Machine learning and DFT calculation compatible workstation

After looking at our structural biology and quantum chemistry research case study No. PC-24001322, the client contacted us to inquire about purchasing a PC to use for analysis using Alphafold3 and ESM Cambrian, as well as for machine learning.
With an eye on using Gromacs and Gaussian in the future, the client requested a high-performance CPU that takes Gaussian into consideration, a GeForce RTX 4090, and 128GB or more of memory.

Reference: Case No. PC-24001322 Machine learning machine

Based on the content of the consultation, we proposed the following configuration.

CPU	AMD Ryzen9 9950X 4.30GHz (up to 5.70GHz at Boost) 16C/32T
memory	Total 192GB DDR5-5600 48GB x 4
Storage 1	1TB SSD S-ATA
Storage 2	4TB SSD M.2 NVMe Gen4
Video	NVIDIA GeForce RTX5090 32GB
network	on board(2.5G x1) Wi-Fi, Bluetooth
Housing + power supply	Mid-tower chassis 1500W 80PLUS PLATINUM
OS	Ubuntu 24.04
Others	27-inch wide WQHD LCD display

The computational performance and memory configuration have been optimized to support both GPU- and CPU-intensive processing, such as deep learning and quantum chemistry calculations.
You can stably and efficiently use research tools such as Alphafold3, scikit-learn, and Gaussian.

Alphafold3 is the main GPU

Alphafold3 is a deep learning-based protein structure prediction software that performs most of its inference processing on a GPU.
Initially, the customer requested a GeForce RTX 4090, but since that model had already been discontinued, we suggested the RTX 4500 Ada, which is in the same generation Ada series.

After that, we received a message from a customer saying, "We have confirmed that Alphafold5090 is working on a GeForce RTX 3."Uses the latest RTX 5090GPUIt was decided to do.
This information was reported by a user on GitHub.Not officially supportedHowever, based on internal testing and multiple case studies, we have confirmed stable operation in a real environment.

Reference: Alphafold3 GitHub page

CPU performance and memory capacity required for comfortable operation of Gaussian

In DFT calculations, the performance is determined by the following three factors:

• Number of cores: Directly linked to parallel processing performance.
• Clock frequency: Single-core performance
• Memory capacity and bandwidth: Ensuring stability for large-scale calculations

This configuration uses the Ryzen 16 32X with 5.7 cores, 9 threads and a maximum speed of 9950GHz. It has an excellent balance of parallel processing and single-threaded performance and is equipped with up to 192GB of memory.
In the calculation of large molecular systemsLack of memory is the cause of performance degradationTherefore, this configuration is equipped with the maximum amount of memory supported by the CPU.
With GPUWell-balanced and capable of handling high-load calculationsThis is the configuration.

Introducing the process leading up to implementation based on actual interactions

This configurationAn example of the process leading up to the proposalThere is also.
We will introduce the consultation content and the process for deciding on the configuration when introducing Gaussian.
Please take a look at this for reference when making your decisions.

Reference: [Case Study] Tegsys's proposal capabilities to respond to the concerns of researchers through "dialogue"

For those who are active in these fields Structural biology Computational Chemistry Drug Discovery Research Machine learning Bioinformatics

We can flexibly customize to suit your desired software and operational policy. Please feel free to contact us for any inquiries, from initial consultation to customization.

Keyword ・What is Alphafold3? AlphaFold3 is a protein structure prediction algorithm developed by DeepMind. It is more accurate and flexible than previous versions, and can also predict complex complexes and molecular interactions. It is used in drug discovery and structural biology research. reference:DeepMind *Jumps to an external site ・What is Gromacs? GROMACS (Groningen Machine for Chemical Simulations) is a molecular dynamics simulation package developed at the University of Groningen in the Netherlands. It is primarily designed for biochemical molecules such as proteins, lipids, and diffusive molecules. reference:Gromacs *Jumps to an external site ・What is Gaussian? GGaussian is software specialized for quantum chemical calculations, and can perform highly accurate theoretical calculations of molecular electronic structures, chemical reactions, and spectroscopic properties. It is used for a variety of purposes, including but not limited to quantum chemical calculations and molecular dynamics calculations, and is widely used in university and corporate research for molecular design and structure analysis. reference:Gaussian *Jumps to an external site

■ Click here for details and inquiries about this PC case Machine learning and DFT calculation compatible workstation * Please enter the name of the case or your desired conditions.