October 19, 2022 Maddie Quach

Look deeper into pricing of render farms

We already have an overview article about the factors to consider before choosing the most suitable render farm. Leaving aside the unique advantages and disadvantages of each render farm on the market, today we will only discuss the issues related to calculate pricing of render farms, which any user should know and understand. So in today’s article, let’s see how render farms calculate prices, and compare with our iRender pricing.

Understand basic unit and definition

First, let’s familiarize ourselves with the basic unit and definition when you’re referencing the price of any renderfarm.

  • GHZ-HR

Consider a computer with a single CPU and only one core. Set the time machine to the past and assume the core runs at exactly 1GHz. If your scene file is  renderedon that machine for an hour, the consumed “compute time” is referred to as 1 Ghz-hr.

Obviously, our computers have many more cores than one, and they all run at much higher clock speeds than 1Ghz, but the concept of a Ghz-hr remains relevant. We multiply the number of cores by their clock speeds, by the render time on that CPU, and by the power rate you specify when you submit the job ($/Ghz-hr). That is the overall cost of your project.

The Ghz-hr charging method is the most widely used by render farms around the world because it is the most reliable for asymmetric farms with varying core counts or clock speeds per node. It’s also a fair way to calculate price, assuming all other variables are equal.

  • OB hour

Octanebench is an Otoy GPU benchmarking utility designed to define the overall compute power provided by any combination of graphics cards in a computer. When you run this utility, it generates a combined “score” for the system being benchmarked, which is denoted as Octanebench points. If a system has 100 Octanebench points and runs for one hour, that system delivered 100 Octanebench points of compute power per hour. By summing their respective Octanebench points and multiplying that by the number of hours those systems were running, any combination of systems and graphics cards can be said to deliver X units of GPU compute power.

Finally, after multiplying the total Octanebench points used (by a GPU-rendered job) by the rate that corresponds to your selected Power level, we have the total job cost.

Remember that Octanebench points per hour is a measurement, just like CPU Gigahertz per hour. It is not a made-up or fictitious number.

Look deeper into pricing of render farms 1

Some other concepts you also often encounter when learning about the price of render farms:

  • Priority is a price plan you can choose to render a job. You have a choice of Low, Medium, and High. Each priority offers you a different set of features and has a different rate. The higher the priority, the faster the rendering.
  • Node is a computer designed and built for network rendering. You can think of it as a specialized server.
  • Node/ hour is just like the OB hour but the difference is it represents the cost for a node and the rate is usually rounded up. It is a price for 1 hour of rendering on a single GPU node.

The problems behind these popular pricing calculation

Accuracy

It’s somehow strange how some render farms can estimate the cost of your project when they don’t even know the PC configuration you are using, just based on the number of frames and the average time it takes a frame to render. The calculator displays the farm render time in the best-case scenario, where your job receives the maximum number of available nodes; however, this number may vary in practice. Furthermore, the calculator only displays the time when the nodes are rendering. It does not include the time spent waiting in the queue or the time required for a scene to open/load on the render node.

We can mention scene/asset loading, pre-processing, rendering, and output saving, all of these factors have an impact on the time it takes to render a project. And any bad scenario could happen when there is technical issues – This problem can come from the service provider when the software they provide has problems, or it can also come from you with settings errors or problems with missing textures, …

Perhaps that is the reason why there will always be a disclaimer in any pricing tab of any render farm which uses Ghz-hr or OB hour as pricing unit. It could be mentioned like the generated results are only estimates and do not reflect the actual cost of a project. Specific factors, such as scene loading time or pre-processing, are unpredictable and thus are not included in the estimation. The difference between the estimate provided here and the job cost on the farm is not grounds for a refund.

Furthermore, the calculator of render farms actually only works with animations, so with a still image, this tool is quite useless.

Transparency

When it comes to hiring external render farms you only pay whilst you’re using their services. Different providers have slightly different methods to calculate the fees but they are all based on how much computing power they are selling to you per hour. However, despite being public about the configuration used as well as the number of nodes assigned to a project when the customer selects a priority, the customer cannot actually check this to ensure that they are paying a fair price for what they get.

Difficulty in price comparison when referring to render farms

As analyzed, due to the inaccuracy of pricing methods as well as price estimates, customers will face many difficulties when comparing prices between render farms to come up with a suitable plan. Even in the best-case scenario and the supplier offers a fairly accurate price estimator, it is possible for the customer to have problems when they give an incorrect Octane Bench number. This is related to the RTX on or RTX off mode when using Octane Bench. At the same time, different lines of graphics cards also produce different OB results. In general, inaccuracy is inevitable, so you should be a conscious customer when making decisions based on cost estimation.

Also, some render farms are having extra for a higher priority which means that you can start your rendering faster than others, and some farms are more expensive because they have superior 24-hours support.

iRender - Uniform and transparent pricing

iRender always considers our model – IaaS render farm to be an advantage because it always provides customers with the most transparent service. With only one pricing based on the time the customer connects to the servers that iRender provides (pay as you go). As we all know, depending on the specific project, there are always different properties, so the most accurate way to calculate the price is to actually render a few frames on our system to see how long it takes, thereby making decisions like upgrading server package (without re-setup working environment) or using more machines at the same time to keep up with the project schedule.

In addition, by providing customers with a separate working environment based on bare-metal server (physical computer server) technology, customers will have the following benefits:

  • Check the benchmark score of all software customers use on the server
  • Check the accuracy of the configuration published on the website.
  • Take full control of your working environment and utilize 100% of the performance of our graphics cards as well.

We provide high-configuration dedicated servers (physical machines/computers) that are specifically configured and optimized for GPU rendering. Our GPU machines are built with 1/2/4/6/8x RTX 3090 and 2x RTX 4090 (the new server package that we just released). Also, they all have a strong AMD Ryzen Threadripper Pro 3955WX processor, which provides increased performance together with GPU. Moreover, the workstations are ideal with 256GB RAM capacity which is a great amount for most users, even high-end ones and 2T NVMe SSD storage that meets all your small to very huge projects’ files. 

Enjoy your work with one of the most anticipated graphics cards of 2022! Currently, iRender has launched the dual GPU RTX 4090 server package, and in the near future we will launch multi-card configuration packages with even more great power.

Final words

In general, ‘Expensive or cheap’ is actually only relative to a project, because sometimes what you need most is to meet the deadline, no matter how affordable a service is, if it does not meet the demand or affects the project schedule, it is too expensive.

We hope that this article will bring you useful information and help you to find a suitable cloud rendering service with suitable pricing of render farms!

Register an account today for FREE TESTING COUPON to experience our service. Or contact us via WhatsApp: (+84) 912 785 500 for advice and support.

iRender – Happy Rendering!

Related Posts

, , , , , , , , , , , , , , , , , , ,

Maddie Quach

Hi everyone. Being an Customer Support from iRender, I always hope to share new things with 3D artists, data scientists from all over the world and learn from them as well.
Contact

INTEGRATIONS

Autodesk Maya
Autodesk 3DS Max
Blender
Cinema 4D
Houdini
Daz Studio
Maxwell
Nvidia Iray
Lumion
KeyShot
Unreal Engine
Twinmotion
Redshift
Octane
And many more…

iRENDER TEAM

MONDAY – SUNDAY
Hotline: (+84) 912-785-500
Skype: iRender Support
Email: [email protected]
Address 1: 68 Circular Road #02-01, 049422, Singapore.
Address 2: No.22 Thanh Cong Street, Hanoi, Vietnam.

Contact
[email protected]