Required cookies

This website uses cookies necessary for its operation in order to provide the user with content and certain functionalities (e.g. language selection). You have no control over the use of these cookies.

Website visitor statistics

We collect visitor statistics on the use of the site. The data is not personally identifiable and is only stored in the Matomo visitor analytics tool managed by CSC.

By accepting visitor statistics, you allow Matomo to use various technologies, such as analytics cookies and web beacons, to collect statistics about your use of the site.

Change your cookie choices and read more about visitor statistics and cookies

CSC

There is a growing interest among researchers in GPU computing. Supercomputers use both the traditional x86 architecture (CPU, Central Processing Unit) and the GPU (graphics processing unit), originally known as graphics processors. GPUs are very suitable for large-scale parallel tasks where the GPU computing power exceeds the CPU.

GPU-computing is used, for example, for artificial intelligence research, training in the AI models, and analysis of large datasets (Big Data). In research, artificial intelligence and machine learning have been used e.g. to reliably predict which drug combinations kill cancer cells, to identify prostate biopsies containing cancer nearly without error, and to create an intelligent annotation pipeline for semi-automated annotation (adding metadata) and enrichment of archived material.

– This extension of the DL2021 environment with 120 A100 based GPUs increases significantly its capability to support AI workloads and converged AI and HPC workloads. By investing both in Mahti-AI and ePouta we can support science done on sensitive data such as genomics. It also exposes scientists in Finland to the latest generation NVIDIA GPU hardware and its new capabilities such as improved tensor cores, enabling application developers to take these in use, said Development manager Sebastian von Alfthan at CSC.

The GPU nodes are based on the Atos BullSequana X2415 GPU blade with

  • two AMD Epyc Rome 7H12 processor (CPU)
  • four next-generation NVIDIA Ampere A100 processors (GPU)
  • each node has 512 GB memory for CPU and 160 GB for GPU (40 GB to each GPU).

In addition to processors, a supercomputer also needs a high-speed interconnect network between nodes. These GPU nodes will have two 200 Gbit/s connections.

The six nodes in ePouta will consist of

  • two AMD Epyc Rome 7402 -processors (CPU) and
  • four NVIDIA A100 -processors (GPU)
  • each node will have 1024 GB of memory for CPU and 160 GB memory for GPU (40 GB to each GPU).

This purchase raises Mahti’s theoretical peak performance by two Petaflop/s. After installation, Mahti’s theoretical peak performance will be 9.5 Petaflop/s. The performance of ePouta will increase by 0.5 Petaflop/s.

The new GPU nodes will be opened into researchers’ use during April, according to the present timetable.

More information

Sebastian von Alfthan, Development manager, CSC
sebastian.von.alfthan at csc.fi, +358405888688

Aalto University: AI predicts which drug combinations kill cancer cells

CSC’s computing environment