AI Supercomputing Research FAQ

Subject: 
AI Supercomputing Research FAQ

Environment:
Fresno State University, OpenAI, AI Supercomputing Research 

Article Summary:
AI Supercomputing National Research Platform FAQs.

Resolution: 

1. What is the National Research Platform (NRP)?

The National Research Platform (NRP) is a collaborative, cutting-edge infrastructure designed to support advanced research and innovation across various scientific disciplines. It brings together a global network of distributed computing resources, including high-performance GPUs, CPUs, FPGAs, and specialized hardware, enabling researchers to conduct experiments and simulations that would be impossible or impractical on traditional systems. The platform is accessible through user-friendly interfaces such as JupyterHub and Coder, providing tools for data analysis, machine learning, and large-scale computing. With its wide-reaching capabilities, the NRP supports research, education, and collaboration among academic institutions, industry, and government organizations worldwide.

2. How do I get support for technical issues on the NRP?

Different ways to reach for technical issues:

  • Call Fresno State Service Desk at 559.278.5000 to open a support ticket
  • Reach out to Fresno State Research Support team at researchsupport@mail.fresnostate.edu
  • Open a TDX ticket using the link: <Pending creation>

3. Who can access the National Research Platform?

The National Research Platform (NRP) is accessible to a wide range of users, including:

  • Academic Researchers, faculty and students – Professors, students, and research teams from universities and colleges who require high-performance computing resources for their projects can gain access.
  • Industry Professionals – Companies and organizations working on cutting-edge research and development in fields such as AI, machine learning, data science, and high-performance computing can utilize the platform for their experiments.
  • Government Agencies – Researchers and teams from government institutions and departments can access the platform for scientific research, simulations, and data analysis.
  • Collaborative Research Networks – NRP also facilitates collaboration between different institutions, making it accessible to research consortia working on global projects.

Access may be granted through an application process or through specific partnerships and grants, depending on the type of user and the intended use of the platform.

4. What resources are available on the National Research Platform?

The National Research Platform (NRP) offers a wide range of resources to support advanced research and experimentation, including:

  • High-Performance GPUs – Powerful graphics processing units (GPUs) designed to accelerate machine learning, AI, simulations, and data-intensive tasks.
  • CPUs – Central processing units (CPUs) for general-purpose computing and a wide variety of scientific and engineering applications.
  • FPGAs – Field-programmable gate arrays (FPGAs) for specialized computing tasks, providing flexibility and high performance for specific research needs.
  • Specialized Storage Solutions – Scalable and high-performance storage to handle large datasets, ensuring quick access and secure management of research data.
  • Cloud-Based Computing Resources – Distributed computational power accessible through flexible cloud-based solutions, enabling global collaboration and resource sharing.
  • Networking and Data Transfer Infrastructure – High-speed networking tools and systems that enable seamless data transfer and support large-scale experiments across distributed nodes.
  • User-Friendly Interfaces – Tools like JupyterHub and Coder to simplify access to computing resources, making it easier for researchers to work with the platform and manage their projects.
  • Collaboration Tools – Features that facilitate collaboration among researchers and institutions globally, including shared workspaces and networking capabilities.
  • AI and Machine Learning Frameworks – Pre-installed software and tools for AI/ML research, making it easier for researchers to run complex models and algorithms.
  • Security and Data Management – Robust systems in place to ensure the safety and privacy of research data, along with tools for efficient data management and governance.

These resources collectively support a wide range of scientific disciplines and provide the computational power needed for large-scale, cutting-edge research.

5. How can researchers access the National Research Platform (NRP)?

Please reach out to the Fresno State Research Support team to get started with your request. Our team will help you to get access to NRP.

Below are the steps involved to access the National Research Platform (NRP):

  • Eligibility and Application

If you are part of a university or research institution, you can apply for access by submitting a request through the NRP’s application portal. You may need to provide details about your research project and the resources you require.

Companies and industry researchers may also gain access by applying for use cases that align with NRP's objectives. Certain industry collaborations or partnerships may provide access.

Government entities may access the platform through specific programs or collaborations.

  • Approval and Resource Allocation
    After submitting your application, the NRP team will review your project and allocate resources based on availability and relevance to the platform’s mission. You may be required to outline how you plan to use the resources (e.g., GPUs, storage, etc.) in your research.
  • Creating an Account
    Once your access is approved, you will need to create an account on the NRP platform. This will allow you to log in and access available computing resources, data storage, and tools like JupyterHub or Coder.
  • Using the Platform
    After logging in, you can start utilizing the platform's resources for your research. The platform provides an intuitive interface for setting up experiments, running simulations, and managing data. Additionally, you'll have access to documentation and support resources to help you navigate the platform.
  • Collaboration and Support
    The NRP encourages collaboration among researchers. If your project involves collaboration with other institutions or professionals, you can use the platform’s networking tools to facilitate shared experiments and data exchange. You can also reach out to the NRP support team for technical assistance and guidance.
  • Compliance and Terms of Use
    Be sure to review and comply with any terms of use, data privacy regulations, and ethical guidelines set by the NRP to ensure responsible usage of the platform's resources.

If you are unsure about the process, please reach out to the Fresno State Research Support team or Service Desk for additional guidance.

6. What types of research are supported on the National Research Platform (NRP)?

The National Research Platform (NRP) supports a wide range of research types, particularly those that require high-performance computing, advanced simulations, and large-scale data processing. Some examples of research areas include:

  • Artificial Intelligence (AI) and Machine Learning (ML) 
    AI/ML model development, training, and testing, including deep learning, reinforcement learning, natural language processing (NLP), and computer vision applications.
  • Data Science and Big Data Analytics
    Processing and analyzing large datasets across various domains, including genomics, social sciences, environmental science, and finance.
  • High-Performance Computing (HPC)
    Complex simulations and computations in physics, engineering, and climate modeling that require massive computational power and parallel processing.
  • Network Research and Simulations
    Testing and evaluating network protocols, distributed systems, and cloud infrastructure. The platform supports experiments in network performance, security, and communication models.
  • AI for Healthcare and Bioinformatics
    Machine learning and AI applications for healthcare research, including medical imaging, drug discovery, and genomics, alongside bioinformatics data analysis.
  • Robotics and Autonomous Systems
    Development of algorithms and simulations for autonomous vehicles, drones, and robotics, including the use of FPGAs for specialized processing needs.
  • Cybersecurity Research
    Simulation and testing of security protocols, vulnerability analysis, threat detection, and the development of advanced security systems.
  • Scientific Simulations and Modeling
    Computational chemistry, molecular dynamics, and physics-based simulations, where large amounts of computational power are required to run complex models.
  • Climate Modeling and Environmental Science
    Research on climate change, weather prediction, and environmental monitoring that requires powerful computational resources for simulations and large-scale data processing.
  • Computational Neuroscience and Cognitive Modeling
    Research into brain simulations, neural networks, and cognitive processes, leveraging AI and computational techniques for neuroscience.
  • Quantum Computing
    Though still in its early stages, the platform can facilitate research on quantum algorithms and quantum system simulations as they develop.
  • Collaborative Research Across Disciplines
    Multidisciplinary research that spans across multiple scientific areas, where the integration of data, experiments, and simulations is needed to advance knowledge in fields like materials science, engineering, and social sciences.
  • Distributed Systems and Cloud Computing
    Research into cloud infrastructure, edge computing, distributed storage, and the management of large-scale distributed systems.

7. Is there any cost associated with using the NRP?

The cost of using the National Research Platform (NRP) can vary depending on several factors, including the type of user, the resources required, and the nature of the research. Here's an overview:

  • Academic and Research Institutions
    In many cases, access to the NRP for academic research projects is free or subsidized. This is particularly true if the research aligns with the platform's mission of advancing scientific discovery and innovation. However, some institutions may need to apply for grants or funding to cover any associated costs, depending on the scale of resources used.
  • Industry Professionals
    For companies and industry researchers, there may be fees associated with accessing the platform. These fees often depend on the level of resources (e.g., computing power, storage, specialized hardware) required and the length of usage. Industry partnerships or collaborations may provide discounted or customized pricing models.
  • Grants and Funding Opportunities
    Some research projects may be eligible for funding through grants from governmental organizations or research institutions. These grants can help cover the cost of using the NRP’s resources. Researchers should check with their institution or the NRP for available funding opportunities.
  • Project-Specific Pricing
    If a research project requires specialized resources, such as large-scale simulations or extended use of GPUs or FPGAs, the cost may be calculated based on the specific usage of these resources. Some platforms use a pay-per-use model, while others may offer bulk access for larger projects.
  • Collaborative Research
    Collaborative projects may benefit from shared costs, especially when multiple institutions are involved. In these cases, the cost structure can be negotiated based on the project's scope and the institutions involved.
    To determine the exact cost associated with the NRP, please consult with the Fresno State Research support team.

8. What is the National Research Platform’s infrastructure and network setup like?

The National Research Platform (NRP) is built on a highly distributed, scalable infrastructure designed to support advanced research across various fields. Here’s an overview of its infrastructure and network setup:

  • Distributed Nodes
    The NRP utilizes distributed computing nodes located in multiple data centers around the world. This global distribution allows researchers to conduct experiments and simulations that require large-scale computing power, while also enabling experiments that would be impossible on centralized clusters.
    The platform hosts a variety of nodes, including general-purpose CPUs, high-performance GPUs, FPGAs, and specialized storage systems. This allows researchers to choose the appropriate resource based on the specific needs of their experiments.
  • High-Performance Computing (HPC) Infrastructure
    The NRP is equipped with cutting-edge HPC resources for large-scale computational tasks. This infrastructure supports heavy processing needs for research areas such as AI/ML, simulations, big data analytics, and scientific modeling.
    The platform provides scalable storage solutions designed to handle vast amounts of data, crucial for research involving large datasets, simulations, and real-time processing.
  • Cloud-Based Architecture
    The platform employs a cloud-based infrastructure, meaning that users can access computational resources on-demand from anywhere in the world. This flexibility is key to fostering collaboration and making high-performance resources accessible to researchers across geographic boundaries.
    The cloud setup allows resources to be dynamically allocated based on the demands of individual research projects, optimizing performance and resource utilization.
  • High-Speed Networking
    The NRP is supported by a robust high-speed network infrastructure that enables rapid data transfer and communication between distributed nodes. This is essential for running simulations, processing large datasets, and supporting collaborative research across institutions.
    The network is designed to facilitate seamless communication between geographically dispersed nodes, ensuring that researchers can work collaboratively without being limited by distance or local infrastructure constraints.
  • Specialized Hardware Support
    The NRP integrates specialized hardware components such as FPGAs (Field-Programmable Gate Arrays), which provide enhanced flexibility and performance for specific research tasks that demand customized computing power.
    AI/ML Accelerators: High-performance GPUs are available to accelerate machine learning and AI research, allowing researchers to train complex models efficiently.
  • User Interfaces and Software Access
    JupyterHub and Coder: Researchers can access the NRP through user-friendly interfaces like JupyterHub for interactive computing and Coder for coding environments. These interfaces simplify the process of managing and running experiments, making it easier for researchers to interact with the platform.
    The platform comes equipped with commonly used research tools, including AI and ML frameworks (e.g., TensorFlow, PyTorch), scientific computing libraries, and data analysis software, further streamlining the research process.
  • Security and Compliance
    The NRP’s infrastructure includes robust security measures to ensure that research data remains protected. This includes encryption for data at rest and in transit, access controls, and secure authentication methods.
    The platform adheres to various regulatory standards to ensure data privacy, security, and ethical research practices, which is especially important for sensitive fields such as healthcare and bioinformatics.
  • Collaboration and Integration
    The NRP encourages collaboration by providing features that enable easy sharing of resources, data, and results across research teams and institutions. Researchers can work together across different geographic locations, pooling their computational resources for large-scale experiments.
    The NRP is designed to integrate with other academic, governmental, and industry research platforms, allowing for enhanced collaboration and data sharing.
    In summary, the NRP’s infrastructure and network setup are built to support high-performance research, provide scalability, foster collaboration, and ensure security and compliance, all while being accessible globally. This makes it a powerful tool for conducting cutting-edge, data-intensive research in various scientific fields.

9. What software and tools are available on the National Research Platform (NRP)?

The National Research Platform (NRP) provides a variety of software and tools to support a wide range of research activities. These tools are designed to enhance the capabilities of researchers in fields such as artificial intelligence (AI), machine learning (ML), data science, network simulations, and more. Below is the list of few software and tools available on the NRP:

  • TensorFlow – A widely-used framework for building machine learning and deep learning models.
  • PyTorch – Another popular deep learning framework used for AI research and development.
  • Scikit-learn – A robust library for traditional machine learning algorithms and data analysis.
  • Keras – A high-level neural networks API, written in Python, capable of running on top of TensorFlow or Theano.
  • CUDA – NVIDIA’s parallel computing platform and application programming interface (API) model that allows software to harness GPU power for parallel computing tasks.
  • Slurm – A job scheduler for allocating resources and scheduling tasks in large clusters, typically used in high-performance computing environments.
  • Hadoop – A framework that allows for the distributed processing of large data sets across clusters of computers using simple programming models.
  • Spark – An open-source unified analytics engine for big data processing, with built-in
  • JupyterHub – A multi-user version of Jupyter Notebooks that enables users to run code in various languages (Python, R, etc.) interactively, supporting collaboration across research teams.
  • Coder – A cloud-based development environment that simplifies coding workflows, allowing researchers to run, test, and deploy their code on the NRP infrastructure.
  • Git – Version control system for managing and collaborating on codebases and research projects, integrated into the platform for seamless collaboration.
  • Kerberos – A network authentication protocol that provides secure authentication for users accessing NRP resources.
  • OpenSSL – A software library for secure communication over networks using cryptographic protocols such as SSL/TLS.
  • Docker and Kubernetes – Containerization technology that ensures reproducibility and portability of research environments, isolating applications to improve security and performance.
  • Terraform – An open-source tool for managing infrastructure as code, enabling researchers to configure and provision resources within the cloud environment.
  • Matplotlib – A plotting library for Python that allows users to create static, animated, and interactive visualizations of data.
  • MATLAB – A high-performance language and environment for numerical computation, visualization, and programming, widely used in engineering and scientific research.

10. How secure is the data on the NRP?

How secure is the data on the National Research Platform (NRP)?
The National Research Platform (NRP) is designed with a strong emphasis on data security, ensuring that sensitive research data is protected from unauthorized access, breaches, or loss. Here's an overview of the key security measures in place on the platform:

  • Data Encryption
    All data transferred between users and the NRP infrastructure is encrypted using industry-standard protocols such as TLS (Transport Layer Security). Data stored on NRP’s servers is encrypted using advanced encryption algorithms
  • Access Control and Authentication
    NRP supports multi-factor authentication (MFA) to ensure that only authorized users can access the platform
  • Secure Data Storage and Backup
    NRP regularly backs up data to prevent data loss in case of system failures. These backups are stored securely and are also encrypted.
    NRP uses distributed storage systems to ensure high availability and fault tolerance. If one part of the infrastructure is compromised or fails, the data is still protected and can be retrieved from another secure location
  • Network Security
    Firewalls and Intrusion Detection Systems (IDS): NRP employs firewalls to prevent unauthorized access to its internal network and intrusion detection systems (IDS) to monitor and detect suspicious activity or potential threats in real-time.
    Virtual Private Networks (VPNs): For additional security, users may be required to connect via a VPN when accessing certain high-security resources on the platform, ensuring that the data transmission remains secure and private.
  • Compliance with Standards and Regulations
    GDPR Compliance: The NRP adheres to international data protection standards, including GDPR (General Data Protection Regulation) for research projects involving data from EU citizens. This ensures that personal data is handled and stored in a secure and compliant manner.
    HIPAA Compliance: For research in healthcare and bioinformatics, the NRP is designed to comply with HIPAA (Health Insurance Portability and Accountability Act), ensuring that sensitive medical data is protected according to U.S. regulations.
    FISMA and NIST Compliance: The NRP infrastructure complies with FISMA (Federal Information Security Modernization Act) and NIST (National Institute of Standards and Technology) cybersecurity standards, which set guidelines for securing federal information systems
  • Data Privacy Policies
    Data Ownership: Researchers retain ownership of their data. The NRP platform implements strict data privacy policies that ensure data is only used for its intended purpose and is not shared without the user’s consent.
    Anonymization and Pseudonymization: For research involving personal or sensitive data, the NRP provides tools for anonymization and pseudonymization to ensure that individual identities are not exposed.
  • Auditing and Monitoring
    Activity Logging: All access to the platform and sensitive data is logged and monitored. Detailed audit trails allow administrators to track who accessed data, when, and what actions were taken. This helps detect unauthorized access and supports accountability.
    Continuous Security Monitoring: The NRP infrastructure is continuously monitored for potential security threats, ensuring quick detection and response to vulnerabilities or attacks
  • Incident Response and Disaster Recovery
    Incident Response Plan: The NRP has a well-defined incident response plan in place to handle any security breaches or data incidents. This includes immediate containment, investigation, and remediation to minimize impact on research projects.
    Disaster Recovery: The platform has disaster recovery protocols to quickly restore services in case of major disruptions. This ensures that data can be recovered, and systems restored promptly

NRP implements a comprehensive and multi-layered security framework that incorporates encryption, access control, network security, compliance with regulatory standards, and ongoing monitoring. These measures ensure that the platform remains a secure environment for researchers working with sensitive and critical data.

11. Can I collaborate with other researchers through the NRP?

Yes, the National Research Platform (NRP) is designed to foster collaboration and facilitate team-based research across institutions and organizations. Different ways to collaborate through the NRP:

  • Shared Resources and Infrastructure
  • Collaborative Workspaces
  • Integration with Collaborative Tools
  • Collaboration Across Institutions
  • Co-Development of Experiments and Tools
  • Collaboration with Industry and Government
  • Remote Collaboration on Network and System Experiments
  • Collaborative Data Analysis
  • Researcher Collaboration via Cloud APIs
  • Collaboration on Publication and Results Sharing

Additional Information:

Need additional information or assistance? Contact the Technology Service Desk at (559) 278-5000. 

TAGS: AI, OpenAI, AI supercomputing

Print Article