Ansys HPC Pack - Ansys Licensing for CPUs and GPUs Explained

Ansys has a quite complex licensing format where enabling more compute nodes or higher performance hardware will require more licensing. Ansys takes into account the number of cores that are enabled for a specific workload called HPC Packs. Price per license fluctuates so we will not be including price here.

What are Ansys HPC Packs?

HPC Packs are license(s) that define the total supported cores enabled for your Ansys calculations. The quantity of HPC packs purchased determines the designated number of CPU cores you can use in a single calculation.

The licensing model typically involves purchasing a certain number of HPC Packs for a single system, which then allow a determined amount of compute to be used. The more HPC Packs purchased, the more cores enabled for faster performance. Determining the number of packs required depends on factors such as the size and complexity of the simulations being performed, as well as the desired speed and efficiency of the analysis.

The number of cores enabled by an HPC pack is calculated with this simple formula where h = number of HPC Pack(s). These cores often refer to physical CPU cores or GPU Streaming Multiprocessors (SMs). More on SMs later. There is also a default of 4 solver cores offered regardless of HPC packs purchased. The formula for total cores offered in a specific HPC pack is as follows, where h is the HPC pack tier:

Therefore, the table is as follows. We will only go up to 5 but you can access the next tier for more allocation if necessary, using the formula above.

CPU Core Count & Ansys HPC Pack Matrix

Ansys HPC Packs are an exponential function gain so the increase from 20 to 2000 cores wouldn’t incur a 100x price cost. These total cores mean that even if your system houses more cores than the given HPC pack amount, those cores will be inactive for the calculation. For example, an individual with 2 HPC packs can only use 36 cores, so if your system has 48 total cores it would only use 36 cores with the rest left idle.

Organizations can choose to split up or consolidate their HPC pack licenses. However, the calculation of cores per system differs. For example, an organization that wants 4 HPC packs distributed between 2 separate and individual systems (2 Packs + 2 Packs) will have 36 + 36 = 72 total cores. If they did not distribute the packs and consolidated them onto a single system 4 HPC packs would yield 516 cores.

Individualized workstations can benefit organizations that have multiple departments running simulations and don’t need/want to have to manage compute resources. A consolidated system can benefit organizations that run large simulations from time to time that require the massive compute. Users still can distribute their simulations across the multiple processors or computing nodes but requires defining solver cores. Both methods accelerating the simulation process, enabling them to tackle engineering problems. This can lead to increased productivity, faster time-to-market for products, and better optimization of designs.

Ansys HPC Packs for GPUs

Ansys HPC Packs don’t just define licensing cost CPU; they also define the cost when running Ansys on GPUs. Many Ansys simulation applications are still CPU reliant but, GPUs are paramount in modern high-performance computing, with some Ansys simulation solvers accelerated or native on GPUs including certain simulation solvers like Fluent, Rocky, and more.

GPU microarchitecture is different than CPUs housing hundreds of times more cores than CPUs. So instead of using core counts as the variable, they use SMs or Streaming Multiprocessors. Not to go into full detail about GPU microarchitecture, but Streaming Multiprocessors or SMs are the building blocks of a GPU that store multiple cores, cache, and controllers. Think of SMs as a group of workers tasked with computations, memory management, instruction pipelines and whatnot and a GPU has a factory of these groups. These SMs enable parallelized computing thus, are what Ansys counts as “cores” in this case.

Finding the SMs of a GPU is not as easy as looking at the spec sheet. We listed current-gen and popular GPU models and their SM counts found by scouring GPU microarchitecture whitepapers and 3rd party sources like techpowerup.com. We will also compare how many HPC packs would equate to each GPU. We also added the GPU memory to help gauge which GPUs suit your potential model size (this does not affect the HPC pack licensing).

Unlike how HPC packs work for CPUs if the GPU configuration exceeds the SM count threshold for HPC packs, additional purchase of the next tier up is required since SMs cannot be disabled by will. Therefore, depending on configuration, careful consideration of single or multiple GPUs is necessary for optimizing each HPC pack license. Some GPUs should be prioritized over others on this list since they maximize the number of available SMs per HPC Pack.

For example, a single RTX 5000 Ada (100 SMs) requires 3 HPC packs (up to 168 SMs). The faster and larger RTX 6000 Ada (142 SMs) would also cost 3 HPC packs. By paying the price of the higher tier card, you can maximize the price you’re paying for Ansys HPC pack licensing while getting better performance. But for multi-GPU configurations, the story is a little different:

In a 4x GPU configuration, the RTX 6000 Ada may not be a smart investment due to the extra license needed to be purchased for you to be able to run it. Be strategic with licensing by maximizing per core or per SM limit without spilling over to the next bracket. The RTX 5000 Ada may be a bit slower, but if the simulation size calls for 4x GPUs, purchasing an extra GPU without the need for an extra license can influence budget and performance considerations.

Also, for 4x GPU configurations, the NVIDIA A800 only requires 4x HPC packs and has native double-precision FP64. The A800 is a more expensive GPU compared to the RTX 6000 Ada, but the A800’s FP64 capability, HBM2e GPU memory, and an unneeded 5th HPC pack can influence decisions depending on the type of simulations and solvers. Of course, there are other considerations like GPU performance between RTX 6000 Ada versus the NVIDIA A800, benchmarks coming soon…

Conclusion

Accelerating computations in Ansys is paramount for engineers in perfecting the product design before putting it on the assembly line and eventually in the hands of the consumer. By running simulations faster, refinements can happen quicker, and innovative products can go to market sooner.

Configuring a high-performance Ansys system has many factors including hardware, licensing costs, performance, and infrastructure. Optimizing price to performance is essential for maximizing value in a large investment like this. Exxact is here to deliver the computing systems and tools for enabling reduced simulation time and success. If you have any questions regarding hardware configurations, contact our team and we will do our best to help guide you in the right direction.