Choose the Best Workstation Components for CAD29 Jan, 2012 By: Cadalyst Staff
When it’s time to upgrade your system, be sure you’re up to speed on CPU, GPU, and RAM technologies.
Editor's note: This article was developed by Cadalyst through a sponsorship by NVIDIA and Dell.
CAD. It represents a breadth of technologies and applications that belie the brevity of the term. There's 2D and 3D design, of course, but the space also encompasses increasingly varied disciplines including engineering analysis, design visualization and product styling, and product lifecycle management (PLM) — each placing additional demands on a company's IT infrastructure.
Today's competitive environment presses a business to get more from all its available resources, be they engineers or workstations. Companies need fewer specialists and more jacks-of-all-trades — nimbler teams with diverse expertise. Investing in versatile workstation hardware that can complement a designer's broader skill set can go a long way toward strengthening the bottom line.
A workstation's versatility and performance are a function of its configuration, most importantly CPU, graphics processing unit (GPU), and memory. Making the best choice for each of these key components will depend on your key applications and the data you typically manage in your design.
Let Your Primary Software Be Your Guide
CAD Software System Requirements
Determine the minimum system requirements for your primary CAD software.
Factors such as your typical CAD file size and complexity and the level of rendering and analysis you do will take your system needs beyond the minimum. Check out BestForCAD, a new web site from NVIDIA and Dell, for guidance.
Once you've identified that app, head to the web to find its suggested minimum hardware requirements. Now, don't settle on these baseline numbers and think you're going to be happy. In most cases, minimum requirements set an extremely low bar. SolidWorks (as of this writing) specifies a minimum of 1 GB RAM, but suggests 6 GB. Well, even 6 GB isn't necessarily the best choice, depending on your budget, and especially given the amazing amount of memory that can be had for not a lot of dollars.
Not only do you need to consider which CAD application you use most, but also how. Factors that call for more than the average specs for RAM and/or graphics horsepower include working with large assemblies and complex parts (such as plant and routing design), rendering, and running first-pass finite-element and kinematic analyses.
Also pay close attention to the list of hardware that your software developer has recommended or certified. Although Autodesk isn't going to stop you from running its apps with a PC gaming graphics card, the company will tell you which cards are optimized for performance and built for reliability when it comes to supporting AutoCAD or Autodesk Inventor. Opt for a professional-caliber card such as the NVIDIA Quadro. Major CAD software vendors will help address issues or bug reports if you're running with a recommended card.
Now that you're familiar with the minimum requirements, read on to determine where — and how much — to dial up your specs.
Tailor Your Choice of CPU for CAD
The dominant thrust in CPU architecture has shifted away from accelerating single-threaded processors and toward developing parallel processing, with the latest professional CPUs moving from dual-core to quad-core and now hex-core processors. Multicore CPU approaches have proved a great avenue to gain theoretical speed-ups, but when it comes to modeling with most mechanical CAD packages, including AutoCAD, Inventor, SolidWorks, and Solid Edge, you won't benefit much from multiple cores, as those applications are not yet optimized for multicore processing. On the other hand, engineering simulation and rendering — whether performed discretely or via a renderer embedded in one of the aforementioned design packages — do tend to utilize multiple processing cores. If that’s the case for your primary rendering engine and if you spend significant time on these tasks, you’ll want to consider shifting more of your budget toward a good graphics card that can handle the workload.
Tap GPU Power for Speedups Across the 3D Modeling Workflow
Perhaps more importantly in today’s 3D environment, you'll want to pay closer attention to the GPU you're choosing to power your MCAD, rendering, and simulation applications. Today's GPUs bear little resemblance to the graphics controllers of old. Forget old-school wireframes and primitive solid models with smooth-shaded sides. Professional graphics cards offer the ability to realistically display parts and models and keep your viewports fluid even when working with complex projects. But graphics are only the beginning of what these new GPUs can tackle. These engines also offer a whole lot more processing horsepower — work that has always tended to bog down CAD pros.
First and foremost, with the right GPU in place, you’ll notice remarkable performance in terms of how your 3D software responds to panning, rotating, zooming, generating animations, and all those other tasks that push the limits of a lesser processor. Such a fundamental improvement in how you interact with your software can bring about dramatic gains in productivity and overall user satisfaction.
The latest GPUs are also rendering raster-based images in real-time that rival anything off-line Hollywood render farms could manage just a few years ago. But more than that, GPUs have broken through all the preconceived limits we'd previously set for them, taking on processing that traditionally was left to the CPU. That means CAD professionals can now leverage their investment in graphics hardware to achieve far more than anything yesterday's graphics cards could have ever hoped to achieve.
Accordingly, you're going to want to spend at least as much time considering your choice of GPU, and quite possibly a bigger portion of your budget as well. Which GPU should today's CAD pros choose? Well, most buyers will want to match the performance and capabilities of the GPU with the rest of the system — that is, an entry-caliber card for an entry-caliber workstation. Good balance, where each component hits a performance level that is supported by the rest of the system, is the best way to maximize ROI for your workstation purchase. To learn more about the return on investment for NVIDIA Quadro professional GPUs, check out the ROI Calculator tool on BestForCAD.com.
GPU Computing. Your questions about selecting a GPU should no longer be limited solely to which GPU, but also how many, thanks to GPU computing. Virtually every corner of the MCAD world utilizes the same basic workflow: design, revise, analyze, then iterate. Two common components in this iterative workflow, finite-element analysis (FEA) and computational fluid dynamics (CFD), employ algorithms that rely on heaps of floating-point arithmetic and can be run in parallel to gain significant increase in throughput. It's a similar story for raytraced rendering, a technique that is so fundamentally different from conventional raster-based rendering that it more naturally falls under the category of GPU computing than graphics. Images rendered by a raytracer immediately catch the eye, making it one of the most sought-after visual tools for product styling. With NVIDIA's GPU compute–enabled iray, a rendering engine found in popular software such as Autodesk 3ds Max, Bunkspeed Shot, and CATIA v6, designers are seeing visualization times shrink from hours to minutes. With that speed, designers can leverage raytracing earlier and more frequently in the development process, fine-tuning designs based on photorealistic feedback.
To take advantage of this new age of GPU computation, you can exploit a single NVIDIA Quadro card or you can invest in a second add-in card, be it a second Quadro or a Tesla-brand GPU optimized for GPU computation.
NVIDIA Maximus. Yet another GPU-powered option is NVIDIA's new Maximus technology, supported in the Dell Precision T5500 and T7500 desktop workstations. Maximus optimizes dual-GPU use for CAD and related professional applications, allowing for the ultimate in simultaneous, ROI-cranking visualization and analysis. Through the combination of NVIDIA Quadro graphics and NVIDIA Tesla coprocessors, you can still design while rendering or performing simulations on the same system. In the next edition of CAD Sense, we’ll dive more deeply into Maximus Technology and the revolutionary improvements it can bring to rendering and simulation performance.
Multi-monitor support. On top of all that, there's now another compelling reason to invest in two GPUs: multi-monitor support. Most workstation motherboards and chassis today support dual slots for graphics, and for CAD professionals, dual slots open the door to a lot more screen real estate. The secret's out that "multi-mon" is the single best way to improve CAD productivity.
You won't necessarily need to buy two cards to run two monitors, so pay attention to the cards you're selecting. And with NVIDIA's Quadro cards you also get the unique Mosaic and nView technologies. Mosaic Technology supports as many as eight monitors and enables seamless taskbar spanning as well as transparent scaling of any application. It works with NVIDIA nView desktop-management software, which allows you to spread your work across a single large display or multiple displays. nView’s Virtual Desktops functionality lets you launch applications on as many as 32 different desktops and seamlessly switch between them; Extended Taskbar lets you see each application button on its specific display, eliminating a single screen with an overly crowded taskbar.
If your performance demands have you buying mid-range or high-end graphics cards, you might get all the screen real estate you want with one card. But if you're much hungrier for pixels and screens than you are for polygons per second, you could opt for two less-pricy, lower-end cards.
Sufficient RAM: a Must for CAD
The size and speed of system memory (RAM) can significantly impact workstation performance — even more so than your choice in CPU or GPU, depending on the software you use and how. In other cases, your choice of system RAM actually depends on the GPU you choose because GPUs have their own internal memory; GPU memory can even free up your RAM to do other tasks.
To gain solid performance within a reasonable budget, the sweet spot for most CAD applications is in the 6–16 GB range. Do your model files tend to be large? Do you tend to run a lot of compute-intensive apps at once, such as rendering and analysis? Are you constantly seeing that hard-disk light blinking on the front panel of your current workstation? If yes, then you'll want to head toward the top end of that range, as much as your budget allows.
If you answered no to all those questions, then opt for less, but try to leave open some available DIMM slots (where the memory slides into the system's motherboard), so you may expand later. Having too much memory is never a bad thing!
And what of Error Correcting Code (ECC), an upgrade that typically allows single-bit memory errors to be detected and corrected? For most CAD applications, ECC is certainly valuable but not essential. If the added cost is modest, go for it.
Making the Right Choice
Your productivity depends on your tools, and the performance of your workstation will only be as good as the components you select. Today, that means you need to give full consideration to your choice of GPU as well as CPU, and your memory needs to be beefy enough to support all that computational horsepower. In the age of massively parallel, general-purpose GPUs, old ROI propositions no longer apply, demanding that we rethink our priorities when it comes to outfitting our next workstation.
Check out BestForCAD, a new web site from NVIDIA Dell that provides a choice of four preconfigured CAD workstations based on your design demands, including typical assembly size, part complexity, and the level of visualization/analysis you do. The site also includes an ROI Calculator to determine hours saved and payback period you can expect from a graphics upgrade. Overall, it can help take the guesswork out of configuring a system that is optimally balanced to meet your needs.