High Performance Computing

West African Science Service Centre on Climate Change and Adapted Land Use (WASCAL). The WASCAL partner countries are Benin, Burkina Faso, Ivory Coast, Gambia, Ghana, Mali, Niger, Nigeria, Senegal and Togo

-Supply of High Performance Computing and IT Equipment for WASCAL, Ouagadougou, Burkina Faso.
-Provide powerful computing system to WASCAL CoC for enablement of further development of regional climate models
-Regional data collection based on collaboration with scientific institutions in partner countries where German scientific community contributes to the establishment of CoC
-Long-term regional climate observation network with high quality automatic weather/climate stations (AWS) and hydrological stations (AHS) should be established in other project stages
-Introduce and implement a state-of-the-art HPC architecture and infrastructure for the high-resolution regional climate simulations and computing projects, resulting in significant improvement in quality, timeliness, reliability, cost-efficiency, and profitability of the delivery of computational services for WASCAL.

Private company order for work on developing self-learning AI capable of working with full genomes of crop plants, on nucleotide level

UAB BAIP as NVIDIA Enterprise partner with AI and HPC competences, delivered NVIDIA DGX Station, world’s fastest workstation for leading-edge AI development. This will allow cusomer to have fully integrated and optimized system which enables team to get started faster and effortlessly experiment with the power of a data center in office.

1.

Client profile:WASCAL (West African Science Service Centre on Climate Change and Adapted Land Use) is a large-scale research-focused Climate Service Centre designed to help tackle this challenge and thereby enhance the resilience of human and environmental systems to climate change and increased variability. It does so by strengthening the research infrastructure and capacity in West Africa related to climate change and by pooling the expertise of ten West African countries and Germany.

Client need:Data for biophysical parameters and socio- economic indicators is collected in meteorological, hydrological, land use, biological/biodiversity, and socio-economic observation networks. To ensure proper data reception, data management and access, analysis and interpretation of data and of the results of scenario simulation a high quality technical infrastructure was needed to implement. A powerful computing system was needed to enable the further development of regional climate models, and using them for numerous simulation experiments and the associated pre- and post-processing of input and result data.

Provided solution: HPC infrastructure architecture’s preparation, delivery, implementation, testing and maintenance services was provided in this project. The implemented HPC solution is based on worldwide known manufacturer for HPC platforms – Dell EMC. The HPC platform provides the computing power of 320 CPU cores and it’s design is fully scalable up to 8EB (exabytes) per file system, supporting up to 512PB (petabyte) of file size. More than 115 hardware items (racks, servers, switches, etc.) were integrated together to start up the whole system. Additionally, one week on-site training workshop was conducted for WASCAL staff and data managers to ensure the proper use of HPC cluster.

2.

Client profile:Vilnius University, the oldest and largest Lithuanian higher education institution with over 430 years of academic traditions and the widest range of study programmes in Lithuania.

Client need: the University had a need to create an Information Technology Open Access Center to provide access to researchers and businesses that need high performance computing in biochemistry, biomedical, optical spectroscopy and other areas. The aim was to open up completely new possibilities for high-performance computing (HPC) in various fields of science and economy, to accelerate the transfer of technological knowledge from the academic level to business, to allow more efficient research and to cooperate more actively with foreign scientists.

Provided solution: the supercomputer’s implementation was phased in at two geographically remote locations (The Digital Research and Computing Center in the Faculty of Mathematics and Computer Science; faculty of Phyciss) that together provide 3200 computing cores and 800 TB of disk space. The cluster is merged and managed with high-performance parallel file system Lustre ™. The symmetric multiprocessor computing system implemented in the high-performance system was the only one of its kind in Lithuania, which is highly efficient and ensures extremely high scalability.

More: https://www.baip.lt/projektai/didelio-nasumo-skaiciavimo-sistemu-superkompiuteriu-projektavimas/

3.

Client profile:Biotechnology company, working on developing self-learning Artificial Intelligence capable of working with full genomes of crop plants on nucleotide level.

Client need: the company analyzes large volumes of sequenced genomes and there was a need for a the new tool thah could works with the „raw“ genome directly at the nucleotide level and can inexpensively extract significant areas hundreds of times faster than current methods.

Provided solution: to achieve the goal the NVIDIA DGX Station was delivered. NVIDIA DGX Station, which is built on NVIDIA Volta GPU architecture, is world’s fastest workstation for leading-edge Artificial Intelligence (AI) development. This solution empowered the customer to get started faster and effortlessly experiment with the power of a data center in the office. Delivered NVIDIA DGX Station calculations are up to 47 times faster if compared to a server with 2 CPUs and 20 times faster if compared to 4X GPU workstation. Because NVIDIA DGX Station is fully assembled in NVIDIA factory, the deployment on client ‘s premises took only several hours.

4.

Client profile: Bangladesh Computer Council, statutory and autonomous government body that aids in the use of information technology and the formulation of related policy for Bangladesh, which has a population of more than 160 million people.

Client need: to integrate all the services provided by Bangladesh national data center to single self-service portal, which will be used by all public sector’s organizations and hundreds of millions of residents.

Provided solution: service catalogue was created to offer client packages on 4 different levels: IaaS, PaaS, SaaS and BPaaS (Business Process-as-a-Service). IT service management model was developed as well as self-service portal with payment and billing capabilities. Training of IT administrators, engineers and managers was organized.

Other – High performance computing (HPC) architecture’s design, preparation, delivery, implementation, testing, training and maintenance. Depending on client’s needs, HPC architecture can be based on CPU, GPU or hybrid technologies. HPC serves as a state-of-the-art architecture platform for the needs of Artificial Intelligence (AI), Deep Learning (DL), Machine Learning (ML) and BigData frameworks and products. HPC is a key element for efficient data management, analysis, simulation and interpretation of data, which arise from the needs of various business’ industries, governmental and science institutions.

Commercial infrastructure – design and implementation of HPC resources’ management platform, helping to publish, manage and account various services and resources of HPC, especially when HPC resources are shared or provided as services. The properly designed and implemented HPC resources’ management platform covers all aspects of HPC’s resource brokerage, SLA monitoring, billing (based on costs resources usage calculations – CPU/GPU, network components, storage and other components), policies and access control. The solution can provide fully automated services invocation through self-service portal for HPC end-customers.

Other – Cloud-based service catalogue creation to offer client packages on 4 different levels: IaaS, PaaS, SaaS and BPaaS (Business Process-as-a-Service). IT service management model could be developed as well as self-service portal with payment and billing capabilities.

TBA

Deep learning and high performance calculations – before purchasing NVIDIA DGX Station, deep learning and high-performance computing applications and frameworks can be tried and tested on any major GPU Cloud providers, including AWS, MS Azure, Google Cloud, IBM Cloud, Oracle Cloud.