Machine for electromagnetic field analysis and deep learning | TEGAKARI, information dissemination media for research and development

Case No.PC-8278A customer who viewed the site asked us about installing a workstation for electromagnetic field analysis.

We plan to introduce a machine that uses Ansys Lumerical FDTD, an electromagnetic field analysis software, as well as Rsoft FullWave and MATLAB.
More specifically, Ansys Lumerical FDTD is scheduled to use a 32-core license, and Rsoft FullWAVE has a usage limit of 20 cores, so the number of CPU cores on the installed machine is expected to be around 32 cores.
We have also heard that you are concerned that the GUI of Ansys Lumerical FDTD feels very slow on the machine you are currently using.

We are also considering using Pytorch for DeepLearning, and we received a request for a configuration equipped with a GPU of the NVDIA Geforce RTX3090.

The conditions expected by the customer are as follows.

・CPU: About 32 cores
・Memory: 1TB or more (2TB or more if possible)
・Storage: M.2 PCIe Gen 4.0 compatible SSD
・GPU: NVIDIA Geforce RTX3090, with plans to add more in the future
・OS: Windows 11 Professional
・Software used: Ansys Lumerical FDTD 2023R1 (32 core license), Rsoft FullWAVE (soft operation upper limit 20 cores), MATLAB, Pytorch
・Budget: About 300 million yen

We proposed the following configuration according to the conditions you contacted us.

CPU	Intel Xeon Gold 6426Y (2.50GHz 16 core) x2
memory	1TB REG ECC
storage	2TB SSD M.2 PCIe Gen 4.0
video	NVIDIA RTX A6000 48GB (can be expanded to a total of 2)
network	on board (10GBase-T x2)
Housing + power supply	Tower type housing + 1600W
OS	Microsoft Windows 11 Professional 64bit

A 4-CPU configuration equipped with the latest 6426th generation Xeon Scalable series Intel Xeon Gold 2Y.A configuration with a total of 32 cores can be achieved with a single socket configuration of the Xeon W-3400 series, but the width of the memory bandwidth is also important for the electromagnetic field analysis you plan to use.Therefore, we have adopted a 1-socket configuration, which is advantageous in terms of memory bus.

If you install 2TB of memory, it will exceed your budget.Therefore, we have chosen a capacity of 1TB, which fits within your budget.

For the video card, we requested a GPU such as Geforce RTX3090, but Ansys products do not support the GeForce series.That's why we chose the NVIDIA RTX A6000, a high-end video card for workstations.
NVIDIA RTX A6000 is a video card equipped with the same type of chip GA3090 as Geforce RTX102, and is a product corresponding to RTX3090 in the workstation video card lineup.It also has 10752 CUDA cores, 336 Tensor cores, and 48GB of GDDR6 VRAM, so it can be used for DeepLearning.

Additionally, one NVIDIA RTX A6000 can be added to this configuration.Regarding the power supply capacity, we also take into consideration what will happen after adding a GPU.

The configuration of this case study is based on the conditions given by the customer.
We will flexibly propose machines according to your conditions, so please feel free to contact us even if you are considering different conditions than what is listed.

■ Keywords

・What is Lumerical FDTD?
Lumerical FDTD is electromagnetic field analysis software using the finite difference time domain method. It is possible to analyze three-dimensional structures in a wide frequency range, and it is possible to consider complex media and various materials. The FDTD method enables high-speed broadband frequency analysis, enabling phase and S-parameter acquisition.The incident wave is abundant and the open boundary is realized by PML.Intuitive operability with support for output formats and circuit connections.It is used for research and development in the fields of microwaves, optical communications, and photonics.

reference:LUMERICAL FDTD (Ansys) *Jump to external page

・What is FullWAVE?
FullWAVE is CAE software that uses the finite difference time domain method (FDTD method).Light propagation in various photonic structures can be analyzed using a high-performance, high-precision simulation engine and a comfortable CAD layout program.It is also possible to handle dispersive materials and materials with nonlinearity as material properties to be analyzed.

reference:FullWAVE Product Overview | Synopsys *Jump to external page

・What is Deep Learning?
DeepLearning is a type of machine learning that uses multilayer neural networks to perform advanced pattern recognition and prediction.Since it generally requires a large amount of data, it is considered to be an effective method when data are abundant.
DeepLearning is also widely used in fields such as image recognition, speech recognition, and natural language processing.Because it can learn complex features and relationships, it can achieve higher accuracy than traditional machine learning methods.

reference:[Special article] What is machine learning? * Jump to our owned media "TEGAKARI"

・What is Python?
Python is an object-oriented programming language copyrighted by the Python Software Foundation (PSF).Its programming syntax is simple, making it highly readable, and it also features a wide variety of components, such as libraries and frameworks, that are suitable for different purposes.A popular language for programming beginners to advanced users.

reference:Python *Jumps to an external site

reference:[Feature Article] Programming language Python Why is it so popular? --Tools to accelerate Python programming * Jump to our owned media "TEGAKARI"

■ Click here for details and inquiries about this PC case
Machine for electromagnetic field analysis and deep learning

* Please enter the name of the case or your desired conditions.