Success Story

Customer

Our client is the research and development department of a large German pharmaceutical company. The identification of potential drug targets often involves the use of bioinformatics tools and databases that have to process enormous amounts of genetic, molecular and clinical data. High-performance computers are crucial for carrying out these processes in a manageable amount of time.

In preclinical drug development, computational design tools are often used to model and optimize potential drug structures. These tools require significant computational resources, especially when dealing with complex molecules or large amounts of data. Similarly, data analysis in clinical trials often requires dealing with huge amounts of data derived from clinical trials. Statistical analysis and machine learning used to interpret this data can also be enormously computationally intensive.

Project Description

Schrödinger life science research and materials science software applications provide solutions for computational drug design, materials science, bioinformatics and more. They use advanced algorithms and machine learning to model and analyze molecular structures, contributing to the discovery and development of new drugs and materials.

GPUs play a special role in the use of Schrödinger software applications (or similar computer-based research tools), since particularly large amounts of data must be processed in parallel in molecular modeling and simulation.

Therefore, the goal of the project was to update the current hybrid HPC/GPU clusters with current Intel CPU and NVIDIA A100-80/PCIe technology. .

Project Realization

With the goal of optimizing research and development results using Schrödinger software applications, we delivered the latest Supermicro Multi GPU SuperServer 220GP-TNR to our customers. This state-of-the-art server is equipped with four NVIDIA A100-80/PCIe GPUs as well as fast U.2 NVMe storage to achieve peak performance. The system also has expansion reserves for two additional dual NVIDIA A100-80s.

In the first phase, we implemented ten such systems. Based on the specific requirements and performance of the state-of-the-art chemical simulation software used in pharmaceuticals, biotechnology and materials science, we plan to gradually increase the number of computing units and further optimize the system.

Again, the Memorysolution Custom Server Solutions team delivered the hybrid CPU/GPU cluster as a turnkey solution and was able to seamlessly integrate it into the customer's existing server landscape.

The Result

The implementation of the Supermicro Multi GPU SuperServer 220GP-TNR has significantly increased computational speeds. This enables researchers to use their bioinformatics tools and databases more efficiently and to identify potential drug targets faster.

In the preclinical phase, the improved computing power helps to accelerate the modeling and optimization of potential drug structures. In clinical trials, the increased computing power makes it easier to handle large amounts of data, making statistical analysis and machine learning more effective for interpreting the data.

With the help of our solution, the customer has been able to accelerate and optimize its research and development processes, which ultimately contributes to faster development of new drugs and treatments.

The project also illustrates the high scalability of the Multi GPU SuperServer platform. Having implemented ten systems in the initial phase, we look forward to further increasing the number of compute units as the customer's performance requirements grow.