Engineering news
Engineers have squeezed more and more transistors onto a single computer chip. For instance, an Intel chip today has more than one billion transistors on a millimetre-sized piece of silicon. These transistors enable the central processing units (CPUs), which are often referred to as the ‘brains’ of a computer, to process instructions.
But many modern computers also feature graphics processing units (GPUs), which can be present on a graphics card or embedded on the motherboard. While the CPU is composed of just a few processing cores that enable it to handle a handful of operations in a serial or linear process, a GPU is composed of thousands of cores that can handle tens of thousands of parallel streams of processes at a time. This makes them ideal for rendering the complex 3D graphics required in gaming, which is what they were invented for.
Since then semiconductor manufacturers have opened up the parallel processing capabilities of GPUs to general-purpose computing. For instance, in terms of engineering software, GPUs have helped accelerate CAD, which could often be bottlenecked by the CPU.
“These first GPUs enabled CAD users to work more smoothly with their digital models,” said Andrew Rink, who leads global marketing strategy for AEC and manufacturing industries at NVIDIA. “For instance, it eliminated the stuttering that often occurred when panning, zooming, or rotating. This not only provided a better experience when working with design software but also helped users become a lot more productive.”
Ray tracing
One of the biggest trends in recent years that GPUs have enabled is realtime ray tracing. Ray tracing brings realistic lighting, shadows and effects to rendered graphics, and now any changes made to the 3D model can be viewed interactively in real time while users are working on their models. Software vendors that have worked with NVIDIA and other GPU manufacturers, predominantly AMD, to support GPU-accelerated ray tracing include Catia with its Catia Live Rendering, SolidWorks with its SolidWorks Visualise 2020, Siemens with its NX Ray Traced Studio, Autodesk with its VRED 2021 and, most recently, Luxion which has introduced GPU acceleration for the first time in KeyShot 9.0.
With the rise in simulation-driven design, GPUs are now also being used for simulation acceleration. For instance, AMD recently announced its Radeon Pro VII GPU for use in demanding simulation-driven design applications. Meanwhile NVIDIA’s realtime engineering simulation tools are being used by Ansys in its CAD-embedded realtime CFD tool, Ansys Discovery Live. And recently Altair announced that two of its tools, Altair AcuSolve and Thea Render, are now supported by NVIDIA GPUs to boost engineering simulation, which it claims is four times faster than on a CPU-only configuration.
A final trend that GPU manufacturers have put substantial investment into, and one that is prevalent with the Covid-19 pandemic, is remote working. The challenge here is for users to access the same high-performance technology they rely on from their office workstations. AMD has introduced its AMD Remote Workstation technology whilst NVIDIA has launched virtualisation software called NVIDIA virtual GPU, which even supports those users who work on compute-intensive workloads.
“The mobile workstations powered by NVIDIA Quadro GPUs now are pretty much reaching the performance levels of desktop machines, which certainly wasn’t the case five years ago,” said Rink. “With most employees working remotely, everyone can access the same technology wherever they are and have a seamless experience just like when they were in the office.”
Want the best engineering stories delivered straight to your inbox? The Professional Engineering newsletter gives you vital updates on the most cutting-edge engineering and exciting new job opportunities. To sign up, click here.
Content published by Professional Engineering does not necessarily represent the views of the Institution of Mechanical Engineers.