SlovakGrid Certification Authority (for academic and Grid usage)
SIVVP production cluster and Grid site (HPC)
IBM system (till 2022) = 8.8 TFlops (776 Intel CPU cores) + 19.7 TFlops (40'832 CUDA cores) + 144 TB disk storage
= 52x IBM dx360 M3 (2x Intel E5645 @ 2.4 GHz, 48GB RAM, 2x500 GB scratch disk) + 2x IBM dx360 M3 (2x Intel E5645 @ 2.4 GHz, 48GB RAM, 2x500 GB scratch disk, NVIDIA Tesla M2070 with 6GB RAM and 448 CUDA cores) + 2x DS3512 with 48+24=72 TB disks + 2x x3650M3 managing servers (2x Intel E5645 @ 2.4 GHz, 48GB RAM, 6x500 GB disks) + 4x x3650M3 data-mananaging servers (2x Intel E5645 @ 2.4 GHz, 48GB RAM, 2x500 GB disks, 2x 8Gbps FC) + 1x x3550M4 server (1x Intel E5-2640 @ 2.5 GHz, 8GB RAM, 2x500 GB disks) + InfiniBand 2x 40 Gbps (in 52+2+2+4 nodes) + console
HPC group software
OMSS - Online monitoring system and statistics for the SIVVP production cluster
Authors: J. Astalos, G. Nguyen and M. Dobrucky
Licence: Apache Software License 2.0.
Publication: M. Dobrucky, J. Astalos: UISAV: SIVVP klaster in HPC focus
Accelerated Computing in FedCloud
HP Blade system: SMART cluster
IBM system (till 2022): 8x IBM dx360 M4 (2x Intel E5-2670 @ 2.6 GHz, 64GB RAM, 2x500 GB scratch disk, 2x NVIDIA Tesla K20 with 5GB RAM and 2496 CUDA cores) + 10Gbps Ethernet (in 8 nodes)
IBM system (since 2023): 756 CPU cores, 3 NVIDIA A100 80GB, 8 NVIDIA A100 40GB, 4 NVIDIA K20, 3.9 TB RAM, 400 TB storage, InfiniBand interconnection network, VM images with preinstalled NVIDIA drivers, CUDA and Docker provided for all users.
With the advent of cloud computing, business models and user expectations are shifting towards on-demand and pay-per-use service provision increasing flexibility and agility. This approach also allows researchers and resource providers to better understand costs to access individual services and would enable the creation of innovative business models and pricing schemes.
IBM system (till 2022): 8x IBM dx360 M4 (2x Intel E5-2670 @ 2.6 GHz, 64GB RAM, 2x500 GB scratch disk, 2x NVIDIA Tesla K20 with 5GB RAM and 2496 CUDA cores) + 10Gbps Ethernet (in 8 nodes)
IBM system (since 2023): 756 CPU cores, 3 NVIDIA A100 80GB, 8 NVIDIA A100 40GB, 4 NVIDIA K20, 3.9 TB RAM, 400 TB storage, InfiniBand interconnection network, VM images with preinstalled NVIDIA drivers, CUDA and Docker provided for all users.
With the advent of cloud computing, business models and user expectations are shifting towards on-demand and pay-per-use service provision increasing flexibility and agility. This approach also allows researchers and resource providers to better understand costs to access individual services and would enable the creation of innovative business models and pricing schemes.
DAS Data Analytics Supercomputer
IKT Information and Knowledge oriented Technologies group
HP Blade system: SMART cluster
Hadoop cluster with 1x server and 14x client. Node specification: 2x Intel® Xeon® Processor E5-2620 (15M Cache, 2.00 GHz, 7.20 GT/s Intel® QPI, 6x cores, 12x threads) + HyperX threading (24 simultaneous tasks per client), 32GB RAM (2 nodes 48GB), 1TB HDD (2 nodes 500GB). Total storage capacity of the cluster: 13TB. MapReduce cluster with Hadoop and Nutch, which are used as a testing environment for developing information processing solutions.
Frameworks: Apache Spark, Apache HBase Apache Hive Apache Pig Cassandra Riak Hypertable CouchDB MongoDB Voldemort
IKT group software
Mobile Computing Lab
- mobile computing, Raspberry Pi, storage servers, ...
- quadcore cluster: 56 cores, 2 TB disk storage
- servers: web, CVS, testing nodes, ...
- testing grid site (xen)
- printers, scanners, ...
