Minimum Hardware Requirements
CPU | Any x64 (AMD64/Intel 64) or compatible |
RAM | 2 GB |
Disk Space | 3 GB |
Graphics Card | Hardware Acceleration + OpenGL 1.2 or higher |
Network | Internet Connection Required |
Please keep in mind that the listed system requirements in this section represent the minimum specifications. If you plan to work with large meshes and numerous result files, additional RAM and hard disk space will be necessary.
Recommended Hardware Requirements
CPU | Most important component for CFD calculations (better CPU better performance) |
RAM | 32 GB - for models under 10M nodes |
Hard Drive | Dual Drive Setup: |
Graphics Card | Any 3D Graphics Card with 4GB or more memory |
Network | Internet Connection Required |
In the following sections you will find our hardware reccomendations suitable for CFD analysis, listed in order of relevance for:
CPU
When it comes to CFD calculations, the performance of the CPU is critical for achieving results quickly.
In CFD simulations, substantial data transfer occurs between the memory and the CPU. While the CPU is typically capable of fast computations, it often spends a significant amount of time waiting for new data. It is crucial to maintain a balance between the core count and the number of memory channels. A commonly suggested guideline is to have approximately 2-4 CPU cores per memory channel. Both a larger cache and an increased number of memory channels (and therefore memory bandwidth) play significant roles and may become limiting factors. For a smaller model, a higher base clock frequency provides better performance per core and scales better.
Selection Criteria for CPU
When selecting CPU, it is best to focus on parameters in this order of importance:
- Memory bandwidth (memory channels)
CFD calculations do not require intricate computations on the processor but involve a substantial volume of data. Hence, more critical than the processor’s individual parameters is the memory bandwidth. This ensures the swift delivery of necessary data for calculations. Memory bandwidth is directly associated with the number of memory channels which allows for data transfer in parallel. - CPU cache size
Cache is a type of memory that a processor uses to store frequently accessed data. A CPU with a larger cache can help speed up CFD calculations by reducing the amount of time it takes to retrieve data. - Number of cores
If you plan on performing simulations of large models, then having a processor with multiple cores can significantly improve the performance of your CFD calculations. However, it’s essential to align the number of cores with the maximum memory bandwidth to ensure there is no limitation to the exchange of data. Often, memory bandwidth tends to be the limiting factor, and the potential of multiple cores is not fully utilized. The suggested guideline is to have 2-4 cores per memory channel. Hyper-threading is not beneficial in CFD calculation. - High clock speed
A less crucial factor appears to be the processor’s frequency. The bottleneck in CFD calculations lies in the information flow, where the majority of time is consumed. Processor operations, being relatively straightforward, typically do not constitute a time-intensive aspect.
SimFlow CPU Reccomendation
Some specific CPUs that are often recommended for CFD calculations include the Intel Core i9 and Xeon processors and the AMD Ryzen and EPYC processors. However, the specific CPU recommendation will depend on the type of CFD simulation you are performing and your budget.
The best CPUs for running long computations are usually server-grade CPUs, like AMD EPYC or Intel Xeon. We recommend taking a look at the AMD EPYC series:
AMD EPYC 4th Generation
12 Memory Channels and up to 460 GB/s Memory Bandwidth
In this generation, we recommend:
AMD EPYC™ 9354 (32 cores and 256 MB L3 Cache) ~ 3500$
AMD EPYC™ 9384X (32 cores and 768 MB L3 Cache) ~ 5000$
AMD EPYC 3rd Generation
8 Memory Channels and up to 204 GB/s Memory Bandwidth
In this generation, we recommend:
AMD EPYC™ 7373X or AMD EPYC™ 7473X ~4000$
AMD EPYC™ 74F3 ~3000$
The AMD EPYC 3rd generation has a worse performance but is compatible with older and more affordable components, such as AM4 motherboards and DDR4 memory.
RAM
RAM is essential for storing the simulation data and software instructions required by the CPU to execute calculations. The necessary RAM for CFD calculations depends on your simulation’s size and complexity. Typically, for simulations with fewer than 10 million nodes, 32 GB of RAM should be adequate. For those running multiple analyses simultaneously on a single computer, increasing the memory to 64 GB could be advantageous. For models, exceeding 10 million nodes, we recommend allocating 2 GB of RAM for every 1 million nodes.
To maximize performance, we advise using one RAM module per memory channel, regardless of the total RAM needed. For example, if the objective is 64 GB of RAM and there are 4 memory channels, 4 x 16 GB modules would be ideal. RAM latency has a minor impact on calculation speed. If the RAM size is not limited for an evaluated case, reducing RAM latency can speed up calculations by a few percent.
If the simulation demands more RAM memory than currently available on the machine, the data will be stored on the hard drive. This will substantially slows down calculations and extends processing times. In such cases, upgrading the RAM is recommended.
SimFlow RAM Recommendation
32 GB - for models under 10M nodes
64 GB or more - for models over 10M nodes (minimum 2 GB of RAM per 1M nodes)
1 RAM module per 1 memory channel
Hard Drive
We advise implementing a dual drive setup to maximize performance: one drive for the operating system and another for storage.
For the main system drive, we recommend a Solid State Drive (SSD) or M.2 drive, which offers significantly faster read and write speeds than traditional Hard Disk Drives (HDDs). For storage, choose a drive with a large capacity, as CFD simulations tend to produce a lot of data. In this case, HDDs are a more economical option.
SimFlow Hard Drive Recommendation
Dual Drive Setup:
Main System Drive - SSD/M2 1TB
Storage Drive - HDD 8TB (or more depending on the individual needs)
Graphics Card
SimFlow CFD computations do not rely on GPU processing. Therefore, any 3D graphics card with at least 4GB of memory should suffice for post-processing tasks, depending on your budget. There’s no requirement to purchase an expensive Quadro card since those are particularly designed for CAD applications.
SimFlow Graphics Card Recommendation
Any 3D Graphics Card with 4GB or more memory