Next-generation scientific instruments and simulations will generate massive amounts of imaging and scientific data, such as data collected from synchrotron radiation facilities, cosmology, and weather simulations. The analysis of these scientific images and datasets requires novel high-performance computing techniques, AI/ML methods, and runtime systems that can provide timely and/or real-time solutions.
The current large-scale computing systems and supercomputers comprise complex, heterogeneous components with varying performance characteristics across different layers and tasks. These include high-performance interconnects such as NVLinks, PCIe, and Ethernet/InfiniBand, as well as specialized compute components such as Tensor/CUDA cores and general-purpose CPUs distributed among multiple sockets (and many nodes). Efficiently using these complex hardware resources is nontrivial.
Within this context, we are interested in articles that address the challenges relevant to "AI/ML-enhanced high-performance computing/communication techniques and runtime systems for scientific image and dataset analysis" that can efficiently use heterogeneous computing platforms. The topics of interest include, but are not limited to, the following:
• New high-performance computing techniques and runtime systems for accelerating scientific data/image analysis and processing.
• Novel AI/ML models that can be incorporated into or replace analysis tasks to accelerate end-to-end execution.
• Techniques for efficient communication between tasks on large-scale systems, e.g., new hierarchical communication and reduction techniques for scientific applications.
• Novel heterogeneous computing techniques to improve the data analysis pipeline: mixed-precision computing, using specialized computing capabilities on accelerators, such as Tensor/CUDA cores.
• Novel applications of AI/ML techniques in diverse scientific domains, such as bioinformatics, materials science, and drug discovery, to accelerate data analysis and enhance scientific insights.
Keywords:
AI/ML, High performance computing, scientific image, runtime systems, data analysis
Important Note:
All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.
Next-generation scientific instruments and simulations will generate massive amounts of imaging and scientific data, such as data collected from synchrotron radiation facilities, cosmology, and weather simulations. The analysis of these scientific images and datasets requires novel high-performance computing techniques, AI/ML methods, and runtime systems that can provide timely and/or real-time solutions.
The current large-scale computing systems and supercomputers comprise complex, heterogeneous components with varying performance characteristics across different layers and tasks. These include high-performance interconnects such as NVLinks, PCIe, and Ethernet/InfiniBand, as well as specialized compute components such as Tensor/CUDA cores and general-purpose CPUs distributed among multiple sockets (and many nodes). Efficiently using these complex hardware resources is nontrivial.
Within this context, we are interested in articles that address the challenges relevant to "AI/ML-enhanced high-performance computing/communication techniques and runtime systems for scientific image and dataset analysis" that can efficiently use heterogeneous computing platforms. The topics of interest include, but are not limited to, the following:
• New high-performance computing techniques and runtime systems for accelerating scientific data/image analysis and processing.
• Novel AI/ML models that can be incorporated into or replace analysis tasks to accelerate end-to-end execution.
• Techniques for efficient communication between tasks on large-scale systems, e.g., new hierarchical communication and reduction techniques for scientific applications.
• Novel heterogeneous computing techniques to improve the data analysis pipeline: mixed-precision computing, using specialized computing capabilities on accelerators, such as Tensor/CUDA cores.
• Novel applications of AI/ML techniques in diverse scientific domains, such as bioinformatics, materials science, and drug discovery, to accelerate data analysis and enhance scientific insights.
Keywords:
AI/ML, High performance computing, scientific image, runtime systems, data analysis
Important Note:
All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.