Call for Pro­pos­al - Large Pro­jects 2023-4

Scientists qualified in their respective field of research can submit proposals for access to HPC resources operated by Paderborn Center for Parallel Computing (PC²). We offer multi-million core-hour projects on a recent generation of HPC systems with low-latency interconnect. PC² is one of nine compute centers in Germany of the NHR alliance (Nationales Hochleistungsrechnen). The main focus of Paderborn is on three important application competence areas and their user communities: Atomistic Simulations, Optoelectronics and Quantum Photonics, and Machine Learning for Intelligent Systems. PC² operates its first NHR large scale HPC system with general availability since April 2022.

PC² provides a portfolio of preinstalled applications and offers support for software installation/configuration of project-specific applications. We provide consulting services for all users and we advise them on using the resources in an efficient way.  In particular, we provide consulting services in the field of Computational Chemistry and Solid-state Physics. Furthermore, PC² encourages approximate computing and the acceleration of applications with Field-Programmable Gate Arrays (FPGA). Proposals from young scientists and institutions where equal opportunities are concerned are particularly encouraged.

Available Resources

Noctua 2
(NHR HPC system and
2nd installation phase of Noctua)

1.061 compute nodes, each 128 processor cores (2x AMD 7763) and 256 GiB main memory (66 nodes with 1 TiB, 5 nodes with 2 TiB and 32 TB fast NVME-SSDs)
32 GPU nodes, each 2x AMD 7763, 512 GiB main memory, and 4x NVIDIA A100 40GB
36 FPGA nodes, each 2x AMD 7713, 512 GiB main memory, with total 80 FPGA-cards (Xilinx Alveo U280 and Bittware 520N)

100/200 Gbps InifiniBand network
6 PByte parallel file system
https://pc2.uni-paderborn.de/hpc-services/available-systems/noctua2/

Noctua 1
(HPC system of the 1st installation phase of Noctua)

256 compute nodes, each 40 processor cores (2x Intel Xeon 6148) and 192 GiB main memory
18 GPU nodes, same as a compute node with additional RTX 2080 Ti (11 GB)
100 Gbps Omni-Path network
720 TByte parallel file system
https://pc2.uni-paderborn.de/hpc-services/available-systems/noctua1/

Proposals can be submitted for large projects with more than 15 million core hours.

Apart from large projects also

  • normal projects (more than 4 million core hours and <= 15 million core hours, resp. less than 0.1 million GPU hours)
  • small projects (4 million or less core hours, resp. 0.01 million or less GPU hours) and
  • test projects (code testing and optimization, limited to a 30 day period)

can be submitted at any time.

For details see https://pc2.uni-paderborn.de/hpc-services/our-services/system-access-application.

Who can apply?

Eligible to submit proposals are researchers with a Ph.D. degree or a comparable degree who are members of a public or state-recognized, institutionally accredited German university.

Small projects can only be requested by scientists from universities located in North Rhine-Westphalia (NRW). Exceptions can be made for small proposals with a special emphasis on the development of FPGA acceleration for HPC codes or projects that directly collaborate with PC².

Scientists from universities with local Tier 3-compute center HPC resources should only apply for normal or large projects.

All proposals should show the potential of achieving results that are intended to be published in scientific journals or presented at conferences.

How to apply?

Please follow out guides for the project categories: test, small, normal or large projects.

Please note that new simplifications for normal and large proposals are possible:

  • Well-known codes: If you use a program that is on our list of well-known codes, then you don't have to include a scaling analysis in the detailed description.
  • Project was already scientifically reviewed: If your reserach project has already been reviewed scientifically by the BMBF, DFG, European Union (EU projects), a GCS center or another NHR center then the detailed description is drastically shorter. For details see simplifications.

The default project duration of large, normal, and small projects is 12 months. Multi-year normal and large projects are possible.

Important Dates for Large Projects

Large Projects > 15 Mio. CPU-core-hours or > 100.000 GPU-hours
Closing date of Call for Proposal 2023-4

extended till August, 7th 2023 (12:00 MET)

Scientific reviews of proposals July/August 2023
Announcement of allocation decision September 2023
Allocation period for awarded large proposals October 2023 until September 2024
Normal Projects <= 15 Mio. CPU-core-hours and <= 100.000 GPU-hours
Closing date of Call for Proposal submission and project start at any time
Allocation period for awarded normal proposals 12 months or multi-year projects

There are four calls for proposals for large projects per year. The follow-up "Call for Proposals 2024-1" will have an allocation period starting on 1st of January 2024.

Formal Regulations 

The normal and large proposals will be ranked based on scientific and technical quality in a peer review process. Small projects are reviewed in a continuous internal process.
Proposals may be granted in their entirety, with a changed volume, or rejected by the Resource Allocation Board of PC².  

Contact

Paderborn University - PC²
PC² - Resource Allocation Board
Warburger Str. 100
33098 Paderborn

pc2-support(at)uni-paderborn(dot)de

Head of the Resource Allocation Board: Prof. Dr. Thomas Kühne