Like organizations in most other industries, companies that rely on CAD technology face increasing pressure to reduce lead times, cut costs, and improve productivity. Balancing those needs while still maintaining a high level of quality is challenging. Advances in CAD processor technology, however, are proving to be a way for firms to meet that challenge.
Workstations built with traditional CPU technology gave way to stations employing GPU processing resulting in a significant uptick in performance. However, an even bigger boost in processing power has come with the use of high end Physics processing cards – more commonly known as Tesla cards for the brand name of the leader in the industry. The use of Tesla cards for visualization and simulation has dramatically changed the mid and high end workstation spaces and is revolutionizing the cluster space. These cards eliminate some of the lineal gains you get from faster processors and make these workstations capable of producing far more analysis than previously possible. With lower prices and better drivers, these powerful tools are more accessible to CAD operators in companies of all sizes.
CPU vs. GPU vs. Tesla
Conventional CPU processing – a series of discrete actions performed sequentially one after the other – was limited and slow. To increase processing speeds, you had to purchase a new computer with a faster processor. The use of GPU cards – short for Graphics Processing Unit – to perform large-scale parallel problems allows certain tasks to be accomplished orders of magnitude faster. In a GPU workstation, resource intensive portions of an application that can be evaluated in parallel are handled by the GPU, leaving the CPU dedicated to run the user end of the application. The result is fast, reliable rendering presented the way CAD users require. Upgrading a GPU is straightforward, performed every few years by swapping out a PCI card on a workstation.
As effective as a GPU is, it pales next to the new generation of Tesla card processor technology. Tesla cards are designed to support specialized graphics engines that deliver the rasterizers and shaders required by professional CAD applications. They have taken the traditional GPU processor and added a supplementary chipset that specializes in the kind of high-end mathematics commonly used in the simulation CAD/CAM design employing finite element analysis.
These high powered cards can rapidly handle advanced simulations such as higher calculus, non-destructive impact testing, linear analysis of a spline bending and flexing, simulating what a part would look like as it bends, and what structural properties it has at various bend points. Performing this with a typical GPU is limited as their primary purpose is to display graphics. Tesla cards are solely dedicated to performing parallel problems rapidly, as they are optimized for calculations only.
Tesla Card Advantages
As an example, there could be a simulation that would take an hour to complete. With a Tesla card, it could be performed much more rapidly on the order of a few minutes. Increased processing speeds provide engineers with several advantages. They are better able to meet deadlines that improve end user services levels. In organizations that share workstations among several users, faster processing means stations will be more available, improving efficiency. Productivity is also improved with the added throughput.
The added memory of Tesla cards are better able to handle the larger scale problems engineers face. This increased computing power enables more rapid analysis. Running Tesla cards allows engineers to test and compare multiple design concepts before going to prototype at a much more rapid speed than typical GPUs. Users can also draw on larger data sets to develop more alternatives and more detailed solutions.
The growing adoption of Tesla cards promises to have a dramatic impact on the CAD world as users find more ways to leverage the advanced technology.