Three Ways to Get the Most Out of Your CAD Workstation13 Mar, 2014 By: Thomas A. Salomone
Hardware configuration may not be a top-of-mind concern, but making educated choices when you choose your next workstation can substantially improve your productivity.
The next consideration is cores. Even though the kernel in the CAD application is single-threaded, other parts of a CAD application — such as the part retrieval or the user interface — are multithreaded. I recommend you choose a machine with a minimum of four cores for this reason; more cores will be necessary if you plan to run multiple CAD applications simultaneously.
Now, suppose your application is computer-aided engineering (CAE), rendering a concept design, or a computer-aided machining (CAM) application. What do you do then? These applications do not have the same type of kernel as CAD; threading takes place within these applications. Threads can run simultaneously, and cores are more important than speed. Though speed is always important, the priority should be on maximizing cores, which requires using a dual-processor system. When sizing your system for CAE, concept rendering, or CAM use, always look to maximize cores rather than simply getting the fastest processor.
Intel also offers a technology to improve the performance of dual-processor systems. Called Hyper Threading, it allows the cores to process two threads simultaneously by taking advantage of unused compute cycles. Think of a car trying to pull out on a busy road — it waits for an open spot, then enters the flow of traffic. Hyper Threading uses cache in the processor in a similar way to manage data traffic and submit threads when there is an open spot in processing. When considering the use of Hyper Threading, there is no hard-and-fast rule, as some applications are already streamlined and the use of Hyper Threading may actually slow down their operation. It’s important to know if your application provider recommends Hyper Threading; follow the provider’s directions on its use to avoid issues that could negatively impact performance.
A graphics processing unit (GPU) technology that uses graphics card architectures has recently emerged. It is important to consider this when configuring equipment for CAE, rendering, or other scientific applications, because these graphics cards are not used for display, but are separate processing units, and the application must be ported to GPU technology to utilize these cards and take advantage of this technology. When using these cards, the application and model are loaded into the card and run in the background so the rest of the system can be used for other tasks. You could also choose to dedicate the CPU processors and the graphics processor to run the analysis as well to get the maximum output. These cards (NVIDIA Tesla cards, for example) are often much faster than multiple cores alone. You will need to check with your application provider to determine whether the applications you run will support them.
Master Your Memory
With your processor strategy set, it’s time to configure memory. For CAD, the simple rule is to make sure you have at least twice the maximum model size as measured in memory — but that’s the bare minimum because CAD files continue to increase in size over time, users add more to the design detail, and the CAD applications themselves become more complex. CAD model sizes tend to double every 2 to 3 years, so allow for growth to ensure your system is equipped to handle the load.
When buying memory, it is important to have one dual in-line memory module (DIMM) in every memory channel – for example, if you have a processor with two memory channels (Xeon E3, i7, etc.), you will buy your memory in groups of 2 DIMMs. If it is a four-memory-channel processor (e.g., Xeon E5), buy your memory in groups of 4 DIMMs.
For CAE, rendering, and other simulation-heavy uses, you’ll need memory to support each thread running in the CPU. For some simulations, such as FEA, this is significant and can require upwards of 4 GB per core. Check to determine what your application provider recommends. For non-FEA applications, CAE and rendering memory could typically be in the 32 GB to 64 GB range.
When choosing a graphics card, note that your selection depends on usage and model size. The larger the model, the more graphics memory is required to hold it, and the more graphic processing power is needed to process it. Graphics cards enable smooth movement on the screen and enhance the quality of the image, both of which are critical to increasing productivity.
Model size typically drives graphics card selection, but the number of pixels used is also important: The larger the number of screen pixels, the more graphics power you’ll need. Creating imagery that includes photorealism and lighting effects requires high-end graphics performance as well. Typically 3D CAD users who are not doing rendering can use midrange graphics, while those working with large models, renderings, or photorealism may need to step up to a higher-end card. If you are spinning a model and onscreen movement is slow, this is a clear sign that you need more graphics power and graphics memory.
Configure to Meet Your Performance Needs
As you can see, configuring a workstation isn’t a simple process. It requires careful consideration of your needs and configuring your system to ensure you’re getting the best performance – and, as a result, the highest productivity. It’s a complex process, but one that can be greatly simplified with a little understanding of how configuration impacts application performance and what additional features you can leverage to give your workstation a boost. With the right configuration, you’ll be positioned for improved productivity and business success.
In her easy-to-follow, friendly style, long-time Cadalyst contributing editor Lynn Allen guides you through a new feature or time-saving trick in every episode of her popular AutoCAD Video Tips. Subscribe to the free Cadalyst Video Picks newsletter, and we'll notify you every time a new video tip is published. All exclusively from Cadalyst!