Select the Appropriate GPU for Your CAD Workflow — and Your Budget, Part 112 Jun, 2014 By: Alex Herrera
The time you invest in choosing the right graphics processing unit will pay off in increased productivity.
There's never been a better time to invest in a new workstation-caliber graphics processing unit (GPU). The industry is exploiting the raw capabilities in graphics silicon more effectively than ever before, while adding critical software and hardware design features to shape that technology into products suited to the needs of professional CAD users, at a wide range of prices.
That may be good news, but professional users want to get on with their projects and meet their deadlines, not spend a lot of time shopping for new hardware. And they definitely don’t want to waste time in their jobs because they bought the wrong tools. Fortunately, there are useful resources and rules of thumb available to guide your selection process. But first, let’s take a look at why you need a professional GPU in the first place.
The Value of a CAD-Specific GPU
It's past time to dispel the popular myth that high-end consumer-grade GPUs — often called gamer cards, because they’re marketed to video-game aficionados — are just as useful for CAD applications as those made for professional uses. Yes, the two products draw off a common foundation of core hardware and software technology. But the similarities end there, because the professional GPU draws on that core base of technology differently, in order to reliably deliver the performance and accuracy necessary for professional-caliber applications.
Specifically engineered for professionals, workstation GPUs render quickly, reliably, and with the particular visual style that users of applications such as AutoCAD, 3ds Max, CATIA, and SolidWorks expect. Their hardware pipelines implement specialized graphics engines optimized for shaders and rasterizers used in engineering and visual effects, not games. Professional GPUs are typically outfitted with more memory than gaming-focused GPUs of similar raw performance, delivering better balance for designers and engineers dealing with large datasets.
And it's not just hardware that makes graphics different for professionals; it's the driver software as well. Optimal driver design takes into account which types of drawing are requested more commonly than others. It fast-tracks the common paths, and those paths are dramatically different for games than they are for a modeling application or stress-analysis tool. The end result? Professional drivers are slower to produce graphics for games, and game drivers are slower when used with CAD applications. Many workstation manufacturers, including HP, Dell, Lenovo, and Fujitsu, outfit their workstations exclusively with professional graphics cards, primarily from GPU vendors NVIDIA (which produces the Quadro and Tesla lines) and AMD (which makes the FirePro brand).
Reliability for Mission-Critical Tasks
Of the many concerns that professional-caliber GPUs address, reliability is the most important. Several factors play into the ultimate reliability of a GPU, including certification, testing, and warranties.
ISV certification. Reliability starts with certification from the independent software vendors (ISVs) who provide the mission-critical software CAD professionals depend on. Many software providers, including Autodesk, PTC, and Dassault Systèmes, certify relatively few GPUs for use with their applications, and include (with very few exceptions) professional-brand GPUs only. For example, if you run AutoCAD, Inventor, 3ds Max, or Revit, you'll want to check Autodesk's list of certified GPU products. If SolidWorks is your principal application, refer to the list of certified cards and drivers for SolidWorks. The exhaustive testing and support that go into ISV certification ensure that your combination of hardware and software has been proven to provide optimal reliability and performance for your mission-critical application.
Qualification and testing. Professional-caliber GPUs undergo thorough chip qualification and testing. Voltage, frequency, and temperature tests, as well as life-test simulations, enable manufacturers to virtually eliminate failures in the field. The GPUs must pass the most stringent durability and electrical integrity tests, including tolerance to shock and vibration, while undergoing extensive thermal qualification, to ensure they can stand up to even harsh environments.
Long-term support. Professional GPU products typically offer extended warranties, along with high-priority response to any bugs or incompatibility issues that, however infrequent, may occur. Furthermore, professional GPU products are specifically designed to serve long, stable lifecycles to suit the longer-duration solutions that are necessary for professional applications, but are meaningless to gamers with much shorter buying cycles.
Evaluate Price and Performance
The range of products to consider from GPU vendors has gotten steadily wider. But for the vast majority of users, there are two overarching metrics to narrow down your search for the right GPU: price point and performance demand. How much performance and which capabilities do you need from your GPU to get your job done effectively? And how many dollars can you — or should you — budget toward the purchase?
The first question may be an obvious one to pose, but for the bulk of users, there is no exact answer. Professional applications tend to have an insatiable thirst for performance. And if you do finally get a machine that enables you to quickly simulate and render today's design, that's great — but what will happen on the next project? You're going to try to do more — create broader-scope models with finer-grained simulations and higher-quality rendering. You’re creating one individual part today, but will it be an entire assembly tomorrow? Plan for an IT infrastructure that will scale with a future filled with inevitably more complex projects and workflows.
As for appropriate costs, no one can speak to your IT budget but you. But if your workflow has you engaging in visual computing for the bulk of your development time — 3D modeling, rendering, and simulation/visualization — the GPU should be a top priority when it comes to allocating dollars for a new workstation.
If you spend most of your time running AutoCAD, in 2D or simple (e.g., wireframe) 3D, then a card in the $100–$300 range should suffice. Do you work with basic 3D assemblies in applications, such as AutoCAD, Inventor, 3ds Max, CATIA, or SolidWorks? Today's cards in the $500 range should suit the majority, offering optimal bang-for-the-buck performance, supported by 2 GB of memory to handle heftier models. Are you tackling large assemblies frequently, and creating high-quality renderings on occasion? Then you’ll want to think about stepping up to cards priced around $800 or so. That’ll buy you very capable performance, coupled to a bigger memory footprint (3–4 GB). And finally, if your typical day has you generating, simulating, and rendering big, complex assemblies, consider a card pairing the highest-performing GPUs available with 6+ GB of memory ($1,400 and up) for the ultimate in visual throughput and compute acceleration.
In the next part of this article, we’ll see how to compare GPU capabilities and features within your price range so you can make a selection that addresses your needs.