Grant: Research Tools and Instruments
University of Toronto
$150,000.00
Developing accurate radiative transport methods with GPU-accelerated computing
FY 2023-24
Summary
In the last few years, the field of astrophysics took a huge leap forward due to the development of GPU-accelerated codes. In particular, the H-AMR code that our team at CITA develops and uses is 1000 times faster than conventional codes running on CPUs. We request funding for five NVIDIA A100 GPU servers to further develop our open-source GPU-accelerated codes including novel radiative magnetohydrodynamics methods, which are essential to model astrophysical emission. The codes will shed light on the emission of multi wavelength and multimessenger radiation from neutron star mergers, as well as accretion disks of black holes and protoplanetary systems. Our team at the University of Toronto brings together world-renowned experts in code development, advise highly qualified personnel (HQP), studying and comparing numerical and observational astrophysical data. Our open-source software will serve a large community of plasma-astrophysicists as it will be the first GPU-accelerated code with an accurate radiation treatment. The development of such code is infeasible on large-scale clusters aimed at ready-to-run simulations, and for which an annual competitive application process is required. Instead, code development requires quick throughput and instant access to a small number of GPUs. In addition, a small local GPU cluster is ideally suited for the analysis of generated data and its comparison to observations. Lastly, the developed methodology and codes will result in competitive applications for (inter)national grants on exascale clusters. The requested equipment will also be a pivotal resource for HQP training in computational physics, code development and efficient data analysis. These skills are indispensable for a career in either academia or industry (e.g., banking, engineering, chip development). We foresee that our open-source code, in addition to a local and accessible GPU cluster, will foster an inclusive and diverse environment for HQP. The impact of the developed methodology will result in competitive applications for (inter)national compute time on exascale clusters. The GPUs will facilitate new research directions, collaborations, and granting opportunities, while providing our HQP with hands-on coding experience, such that both our students and research programs will remain internationally competitive. The research that can be carried out with the requested GPU nodes and its resulting developed open-source code will have a significant impact on the modeling and understanding of recent groundbreaking observations by the Event Horizon Telescope (EHT), the Laser Interferometer Gravitational-Wave Observatory (LIGO), the Canadian Hydrogen Intensity Mapping Experiment (CHIME), and the IceCube Neutrino Observatory by our group and by the larger astrophysics community. Now is the time to invest in a local GPU cluster and the development of the next-generation GPU-accelerated codes essential to solve longstanding questions in astrophysics.
Program
Research Tools and Instruments
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
RTI-2024-00306
Name nom
Ripperda, Bart
Department département
Canadian Institute for Theoretical Astrophysics
Organization id
31
Institution établissement
University of Toronto
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2023
Competition year année de concours
2024
Award amount
150000
Program id
RTI
Program name en
Research Tools and Instruments
Program name fr
Outils et instruments de recherche
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1605
Committee name en
Physics (RTI)
Committee name fr
Physique (OIR)
Area of application code
801
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Computer software
Area of application fr
Logiciel
Research subject code
3107
Research subject group en
Physics
Research subject group fr
Plasma physics
Research subject en
Physique
Research subject fr
Physique des plasmas
Application title
Developing accurate radiative transport methods with GPU-accelerated computing
Keyword
accretion disks; black holes; computational physics; GPU acceleration; magnetohydrodynamics; numerical methods; plasma physics; radiation transport
Application summary
In the last few years, the field of astrophysics took a huge leap forward due to the development of GPU-accelerated codes. In particular, the H-AMR code that our team at CITA develops and uses is 1000 times faster than conventional codes running on CPUs. We request funding for five NVIDIA A100 GPU servers to further develop our open-source GPU-accelerated codes including novel radiative magnetohydrodynamics methods, which are essential to model astrophysical emission. The codes will shed light on the emission of multi wavelength and multimessenger radiation from neutron star mergers, as well as accretion disks of black holes and protoplanetary systems. Our team at the University of Toronto brings together world-renowned experts in code development, advise highly qualified personnel (HQP), studying and comparing numerical and observational astrophysical data. Our open-source software will serve a large community of plasma-astrophysicists as it will be the first GPU-accelerated code with an accurate radiation treatment. The development of such code is infeasible on large-scale clusters aimed at ready-to-run simulations, and for which an annual competitive application process is required. Instead, code development requires quick throughput and instant access to a small number of GPUs. In addition, a small local GPU cluster is ideally suited for the analysis of generated data and its comparison to observations. Lastly, the developed methodology and codes will result in competitive applications for (inter)national grants on exascale clusters. The requested equipment will also be a pivotal resource for HQP training in computational physics, code development and efficient data analysis. These skills are indispensable for a career in either academia or industry (e.g., banking, engineering, chip development). We foresee that our open-source code, in addition to a local and accessible GPU cluster, will foster an inclusive and diverse environment for HQP. The impact of the developed methodology will result in competitive applications for (inter)national compute time on exascale clusters. The GPUs will facilitate new research directions, collaborations, and granting opportunities, while providing our HQP with hands-on coding experience, such that both our students and research programs will remain internationally competitive. The research that can be carried out with the requested GPU nodes and its resulting developed open-source code will have a significant impact on the modeling and understanding of recent groundbreaking observations by the Event Horizon Telescope (EHT), the Laser Interferometer Gravitational-Wave Observatory (LIGO), the Canadian Hydrogen Intensity Mapping Experiment (CHIME), and the IceCube Neutrino Observatory by our group and by the larger astrophysics community. Now is the time to invest in a local GPU cluster and the development of the next-generation GPU-accelerated codes essential to solve longstanding questions in astrophysics.