) who are always processing something and have local IT staff on-site, or have to contend with HIPPA policies, cloud rendering may not always be an option


It’s been fantastic for us as well. I’ve even hit myself in the head for not using it 2 years earlier than we did since it was available.


How big is your team?
1 (myself)

How many render nodes do you have?
6 + Workstation

What type of nodes?
4 mini ITX Boxes with 8 core xeons and 16 GB RAM
2 Thinkstation dual CPU 8 core xeons and 32 GB RAM

How are the nodes arranged/installed?
There is no server room available, they sit on a shelf in my room.
Since they are built to be quiet, the noise level is not a problem when they are rendering. They shut themselves off if not rendering.
Deadline 10

What type of file server?

What type of cooling?
just standard PC coolers.

I am thinking about using amazon EC2 periodically, but never managed to dive into it. It always seemed to much trouble to set up and to manage for a single guy.


how does the pricing work with that?


How big is your team?

  • Only me and hiring extra people when specific skills are needed

How many render nodes do you have?

  • Main workstation and 5 render nodes/workstations

What type of nodes?

  • Desktops, focused on GPU rendering, 4 GPUs per workstation, total of 24 GPUs at the moment

How are the nodes arranged/installed?

  • Custom made render room in basement

What type of file server?

  • Synology NAS

What type of cooling?

  • Room got specific ventilation system that mostly uses external air and it is 4-5 times cheaper then using standard AC + there is huge AC for days when outside air temp is to high for 1st system.
    Saving a lot of cost on power spent on cooling over a year.

Also when farm is idle I rent/give usage to whoever need as well. Helping couple outisde studios with render needs plus students an with some personal projects around when they don’t have budget for online farms and similar.
So made this as small side project as well www.gpuoven.com