What do workstations for 3D modeling and design typically look like? If you’re building your own 3D workstation for CAD, engineering, CGI, and scientific modeling, you’ll want to understand how your hardware relates to specific facets of the software that you use. Here we’ll discuss how your CPU and GPU specifications affect performance during design and rendering. This will clarify the type of components you will need to run the tools and applications you use to create. Included are recommendations which you may use as a starting point for selecting computer parts for your workstation. Here’s what you’ll want to know.
Most 3D applications are single-threaded for designing. This means that CPU clock speed determines performance for the workstation when designing. In the design phase, a workstation will not benefit much from a processor with a high physical core count. Having a CPU with turbo features tends to make a positive impact on performance in design phases—so if you’re looking for a reason to overclock at work, this is it.
Rendering, on the other hand, is a different process that utilizes multiple CPU cores and threads. A rendering engine takes advantage of multiple cores and threads on a CPU. Many pro design applications—3ds max, Maya, Blender, Cinema 4D, among others—have built-in external render engines in them. Unlike 3D design and animation, a rendering engine takes advantage of multiple cores on a CPU.
So is it impossible to optimize a CPU for all facets of the software—both design and rendering? Not really. In years past, you couldn’t get high clock speed and lots of cores in one CPU. That’s not the case today. Current-generation Intel Core i7 and Core i9, and R5- and R7-series AMD Ryzen—so-called High End Desktop (HEDT) processors—are jack-of-all-trades CPUs that deliver multiple (4, 8, 10) cores plus a turbo-boosted clock speed upwards of 4.0 GHz.
|Base Clock (GHz)||Turbo Clock (GHz)||Cores / Threads||Price (March 2018)|
|Intel Core i9-7900K||3.3||4.3||10/20||$930|
|AMD Ryzen Threadripper 1920X||3.5||4||12/24||$700|
|Intel Core i7-8700K||3.7||4.7||6/12||$360|
|AMD Ryzen 7 1800X||3.6||4||8/16||$330|
|Intel Core i5-8600K||3.6||4.3||6/6||$240|
|AMD Ryzen 5 1600X||3.6||4||6/12||$200|
CG design applications that utilize the Arnold renderer work best with a processor with SSE4 support, which almost every new CPU does. The earliest CPUs that support SSE4.2 include AMD FX-series and Operteron, and Nehalem Core-series Intel CPUs. The past few years have seen a shift in software programs for CG which started taking advantage of the graphics processor (GPU) a workstation. In 3D graphics production, GPU architecture speeds up processor-intensive work like rendering or scientific simulation.
The mental ray rendering application integrated in many pro-grade 3D CG suites (Maya, 3D Studio Max, AutoCAD, Revit, Cinema 4D, SolidWorks, others) is owned by NVIDIA. Most 3D software takes advantage of the CUDA cores in GPU architecture to optimize rendering and scientific simulation. Some GPU-render engines require CUDA architecture, notably Octane, Redshift 3D, and VRAY-RT.
Does this mean you might only use NVIDIA for 3D design? Not necessarily. The Stream/OpenCL processors inside AMD graphics processors aren’t much different from NVIDIA’s CUDA cores. Both are designed for executing parallel programs. Stream cores are smaller and run at a lower frequency; CUDA cores are comparatively larger and run on a higher frequency. Benchmarks, in general, are a close match for cost-performance ratio for graphics work. The real edge that NVIDIA has comes from its popularity with developers, and thus a bigger library, more code snippets, and available resources for creators. Also, NVIDIA cards improve their running OpenCL framework every generation.
What are ideal specifications for GPU hardware used in a 3D design capacity? This generally is a bang-for-buck consideration. As of this writing, that consideration is thrown off kilter by cryptocurrency mining driving demand for GEFORCE gaming graphics cards. Historically, the NVIDIA Quadro family of processors is branded as a workstation graphics card geared toward engineering and CGI professionals. Newer Telsa architecture improves upon Quadro for this purpose.
Without getting too deep in the weeds, (deep dive here!) drivers for workstation GPUs offer precision and accuracy for creating graphics and GeForce offers speed (framerates) for consuming them. Quadros have optimizations for CAD and 3D software utilize architecture that is unique to those cards. The lines blur at NVIDIA’s Titan series for “best” performing card for either gaming or a creative workstation. If you ever wanted to switch hit between work and play with your workstation, look toward the upper reaches of NVIDIA’s product line.
When you’re picking a GPU for your 3D design workstation, you want to pay attention to the VRAM aboard a graphics card.
The VRAM is important because it stores textures and 3D meshes. If you’re creating complex scenes, performance will benefit a card with more VRAM. A graphics card with 12 GB of VRAM processes high-poly scenes with around 200 million unique objects, according to pro benchmarks.
|VRAM||Price (March 2018|
|NVIDIA Quadro K1200||4 GB||$340|
|NVIDIA Tesla C2075||6 GB||$625|
|NVIDIA GeForce GTX 1070||8 GB||$550|
If you’re using a high-clock CPU for design work, and a GPU rending engine, the PCIE lanes on your workstation motherboard play an important role in your rig.
PCIE lanes are how your CPU communicates with your GPU in a workstation system. A high-end graphics card connects to the PCIEx16 interface on a motherboard. If you’re using two graphics cards in SLI, that’s 32 PCIE lanes funneling into your CPU socket, and you want to make sure your CPU has enough PCIE lanes to take advantage of all that processing power.
A high-clock CPU with only 16 PCIE lanes (i7-7700K/8700K) will bottleneck in this situation. You’ll want to check for a processor with 32 or more PCIE lanes. If this is your setup, check out a CPU like an AMD Threadripper (64 PCIE lanes) or X-series Intel Core i9 7900X (44 PCIE lanes).
How much RAM should you have?
A pro workstation needs a baseline of 16 GB of system memory. Most applications for 3D and CAD suggest 8 GB minimum to run. With 16-32 GB you will notice snappier performance. Just understand that maxing out the DIMM slots with system memory will not compensate for an underperforming or bottlenecked CPU and GPU.
Do you need a solid state drive?
Installing an SSD into your workstation will certainly speed up your system. A 7200 RPM hard drive will work fine. The most notable performance increases a SSD offers over a regular spinning hard drive are experienced when programs launch or during the startup sequence for the computer. A popular drive configuration for workstation incorporates both solid state storage and HDD storage. The operating system and most-used applications and working copy files under process are installed on a small (250-500 GB) SSD. The less expensive HDD stores larger files that might not currently be in use.
Keeping your workstation cool is important.
Demanding applications cause computers to run hot. The more hardware that is inside a workstation, and the harder it works, the hotter the gear. If you’re overclocking the CPU—a proven way to boost performance with 3D design applications—you will need to purchase an aftermarket CPU cooler. Having a spacious ATX computer case with case fans configured for optimal airflow is critical for keeping your PC at ideal temperatures.
What are important or specialized peripherals to note?
Note that most design applications utilize a three button computer mouse for the entire feature set. A dual monitor setup—or one extra wide monitor—gives creatives the necessary real estate to do professional editing work. If you’re designing 4K content, then a 4K monitor is requisite. When choosing a monitor for professional design, IPS-panel monitors lend themselves to better color accuracy.
Want to buy a finished system instead of building your own 3D workstation? Compare and shop here for the best prebuilt 3D design workstations.