Computer Science & AI
In support of the research and education, the Leiden Institute of Advanced Computer Science (LIACS) has a Research and Education laboratory at its disposal. Within this environment, we can offer machines that go beyond normal office automisation and production.
The Research and Education (R&E) Laboratory is a special computer network with a number of general facilities and some smaller labs. The labs house the devices for the different research groups. Of course all groups also use each other's machines.
The general facilities are mainly meant for access to the separate computer network and in support of the labs. Facilities that are available:
- SSH to get access to the R&E Lab.
- VPN to get access to the network services of the university like the digital library, e-mail and other internal services.
- Wired and wireless network access within the Snellius building.
- Hosting of research servers, in order to allow researchers to install and run their own servers for their research and experiments.
- Bulk data storage, to store large scale datasets.
- Virtual Machine cluster, in order to offer simple services, like the webserver of a research group, or as a temporary server to be able to experiment on a short notice.
Data Science Lab
The Data Science Lab consists of a number of modern computers with unusual configurations that enable us to perform complex calculations and do large scale data analyses. The computers are accessible to all staff and students of LIACS, and to members of the Leiden Center of Data Science (LCDS).
Special facilities that are present here:
- Machines with a large amount of internal RAM memory (1TB to 1.5TB). For example to perform computations on very large graphs and/or social networks.
- Machines with a large number of cores (64 cores, 128 threads), for intensive computation in parallel or research on multi-core calculations on many cores.
- Machines with graphical cards (GPUs) of the latest generation, for research on the increased velocity of calculations with the aid of GPUs (e.g., a machine with 12.000 GPU cores over 4 GPU's).
- achines with multiple terabytes of fast SSD storage, to fasten database manipulation.
- Shared secure data storage (over 100TB).
- Possibility of large scale data storage for research support.
High Performance Computing Lab
Within the High Performance Computing Lab, several small cluster computers have been placed, that are being used in cluster computing, the research on the division of calculations among several computers.
LIACS is a participant of the Distributed ASCI Super computer (DAS), a cluster computer spread among five Dutch universities that are members of the ASCI graduate school.The cluster computers at the various locations are attached to one another, to enable experiments with ‘grid’ systems, in which an application runs on several cluster computers at a time.
At this moment, the following DAS systems are active:
- DAS-3 is, at this moment, used by LIACS for education in the area of Distributed Data Mining with the use of Hadoop. The machine was originally installed in 2007 and consists of 32 dual-CPU machines.
- DAS-4 consists at LIACS of 16 dual quad-core machines, each with 48 GB RAM memory and 10 TB local storage. We are using DAS-4 for several research projects.
- DAS-5 is the most modern machine, installed in 2015 and at LIACS consisting of 24 dual 8-core machines, each with 64 GB memory. There is a file server with a capacity of 128 terabytes available.
Furthermore, LIACS has the ‘Little Green Machine’ (LGM) at its disposal. LGM is a cluster computer equipped with GPUs. This machine was brought into use in 2011 and was one of the largest GPU clusters in the Netherlands. Until now, the machine is used for research on distributed GPU calculations.
Within the LIACS Media Lab, research on multimedia analyses, artificial intelligence and internet technologies is performed. The media lab has a separate room at its disposal with special provisions for adjusting the light. This makes it possible to create both a totally dark and an extra clear environment, which is necessary for performing controlled scientific experiments.
Furthermore, the room is equipped with heavy power supply to facilitate special multimedia devices.
The media lab is also equipped with computer and video cards of the latest generation, providing the possibility of pioneering research in the area of deep learning.
The Imaging Lab is mainly used by the Bioimaging group. Within the lab, a cluster computer has been placed: the LIACS Life Science Cluster (LLSC), available for running Bioinformatics software. We are performing research on how we can simplify the use of such software and significantly fasten the calculations with the aid of the cluster, in order to apply this technology in support of the physical laboratory work at Biology and NeCen.
The pride of the Imaging Lab is the video wall, which exists of twelve 42 inch Full HD screens. The screens are directed by a single computer that has been equipped with three powerful video cards. This screen can accomplish a very high resolution, which makes it possible to, for example, show microscopic images very sharply. Using the video wall, research is performed on special interactive arrangements, in which researchers can view and manipulate their microscopic images with the latest techniques of interaction.
Natural Computing Group
The Natural Computing Group focusses on solving problems using ideas from nature, such as evolutionary algorithms and neural networks. Within this group, somne dedicated computers are available for certain projects that are frequently running long experiments. These can include the training of machine learning models, performing wind tunnel simulations and letting evolutionary algorithms solve design problems.
For these experiments, we have several machines with 32+ cores, plenty of internal RAM (up to 512GB) and access to several modern GPUs (1080Ti's and 2080Ti's). Every machine also has NVME SSD's for fast local storage and is connected to an archive machine that can hold over 100TB of data.