PNY has submitted this article.
The way designers and engineers work is undergoing drastic changes. Industries utilizing employees as computer-aided design (CAD) and computer-aided engineering (CAE) specialists face many challenges including increasing workflow complexity. Data-intensive, multi-application workflows are now common, and CAE simulation is being performed much earlier in the design process. In addition, the volume and size of datasets increase as designs become more complex. Generating photorealistic images directly from CAD files for customer research or marketing and sales purposes is increasingly common.
Work environments continue shifting. Now distributed teams and remote workers are the norm, which increases the need for real-time, virtual collaboration. An essential requirement in today’s business reality is the fact that engineers and designers must be able to work from anywhere. There is also a growing demand for faster visualization iterations.
Emerging technologies are rapidly advancing. The growing use of generative design drives innovation while the demand for AI-enhanced solutions is intensifying. Rich, immersive experiences including augmented reality (AR) and virtual reality (VR) drive the demand for extremely high-performance graphics and rendering capabilities supplemented by artificial intelligence (AI). This article describes why accelerated computing with the latest professional GPU products is so effective for AI and machine learning (ML) applications.
The GPU Advantage Over Central Processing Units (CPUs) for AI/ML
Using CPUs for AI-driven tasks or capabilities is simply not viable compared to GPUs. GPUs are replacing CPUs for AI/ML because GPUs have parallel processing capabilities and can run massive numbers of computational threads simultaneously. NVIDIA RTX professional GPUs have Tensor Cores dedicated to accelerating the AI matrix multiplication process, used for deep learning (DL) training and inferencing.
The leading open source and proprietary AI solutions for engineers require GPUs to function. AI/ML and DL would not be viable without the compute and graphics power delivered by GPUs, used in engineering CAD, CAE, graphics, simulations and accelerated visualization.
GPU Features That Support AI
Organizations must consider their unique needs when selecting GPUs. GPU solutions must be AI-enabled and have these capabilities. The latest NVIDIA RTX products, available from PNY, are enhanced for AI-enabled CAD, CAE and 3D product design applications. The new generation of NVIDIA GPU products includes NVIDIA RTX 6000 Ada Generation, NVIDIA H100, and NVIDIA L40 data center GPUs.
CAE-enabled engineering simulation: Engineering simulations require GPUs with large cache and memory to run 3D design and real-time simulations simultaneously. The NVIDIA® RTX 6000 Ada CUDA® Cores and a larger L2 cache improves compute performance for faster, more complex, higher fidelity engineering/CAE simulations. The RTX 6000 Ada is a powerful workstation GPU offering high-performance, real-time ray tracing and AI-accelerated visualization or photorealistic rendering. Optimized for today's professional workflows, the RTX 6000 Ada combines 142 third-generation RT Cores, 568 fourth-generation Tensor Cores, and 18,176 CUDA Cores with 48GB of GDDR6 ECC graphics memory. Ada-generation shaders provide a significant graphics performance boost to enable smooth, real-time interaction with larger, more complex models and assemblies.
AI-enabled applications can now handle large CAD datasets: AI applications need GPUs with large amounts of GPU memory for AI-enabled, multi-application workflows or AI-application development systems. Powerful GPUs and memory subsystems allow designers to work with complex photorealistic renders for faster product design iteration and project reviews.
GPUs with large cache and memory are required to run 3D design and real-time simulations simultaneously. 48GB of GPU memory in the RTX 6000 enables creators to work seamlessly with larger 3D datasets and scenes without having to reduce design complexity or display resolution. The GPU also provides improved graphics performance to visualize photorealistic results of complex simulations.
Data center workloads: Data centers require GPUs that can handle various workload capabilities. The NVIDIA H100 Tensor Core GPU is designed for 24/7 data center operations requiring reliable power-efficient hardware. The NVIDIA H100 provides large-scale AI and high-performance computing (HPC) with unprecedented performance and scalability. The product can be partitioned down to right-sized Multi-Instance GPU (MIG) partitions for multiple users running smaller tasks. Hopper Confidential Computing secures sensitive applications on shared data center infrastructure. The NVIDIA H100 includes the NVIDIA AI Enterprise software suite to streamline AI development and deployment across large organizations, including engineering and design groups.
Accelerated visualization: Organizations need GPUs with high-performance capabilities to support simulations, speed up data analysis and support scientific research. The NVIDIA L40 GPU is uniquely suited to provide the visual computing power engineers require from modern data centers. It provides extreme performance and scalability to meet the diverse computing demands of complex workloads. It uses third-generation RT Cores that enhance real-time ray tracing capabilities. Fourth-generation Tensor Cores with support for the FP8 data format can deliver over a petaflop of inferencing performance. The GPU uses the latest CUDA Cores and 48GB of GDDR6 ECC graphics memory to accelerate visual computing workloads on high-performance virtual workstation instances. The L40 also supports visual computing requirements of large-scale digital twins in NVIDIA Omniverse Enterprise.
Ansys Car Air Flow Simulation
The NVIDIA RTX 6000 GPU is used in high-end design, manufacturing and product development CAE simulations. Ansys used the RTX 6000 for wind tunnel analysis of air flow over a car using their Ansys Discovery application in real-time. Ansys Discovery is built on NVIDIA CUDA technology running on NVIDIA RTX GPUs.
The external aerodynamic simulation looked at flow inlet, pressure outlets and wall boundary conditions. The Ansys software allows “what if” analysis and scenarios at the speed of thought using the FP32 compute performance, visualization capabilities and 48GB GDDR6 ECC memory. The Ansys engineering external air flow simulation ran faster, with over 7.5 hours saved on a typical run with >10 million cells.
Meeting Advanced AI/ML Needs with the Latest GPUs
There are many challenges facing CAD and CAE professionals, and the way engineers and CAE specialists work is rapidly changing. Workflows are data intensive, and the size and volume of datasets have increased. Simulations are now performed significantly earlier in the design process. The use of generative design, computer simulations and technologies such as AR and VR requires the use of high-performance GPUs and AI-enabled solutions. GPUs with parallel processing capabilities are required to run massive amounts of computations simultaneously, in order to meet advanced graphics and AI/ML needs. Click here for an ebook describing how NVIDIA GPUs aid in graphics acceleration in CAE applications.
The latest NVIDIA RTX professional GPUs, available from PNY, meet the requirements for today’s professional workflows including unprecedented AI/ML and 3D design/simulation processing performance. These GPUs provide advanced graphics capabilities including rendering and tracing to deliver photorealistic images.
For more information on how PNY can help in choosing the right NVIDIA RTX GPU for CAD and CAE engineering workstations, visit www.pny.com/manufacturing.