Reducing CAE's carbon footprint is good for the planet and for the bottom line.
Altair has submitted this post.
Written by: Altair
Sustainability has emerged as a key requirement in modern CAE environments. Driven by consumer demand, engineers strive to create more eco-friendly, energy-efficient, lighter and less costly products without compromising quality or durability. Besides the products themselves, boosting the efficiency of data centers where products are designed is also critical.
According to the International Energy Agency (IEA), data centers and the transmission networks that connect them account for over 2.1 percent of global electricity demand, and high-performance computing (HPC) is a significant contributor.[1] Most manufacturers have corporate sustainability goals and have pledged to reduce greenhouse gas (GHG) emissions. With increased regulation, and carbon taxes appearing in some jurisdictions, reducing CAE’s carbon footprint is not just good for the planet—it’s good for the bottom line. Fortunately, there are steps that engineering managers can take to help put their CAE environment on a path to better energy efficiency.
1. Leverage More Energy-Efficient Servers
One way to reduce CAE energy requirements is to use more efficient servers. Boosting throughput per watt can help reduce power and cooling requirements by reducing the number of servers needed to meet performance goals. Recent benchmarks conducted by Altair measured the performance of standard finite-element analysis (FEA) and computational fluid dynamics (CFD) workloads on the latest AMD EPYC processors with AMD 3D V-Cache Technology. These workloads ran up to 1.5x to 1.8x faster than on previous generation processors.[2] If a simulation runs 1.8x faster, the same result can be achieved with a 44 percent reduction in server capacity resulting in potential savings in power, cooling and data center footprint.[3] Not every application will deliver these gains, but every bit helps. When planning data center capacity, engineers can look to standard benchmarks such as SPECpower_ssj 2008 to measure energy efficiency and look for the most efficient servers that meets their performance goals.[4]
2. Improve Power Usage Efficiency (PUE)
PUE is used to express the energy efficiency of a data center. It is found by dividing the total power used by the data center by the power used by the servers themselves. While an ideal PUE is 1.0, values typically range from 1.8 to 1.2. Focusing on PUE is the next logical step after selecting power-efficient servers. Cooling is the main driver of high data center PUEs, and cooling costs can be hard to reduce. This is especially true in simulation environments where servers operate near peak clock frequencies for sustained periods.
As power requirements for top-bin CPUs and GPUs continue to increase, cooling is a growing challenge. One strategy that can help improve PUE is the use of liquid cooling. Because liquid cooling provides more effective heat capture than air cooling, power requirements can often be substantially reduced. Ongoing savings from liquid cooling can help to justify upfront investments. CAE operators can reduce overall power consumption, deploy more servers per rack to improve density and even prolong the life of components by avoiding overheating.
3. Employ Smart Scheduling
The old adage “work smarter, not harder” is alive and well in CAE data centers. Implementing a workload manager is one of the most important steps engineers can take to boost efficiency. Workload managers place simulation jobs on the servers where they run most efficiently. Schedulers seek to maximize throughput while keeping servers fully utilized.
Modern processors typically run multiple simulations per socket. Finding the “sweet spot” where the maximum simulations run concurrently before bottlenecks occur is critical to efficiency. Workload placement that considers socket and core affinity can substantially impact both throughput and energy efficiency. Schedulers such as Altair PBS Professional with green provisioning features can even shut down cluster nodes when not in use to further reduce power consumption.
4. Cloud Bursting
Most cloud providers already claim to be carbon neutral. Rather than provisioning for peak capacity on-premises, engineers can employ cloud bursting to tap more eco-friendly cloud resources during busy periods while minimizing on-prem infrastructure.
The same workload management tools that help optimize the use of on-premises resources and software licenses typically support hybrid cloud deployment models. For example, policy-based cloud bursting built into Altair PBS Professional enables CAE users to augment on-premises capacity by bursting seamlessly to their preferred public cloud.
5. Effective Use of GPUs
While CPU-based solvers are still the norm, simulation tools increasingly benefit from high-performance GPUs. GPUs can be a game changer, offering orders of magnitude better performance and power efficiency than CPUs for numerically intensive calculations. It is no accident that both the Top500 and Green500 lists of the most powerful and energy-efficient supercomputers are dominated by GPU-based systems. Some CAE centers may elect to use the cloud bursting approaches described above to tap specialized cloud-based GPUs rather than provision GPU capacity on-prem.
Modern multiphysics simulators can offload linear algebra computations to GPUs to improve both throughput and energy efficiency. For example, Altair CFD (which includes AcuSolve, a general-purpose Navier-Stokes solver) runs on the latest AMD GPUs and has been shown to deliver real-world performance gains ranging from 3x to 5x.[5]
6. Better Tools for Energy-Efficient Designs
Reducing a data center’s carbon footprint is important, but the real payoff comes from tools that deliver more energy-efficient designs. Modern structural optimization and simulation tools can help engineers realize greener, more sustainable products by minimizing material requirements. They can also help achieve longer-lasting designs that require fewer resources to manufacture and operate over their lifecycle.
Based on the idea that reducing is better than recycling, Altair helps customers design long-lasting products that require fewer resources to manufacture and operate. The Altair Enlighten Award honors advancements in sustainability, recognizing lightweight designs that boost mileage in the automotive industry and result in more eco-friendly designs.
To learn more about Altair HPC solutions, visit altair.com/hpc-cloud-applications.
To learn more about third-gen AMD EPYC processors, visit amd.com/en/processors/epyc-7003-series.
1 International Energy Agency, November 2021 – Data Centres and Data Transmission Networks – 1% of global electricity use is attributable to data centers and another 1.1-1.4% is attributable to data transmission networks.
2,3 See the article Breakthrough Computing Performance with Altair and 3rd Gen AMD EPYC Processors with AMD 3D V-Cache Technology for details. A1.8x throughput gain corresponds to roughly a 44% reduction in infrastructure – (1-(1/1.8)) = 44%.
4 See details in AMD EPYC family SPECpower_ssj 2008 benchmark results – EPYC-028.
5 Based on a comparison of an AMD Radeon Pro VII graphics card vs a processor with 32 CPU cores. See the article AMD Radeon Pro VII speeds up steady-state simulations with Altair AcuSolve.
