Skip to main content Skip to search Skip to main navigation

High-Performance Computing Infrastructure to Support Molecular Research

Customer

Our customer is a globally operating research and development company specializing in drug discovery. In modern molecular research, powerful computing resources are essential for running highly complex simulations and developing new pharmaceuticals. Particularly compute-intensive tasks such as virtual screening, protein structure analysis, and drug candidate optimization require specialized high-performance computing (HPC) solutions.

With increasing demands for data processing and modeling accuracy, the customer sought a scalable, GPU-accelerated solution optimized for cutting-edge AI and simulation applications.

Project Date: 2024
Project Volume: approx. 1,200,000 €

Project Description

The goal of the project was to implement a high-performance computing infrastructure of the 4th expansion stage to support molecular research with maximum computational power. The main objectives included:

  • Accelerating virtual screening processes to identify potential drug candidates
  • Enhancing protein structure analysis using AI-powered simulation models
  • Optimizing high-performance data processing for analyzing and modeling complex molecular interactions

To meet these requirements, we developed a customizable and scalable HPC solution based on the latest GPU technology.

Project Implementation

Memorysolution delivered and integrated nine Supermicro GPU SuperServer 420GP-TNAR systems, featuring:

  • 2x 32-Core Intel Xeon Scalable Processors (3rd Gen, Gold 6338)
  • 512 GB DDR4 RAM for high-speed data processing
  • 12 TB NVMe storage to optimize data throughput
  • 8x NVIDIA H100 SXM GPUs with NVSwitch, providing up to 900 GB/s of internal bandwidth for ultra-fast GPU communication

This system configuration enables highly complex computations to run in parallel while efficiently processing large datasets. The NVSwitch technology allows direct GPU-to-GPU communication, minimizing bottlenecks and maximizing overall system performance.

Memorysolution Custom Server Solutions delivered and integrated the new turnkey HPC cluster seamlessly into the customer's existing infrastructure. The solution was designed to be scalable and future-proof, ensuring that it can be expanded to meet growing computational demands.

The Result

With the new GPU-accelerated HPC platform, our customer achieved significant improvements in computational speed and reduced the simulation time required for molecular models.

  • Faster identification of potential drug candidates through optimized virtual screening processes
  • Increased efficiency in protein structure analysis, enabling more accurate predictions and accelerating drug development
  • Maximum scalability to meet future demands, thanks to the modular system architecture

By implementing this powerful and scalable HPC infrastructure, the customer is now well-equipped to support future advancements in drug discovery. The system is designed to grow alongside the increasing demands of molecular research while efficiently handling next-generation AI-powered simulation models.

This project highlights the exceptional scalability and performance of our Supermicro GPU solutions and underscores our expertise in delivering high-performance computing solutions for research and development.

Do You Still Have Questions?

Our support staff will be happy to assist you with all enquiries relating to information on availability, technical product information or advice on product selection. You can reach us by phone, e-mail or via the contact form. You can always find our contact details in the footer.