Johns Hopkins IT develops and supports high-performance computing resources, pathways, and operations that enable research, discovery, and learning for the Johns Hopkins research community.
Supported by the Research IT infrastructure environment, JH IT provides investigators with computing resources that are suitable for varying compliance levels, including HIPAA, PHI/PII, LDS, and non-PHI/PII, as well as more strict security standards such as National Institute of Standards and Technology Special Publication 800-171 and Cybersecurity Maturity Model Certification (CMMC) Level 2 requirements to meet the needs of protected datasets.
Johns Hopkins IT provides the following Research Computing services:
High-performance Computing
Johns Hopkins IT offers advanced, on-premise computing resources through its shared computing facility Advanced Research Computing at Hopkins (ARCH)–comprised of the Rockfish and DSAI clusters–and its DISCOVERY high-performance computing offering.
| GET STARTED WITH ARCH | GET STARTED WITH DISCOVERY |
Access
ARCH is accessible* to JHU faculty, staff, and students, as well as external universities and institutions.
*Note: Access requests require sponsorship by a Johns Hopkins University Principal Investigator.
DISCOVERY access requests require a principal investigator’s approval.
Data Security & Use
Data use within ARCH is subject to restrictions. Data containing–but not limited to–HIPAA, PHI, or CUI is not permitted on ARCH systems. If your research involves an IRB and the data is de-identified, please reach out to [email protected] for further guidance prior to storing or processing any data.
DISCOVERY is PHI-compliant for both analysis and storage and is suitable for projects that require NIST SP 800-171 and/or Cybersecurity Maturity Model Certification (CMMC) Level 2 data compliance, in addition to HIPAA, PHI/PII, LDS, and non PHI/PII.
Capacity
ARCH
| ARCH | |||||||
|---|---|---|---|---|---|---|---|
| Node Type | Number of Nodes | Processor | GPUs | RAM | Storage | Total Cores | Total GPUs |
| Compute | 720 | Intel Xeon Gold 6248R (Cascade Lake) | N/A | 192 GB DDR4 2933MHz | 1 TB NVMe SSD | 36,864 | N/A |
| Compute | 80 | Intel Xeon Platinum 8480+ 56 Core | N/A | 512GB | N/A | 8,960 | N/A |
| Compute (Next-Gen) | 46 | Intel Xeon Gold 6448Y (Sapphire Rapids) | N/A | 256 GB DDR5 4800MHz | 2 TB NVMe SSD | 2,944 | N/A |
| Large Memory | 25 | Intel Xeon Gold 6248R (Cascade Lake) | N/A | 1.5 TB DDR4 2933MHz | 1 TB NVMe SSD | 1,200 | N/A |
| GPU (A100 40GB) | 18 | Intel Xeon Gold 6248R (Cascade Lake) | 4 × NVIDIA A100 40GB | 192 GB DDR4 2933MHz | 1 TB NVMe SSD | 864 | N/A |
| Fast Compute (Emerald Rapids) | 18 | Intel Xeon Gold / Platinum 8592+ | N/A | 512 GB TruDDR5 5600MHz | 1.8 TB NVMe SSD | 2,304 | N/A |
| GPU Nodes | 16 | AMD EPYC 9534 64 Core | 4x Nvidia H100 80GB | 1.5TB | N/A | 2,048 | 64 |
| GPU Nodes | 16 | AMD EPYC 9534 64 Core | 4x Nvidia H100-NVL 96GB | 1.5TB | N/A | 2,048 | 64 |
| GPU | 15 | AMD EPYC 7443 24 Core | 8x Nvidia A100 80GB | 1TB | N/A | 720 | 120 |
| GPU (A100 80GB) | 10 | Intel Xeon Gold 6338 (Icy Lake) | 4 × NVIDIA A100 80GB | 256 GB DDR4 3200MHz | 1.6 TB NVMe SSD | 640 | N/A |
| GPU | 8 | AMD EPYC 9534 64 Core | 8x Nvidia L40S 48GB | 768GB | N/A | 1,024 | 64 |
| GPU (L40S) | 4 | Intel Xeon Gold 6338 | 8 × NVIDIA L40S 48GB | 512 GB DDR4 3200MHz | 3.5 TB NVMe SSD | 256 | N/A |
DISCOVERY
| DISCOVERY | |||||||
|---|---|---|---|---|---|---|---|
| Node Type | Number of Nodes | Processor | GPUs | RAM | Storage | Total Cores | Total GPUs |
| Compute | 86 | Intel,Platinum-8168 | N/A | 716GB-1.5TB | 0 | 4132 | N/A |
| Compute | 14 | Intel,Platinum-8270 | N/A | 716GB | 0 | 728 | N/A |
| Compute | 37 | Intel,Platinum-8480+ | N/A | 512GB | 0 | 4144 | N/A |
| Compute | 7 | Intel,Xeon-6248R | N/A | 1.5TB | 0 | 336 | N/A |
| GPU Nodes | 27 | Intel,Gold-6248R | NVIDIA T4 | 1.5TB | 0 | 912 | 162 |
| GPU Nodes | 3 | Intel, Xeon Silver 4314 | NVIDIA A100 | 174GB | 0 | 72 | 3 |
| GPU Nodes | 16 | Intel Xeon Platinum 8592 | NVIDIA H100 | 1TB | 0 | 2048 | 64 |
| GPU Nodes | 18 | Intel,Platinum-8480C | NVIDIA H200 | 1.5TB | 42GB | 1728 | 144 |
Cost Structure
A subsidized cost model for JH IT-managed high-performance compute resources is forthcoming.
Facility Specifications
ARCH provides several resources to enhance researchers’ computational needs including:
- Research planning, strategies for resource (allocation) requests and effective utilization
- Traditional HPC simulations that need large amounts of CPU cycles, memory and I/O intensive
- Interactive data analysis (visualization, Jupyter notebooks, R-Studio)
- Use of Parallel, scalable file systems to improve performance and analyze much larger datasets
- Fast compute node and storage connectivity to reduce latency
- Design and implementation of research workflows
- Scientific support on emerging fields and disciplines (ML, AI)
- Scientific applications support, installation and maintenance
- Reliable file systems for short-term data storage, including data management
- Development of data intensive computing tools (Containers, VMs)
- Scientific support and consulting on HPC related topics (python, go)
- Fast data transfer via HORNet or DMZ using managed Globus end points
- Education/training resources
- Server colocation facilities
DISCOVERY provides several benefits for researchers’ high-performance computing needs. Offering a home directory with 100 GB allocations, DISCOVERY provides a secure location for custom libraries, packages, and environment builds for managed workflows. DISCOVERY also has a project directory with 1 to 50 allocations, enabling a collaborative environment for research groups.
Other benefits provided by DISCOVERY include:
- A SLURM Workload Manager with multidisciplinary modules for:
- Genomics/Transcriptomic alignment and analyses
- DNA Sequencing quality control and annotation
- Machine/Deep Learning algorithm development
- Data engineering and management
- Multimodal Image analyses
- An OnDemand Web Interface for secure interactions with the HPC to:
- Create SLURM job allocations for Graphical User Interface (GUI) applications such as Rstudio, Jupyter Lab, Jupyter Notebook, MatLab, and VS Code
- Provide an Secure Shell (SSH) protocol to modify and manage data structures within the HPC environment
- Run a Virtual Desktop using RHEL 9 with SLURM allocated resources for a point and click interface for image/video manipulation, segmentation analyses, and more
- Globus High Assurance transfer protocols to securely transfer data between RIT-managed HPC endpoints and researcher controlled devices
- Various application modules and extensions*
*Note: JHED authentication is required to view this page.
Cloud Computing
IT@JH provides managed cloud services tailored for specialized research needs. These secure and flexible solutions are set up on cloud infrastructure with Microsoft Azure and AWS.