Short Description
MUSICA is a distributed high-performance computer at the locations in Vienna, Linz and Innsbruck. At each location there is a computer-cluster. Each of the tree clusters consists of 128 (Vienna: 184) individual computers (nodes). The individual computers each have 2 AMD CPUs with 96 cores each and 768 GB of main memory. 80 (Vienna: 112) nodes are aditionally equipped with 4 NVIDIA Hopper H100 GPUs per node. In total, 84,480 CPU cores and 1088 H100 GPUs are available. The nodes at a location are connected to each other and to the file systems with a high-performance network of 400 GBit/sec. In addition to the normal connection to a file system, each location has a high-performance file system with a capacity of 4 PByte each and a throughput of around 1 TByte/sec. The locations are connected to each other with a private network of 100 Gbit/sec.
The systems have a software stack that enables both batch processing and interactive work. In addition to a number of application programs, the extensive software includes the necessary tools for developing and optimizing programs. Access is available to users from all areas of research, as well as from business and public administration. An extensive course program and an email hotline are offered to support users.
Equipped with a large number of extremely powerful GPUs and high-performance file systems, MUSICA is particularly suitable for high-performance data analysis (HPDA), machine learning (ML), artificial intelligence (AI) and other data-intensive tasks in addition to classic HPC tasks.
The spatial distribution increases the resilience and availability of the system. The connection of a quantum computer is also planned at the Innsbruck site.
Contact Person
Dr. Ernst Haunschmid
Research Services
Provisioning of computing power at the highest level as well as training and support
Methods & Expertise for Research Infrastructure
MUSICA is open to users from research, industry and public administration who have a need for very high computing power. The VSC team (Org Unit VSC Research Center E057-09) operates the system and supports users. To support users, an extensive course program and an email hotline are offered.
service@vsc.ac.at
BOKU University Vienna
University of Innsbruck
TU Vienna
TU Graz
Johannes Kepler University of Linz
Further co-operation possible based on agreements