Written and maintained by CASRAI Editorial Board
Last updated
The ASCR Leadership Computing Challenge (ALCC) is a competitive high-performance computing allocation program administered by the U.S. Department of Energy (DOE) Office of Science. Managed by the Advanced Scientific Computing Research (ASCR) program office, ALCC awards millions of node-hours on leadership-class supercomputers to advance mission-critical scientific discovery, national emergency response efforts, and large-scale artificial intelligence campaigns.
What it is used for in research
The ALCC program provides compute allocations for projects of interest to the Department of Energy, with a specific mandate to support high-risk, high-payoff scientific research. In contemporary computational science, ALCC allocations serve as a cornerstone for massive simulation workflows, large-scale data analytics, and computational AI for science.
Modern ALCC campaigns frequently focus on foundation models for scientific discovery, training multi-modal architectures across materials science, genomics, climate modeling, and particle physics. Researchers deploy these resources for distributed machine learning workflows, such as scaling physics-informed neural networks to emulate complex fluid dynamics, coordinating GPU-accelerated computing pipelines using CuPy, and handling extreme-scale parallel I/O with h5py across parallel Lustre and Orion filesystems.
Computing time under ALCC is allocated across three primary DOE user facilities:
- Oak Ridge Leadership Computing Facility (OLCF): Located at Oak Ridge National Laboratory, hosting the Frontier exascale supercomputing platform.
- Argonne Leadership Computing Facility (ALCF): Located at Argonne National Laboratory, hosting the Aurora exascale system and the Polaris accelerated cluster.
- National Energy Research Scientific Computing Center (NERSC): Located at Lawrence Berkeley National Laboratory, hosting the Perlmutter GPU-accelerated and CPU system.
The DOE Office of Science reserves between 10% and 30% of total allocatable capacity on these machines for the ALCC program. Unlike conventional competitive research grants funded under DOE ASCR programs, ALCC awards compute node-hours rather than direct financial awards, enabling research groups in academia, industry, and national laboratories to execute computing campaigns that exceed standard institutional capacity.
Install or access (official source only)
ALCC is not an installable software package; access is granted through a formal, peer-reviewed allocation proposal submitted through the official DOE Office of Science intake platform. Proposals are managed electronically via the ASCR PeerNet submission system.
Allocations run on a annual cycle operating from July 1 through June 30 of the following year. Multi-year awards can accommodate project requests for up to three years, subject to annual renewal proposal reviews. To ensure equitable access across the research community, ASCR enforces strict allocation request caps per project across each computing platform.
| HPC Platform and Facility | Total Available Pool | Individual Request Cap | System Architecture |
|---|---|---|---|
| Frontier (OLCF) | 20.0 × 106 node-hours | 2.0 × 106 node-hours | AMD EPYC CPUs + AMD Instinct MI250X GPUs |
| Aurora (ALCF) | 16.0 × 106 node-hours | 2.0 × 106 node-hours | Intel Xeon CPU Max + Intel Data Center GPU Max |
| Polaris (ALCF) | 0.75 × 106 node-hours | 0.15 × 106 node-hours | AMD EPYC CPUs + NVIDIA A100 GPUs |
| Perlmutter-GPU (NERSC) | 1.25 × 106 node-hours | 0.30 × 106 node-hours | AMD EPYC CPUs + NVIDIA A100 GPUs |
| Perlmutter-CPU (NERSC) | 2.25 × 106 node-hours | 0.55 × 106 node-hours | Dual-socket AMD EPYC 7763 CPUs |
Three rigid compliance rules apply to all ALCC applications:
- GPU utilization requirement: Proposals requesting time on Frontier, Aurora, Polaris, or Perlmutter-GPU that do not actively leverage GPUs are deemed non-responsive and will not be reviewed.
- Request cap compliance: Any proposal requesting more than the published ceiling for a given platform is declared non-responsive and declined without merit review.
- Physical node-hour units: Requests must be calculated in raw physical node-hours without facility site charging factors, entered into the submission form scaled as values × 106.
Step-by-step task
Securing and utilizing an ALCC allocation follows a structured operational workflow spanning benchmarking, submission, account creation, and job execution:
- Conduct benchmark scaling: Execute preliminary scaling tests on target hardware (via institutional clusters or Director’s Discretion access) to prove weak and strong scaling. Record execution wall-clock time, GPU memory residency, and parallel efficiency across multi-node topologies.
- Calculate total node-hour requests: Compute node requirements using the formula:
Physical Node-Hours = Total Nodes Required × Production Wall-Clock Hours per Run × Total Planned Production RunsConfirm that the sum falls strictly within the machine’s individual cap (for example, ≤ 2.0 × 106 node-hours on Frontier).
- Prepare official proposal templates: Complete the standardized proposal package:
- Project Information, Area of Alignment, and Science Category Summary.
- Funding Project Financial Support (identifying sponsoring grants and grant numbers).
- HPC Resource Request: code architecture description, scaling curves, I/O bandwidth needs, and Integrated Research Infrastructure (IRI) relevance.
- Milestone Table and Curriculum Vitae for Principal Investigators and Co-PIs.
- Submit through the PeerNet intake portal: Log into the PeerNet portal during the open submission window and upload all required documents before the 8:00 PM Eastern Time deadline. ALCC operates a single-step submission without a preliminary pre-proposal stage.
- Complete facility account onboarding: Once awarded, the Principal Investigator receives an official project ID. Project collaborators submit individual user account requests through facility portals (myOLCF at Oak Ridge, Iris at NERSC, or the ALCF Account Management system) and configure required multi-factor authentication (such as RSA SecurID tokens).
- Launch batch jobs with project billing: In the Slurm workload manager, assign all production runs to the awarded ALCC project ID using the
-Aor--accountdirective:#!/bin/bash #SBATCH -A proj1234 #SBATCH -J ai_training_run #SBATCH -N 128 #SBATCH -t 06:00:00 #SBATCH -p batch srun ./train_foundation_model --config config.yaml
Output and how to read it
Upon review completion, Principal Investigators receive an official decision notice detailing the awarded platform, total approved physical node-hours, and start date (July 1). Facility project management dashboards track allocation usage:
- OLCF (myOLCF): Displays overall project consumption, remaining balances, and graphical burn-rate trajectories against historical averages.
- NERSC (Iris): Provides real-time tracking of raw node-hours and repository allocations through the
checksharescommand line utility. - ALCF (sbank): Tracks project balances through the
sbank-detailcommand line tool, showing node-hour deductions per completed queue job.
Researchers must regularly monitor burn rates. Allocations expire at the conclusion of the allocation year on June 30 and do not roll over to subsequent cycles.
Troubleshooting
The most common operational challenge for ALCC awardees is maintaining the required computational pace to avoid allocation clawbacks under ASCR’s mid-year forfeiture policy:
- Mid-year allocation reductions: To ensure high utilization of exascale supercomputing infrastructure, ASCR executes semi-annual allocation reductions on November 1 and March 1:
- If a project has consumed less than 10% of its allocation by the review checkpoint, ASCR removes 25% of the unused balance.
- If usage is between 10% and 25%, ASCR removes 15% of the unused balance.
- At the second checkpoint, projects utilizing less than 25% forfeit 50% of the remaining balance, while usage between 25% and 50% forfeits 25%.
- Exemption appeals: Principal Investigators receive email notifications two weeks prior to scheduled reductions. PIs encountering unexpected technical barriers (such as major system outages or code refactoring) must submit an exemption request in writing at least one week prior to the reduction date via email to
[email protected]. - GPU runtime failures: Jobs terminating with GPU initialization errors usually indicate missing vendor runtime modules. Ensure the appropriate programming environment is loaded before submission (for example, ROCm modules on Frontier, oneAPI compilers on Aurora, or NVIDIA CUDA modules on Polaris and Perlmutter).
Free alternatives
Researchers seeking supercomputing time can explore several peer competitive allocation mechanisms across national laboratories and federal research programs:
| Allocation Program | Governing Agency | Allocation Scope and Target Systems | Access Model | Best Suited For |
|---|---|---|---|---|
| ALCC | DOE Office of Science (ASCR) | 10% to 30% of capacity at ALCF, OLCF, and NERSC | Annual single-step competitive peer review via PeerNet | High-risk mission science, national emergencies, and scalable AI workflows |
| INCITE | DOE Office of Science (ASCR) | Up to 60% of leadership capacity at ALCF and OLCF | Annual competitive international review | Grand-challenge, highly scaled exascale simulations requiring massive node concurrency |
| Director’s Discretion (DD) | Individual Facilities (ALCF / OLCF / NERSC) | Small exploratory allocations (thousands of node-hours) | Rolling facility application portal | Code porting, GPU profiling, scaling verification, and proposal preparation |
| NSF ACCESS | National Science Foundation | National academic cyberinfrastructure (Stampede3, Bridges-2, Delta) | Tiered proposal submissions (Explore, Discover, Accelerate, Maximize) | Academic researchers funded across general NSF disciplines |
FAQ
What does ALCC stand for in high-performance computing?
ALCC stands for the ASCR Leadership Computing Challenge, a program administered by the Advanced Scientific Computing Research division of the U.S. Department of Energy Office of Science.
How does ALCC differ from the INCITE program?
While both programs allocate time on leadership-class supercomputers, INCITE focuses on computationally grand-challenge projects demanding massive node scaling (awarding up to 60% of facility time), whereas ALCC specifically supports DOE mission-aligned campaigns, rapid responses to national emergencies, and projects that expand access to leadership computing for AI and data science.
Who is eligible to apply for an ALCC allocation?
The ALCC program is open to researchers from academia, national laboratories, and commercial industry worldwide. Applicants do not need existing DOE grant funding, provided the proposed research demonstrates clear scientific merit and alignment with DOE mission areas.
Can an ALCC allocation be renewed across multiple years?
Yes. ALCC accommodates multi-year project requests for up to three years. However, awarded projects must submit an annual renewal proposal demonstrating progress against established milestones to secure allocation hours for subsequent years.
What are the mandatory publication acknowledgment requirements for ALCC?
All publications resulting from ALCC awards must explicitly acknowledge both the ALCC program and the host user facility using the official format: “An award of computer time was provided by the ASCR Leadership Computing Challenge (ALCC) program. This research used resources of the [Argonne Leadership Computing Facility / Oak Ridge Leadership Computing Facility / National Energy Research Scientific Computing Center], a DOE Office of Science User Facility.”
Sources
- U.S. Department of Energy Office of Science, Advanced Scientific Computing Research (ASCR), ALCC Program Overview (accessed 2026-11-08): https://science.osti.gov/ascr/Facilities/Accessing-ASCR-Facilities/ALCC
- U.S. Department of Energy Office of Science ASCR, Call for Proposals and Allocation Request Cap (accessed 2026-11-08): https://science.osti.gov/ascr/Facilities/Accessing-ASCR-Facilities/ALCC/Allocation-Request-Cap
- U.S. Department of Energy Office of Science ASCR, Proposal Submission Information (accessed 2026-11-08): https://science.osti.gov/ascr/Facilities/Accessing-ASCR-Facilities/ALCC/Proposal-Information
- U.S. Department of Energy Office of Science ASCR, Allocation Reduction Information (accessed 2026-11-08): https://science.osti.gov/ascr/Facilities/Accessing-ASCR-Facilities/ALCC/Allocation-Reduction-Information
- Oak Ridge Leadership Computing Facility (OLCF), Accounts and Projects User Documentation (accessed 2026-11-08): https://docs.olcf.ornl.gov/accounts/accounts_and_projects.html
