Sparse observations and coarse-resolution climate models limit effective regional decision-making, underscoring the need for robust downscaling. However, existing AI methods struggle with generalization across variables and geographies and are constrained by the quadratic complexity of Vision Transformer (ViT) self-attention. We introduce ORBIT-2, a scalable foundation model for global, hyper-resolution climate downscaling. ORBIT-2 incorporates two key innovations: (1) Residual Slim ViT (Reslim), a lightweight architecture with residual learning and Bayesian regularization for efficient, robust prediction; and (2) TILES, a tile-wise sequence scaling algorithm that reduces self-attention complexity from quadratic to linear, enabling long-sequence processing and massive parallelism. ORBIT-2 scales to 10 billion parameters across 65,536 GPUs, achieving up to 4.1 ExaFLOPS sustained throughput and 74–98% strong scaling efficiency. It supports downscaling to 0.9 km global resolution and processes sequences up to 4.2 billion tokens. On 7 km resolution benchmarks, ORBIT-2 achieves high accuracy with R2 scores in range of 0.98–0.99 against observation data.

We present our work on developing and training scalable, trustworthy, and energy-efficient predictive graph foundation models (GFMs) using HydraGNN, a multi-headed graph convolutional neural network architecture. HydraGNN expands the boundaries of graph neural network (GNN) computations in both training scale and data diversity. It abstracts over message passing algorithms, allowing both reproduction of and comparison across algorithmic innovations that define nearest-neighbor convolution in GNNs. This work discusses a series of optimizations that have allowed scaling up the GFMs training to tens of thousands of GPUs on datasets consisting of hundreds of millions of graphs. Our GFMs use multitask learning (MTL) to simultaneously learn graph-level and node-level properties of atomistic structures, such as energy and atomic forces. Using over 154 million atomistic structures for training, we illustrate the performance of our approach along with the lessons learned on two state-of-the-art US Department of Energy (US-DOE) supercomputers, namely the Perlmutter petascale system at the National Energy Research Scientific Computing Center and the Frontier exascale system at Oak Ridge Leadership Computing Facility. The HydraGNN architecture enables the GFM to achieve near-linear strong scaling performance using more than 2000 GPUs on Perlmutter and 16,000 GPUs on Frontier.

Earth system predictability is challenged by the complexity of environmental dynamics and the multitude of variables involved. Current AI foundation models, although advanced by leveraging large and heterogeneous data, are often constrained by their size and data integration, limiting their effectiveness in addressing the full range of Earth system prediction challenges. To overcome these limitations, we introduce the Oak Ridge Base Foundation Model for Earth System Predictability (ORBIT), an advanced vision transformer model that scales up to 113 billion parameters using a novel hybrid tensor-data orthogonal parallelism technique. As the largest model of its kind, ORBIT surpasses the current climate AI foundation model size by a thousandfold. Performance scaling tests conducted on the Frontier supercomputer have demonstrated that ORBIT achieves 684 petaFLOPS to 1.6 exaFLOPS sustained throughput, with scaling efficiency maintained at 41% to 85% across 49,152 AMD GPUs. These breakthroughs establish new advances in AIdriven climate modeling and demonstrate promise to significantly improve the Earth system predictability.

The pace of advancement of the top-end supercomputers historically followed an exponential curve similar to (and driven in part by) Moore's Law. Shortly after hitting the petaflop mark, the community started looking ahead to the next milestone: Exascale. However, many obstacles were already looming on the horizon, such as the slowing of Moore's Law, and others like the end of Dennard Scaling had already arrived. Anticipating significant challenges for the overall high-performance computing (HPC) community to achieve the next 1000x improvement, the U.S. Department of Energy (DOE) launched the Exascale Computing Program to enable and accelerate fundamental research across the many technologies needed to achieve exascale computing.

Sparse solvers are heavily used in computational fluid dynamics (CFD), computer-aided design (CAD), and other important application domains. These solvers remain challenging to execute on massively parallel architectures, due to the sequential dependencies between the fine-grained application tasks. In particular, parallel sparse solvers typically suffer from substantial scheduling and dependency-management overheads relative to the compute operations. We propose adaptive task aggregation (ATA) to efficiently execute such irregular computations on GPU architectures via hierarchical dependency management and low-latency task scheduling. On a gamut of representative problems with different data-dependency structures, ATA significantly outperforms existing GPU task-execution approaches, achieving a geometric mean speedup of 2.2X to 3.7X across different sparse kernels (with speedups of up to two orders of magnitude).

On recent GPU architectures, dynamic parallelism, which enables the launching of kernels from the GPU without CPU involvement, provides a way to improve the performance of irregular applications by generating child kernels dynamically to reduce workload imbalance and improve GPU utilization. However, in practice, dynamic parallelism does not improve performance due to high kernel launch overhead and low child kernel occupancy. Consequently, most existing studies focus on mitigating the kernel launch overhead. As the kernel launch overhead has decreased due to algorithmic redesigns and hardware architectural innovations, the organization of subtasks to child kernels becomes a new performance bottleneck.

Data movement in high-performance computing systems accelerated by graphics processing units (GPUs) remains a challenging problem. Data communication in popular parallel programming models, such as the Message Passing Interface (MPI), is currently limited to the data stored in the CPU memory space. Auxiliary memory systems, such as GPU memory, are not integrated into such data movement standards, thus providing applications with no direct mechanism to perform end-to-end data movement. We introduce MPI-ACC, an integrated and extensible framework that allows end-to-end data movement in accelerator-based systems. MPI-ACC provides productivity and performance benefits by integrating support for auxiliary memory spaces into MPI. MPI-ACC supports data transfer among CUDA, OpenCL and CPU memory spaces and is extensible to other offload models as well. MPI-ACC's runtime system enables several key optimizations, including pipelining of data transfers, scalable memory management techniques, and balancing of communication based on accelerator and node architecture. MPI-ACC is designed to work concurrently with other GPU workloads with minimum contention. We describe how MPI-ACC can be used to design new communication-computation patterns in scientific applications from domains such as epidemiology simulation and seismology modeling, and we discuss the lessons learned. We present experimental results on a state-of-the-art cluster with hundreds of GPUs; and we compare the performance and productivity of MPI-ACC with MVAPICH, a popular CUDA-aware MPI solution. MPI-ACC encourages programmers to explore novel application-specific optimizations for improved overall cluster utilization.

OpenCL is a portable interface that can be used to program cluster nodes with heterogeneous compute devices. The OpenCL specification tightly binds its workflow abstraction, or "command queue," to a specific device for the entire program. For best performance, the user has to find the ideal queue -- device mapping at command queue creation time, an effort that requires a thorough understanding of the match between the characteristics of all the underlying device architectures and the kernels in the program. In this paper, we propose to add scheduling attributes to the OpenCL context and command queue objects that can be leveraged by an intelligent runtime scheduler to automatically perform ideal queue - device mapping. Our proposed extensions enable the average OpenCL programmer to focus on the algorithm design rather than scheduling and automatically gain performance without sacrificing programmability. As an example, we design and implement an OpenCL runtime for task-parallel workloads, called MultiCL, which efficiently schedules command queues across devices. Within MultiCL, we implement several key optimizations to reduce runtime overhead. Our case studies include the SNU-NPB OpenCL benchmark suite and a real-world seismology simulation. We show that, on average, users have to apply our proposed scheduler extensions to only four source lines of code in existing OpenCL applications in order to automatically benefit from our runtime optimizations. We also show that MultiCL always maps command queues to the optimal device set with negligible runtime overhead.

Power-hungry Graphics processing unit (GPU) accelerators are ubiquitous in high performance computing data centers today. GPU virtualization frameworks introduce new opportunities for effective management of GPU resources by decoupling them from application execution. However, power management of GPU-enabled server clusters faces significant challenges. The underlying system infrastructure shows complex power consumption characteristics depending on the placement of GPU workloads across various compute nodes, power-phases and cabinets in a datacenter. GPU resources need to be scheduled dynamically in the face of time-varying resource demand and peak power constraints. We propose and develop a power-aware virtual OpenCL (pVOCL) framework that controls the peak power consumption and improves the energy efficiency of the underlying server system through dynamic consolidation and power-phase topology aware placement of GPU workloads. Experimental results show that pVOCL achieves significant energy savings compared to existing power management techniques for GPU-enabled server clusters, while incurring negligible impact on performance. It drives the system towards energy-efficient configurations by taking an optimal sequence of adaptation actions in a virtualized GPU environment and meanwhile keeps the power consumption below the peak power budget.

We present a fully automated approach to project the relative performance of an OpenCL program over different GPUs. Performance projections can be made within a small amount of time, and the projection overhead stays relatively constant with the input data size. As a result, the technique can help runtime tools make dynamic decisions about which GPU would run faster for a given kernel. Usage cases of this technique include scheduling or migrating GPU workloads over a heterogeneous cluster with different types of GPUs.

Graphics processing units (GPUs) have been widely used to accelerate algorithms that exhibit massive data parallelism or task parallelism. When such parallelism is not inherent in an algorithm, computational scientists resort to simply replicating the algorithm on every multiprocessor of a NVIDIA GPU, for example, to create such parallelism, resulting in embarrassingly parallel ensemble runs that deliver significant aggregate speed-up. However, the fundamental issue with such ensemble runs is that the problem size to achieve this speed-up is limited to the available shared memory and cache of a GPU multiprocessor. An example of the above is dynamic programming (DP), one of the Berkeley 13 dwarfs. All known DP implementations to date use the coarse-grained approach of embarrassingly parallel ensemble runs because a fine-grained parallelization on the GPU would require extensive communication between the multiprocessors of a GPU, which could easily cripple performance as communication between multiprocessors is not natively supported in a GPU. Consequently, we address the above by proposing a fine-grained parallelization of a single instance of the DP algorithm that is mapped to the GPU. Our parallelization incorporates a set of techniques aimed to substantially improve GPU performance: matrix re-alignment, coalesced memory access, tiling, and GPU (rather than CPU) synchronization. The specific DP algorithm that we parallelize is called Smith-Waterman (SWat), which is an optimal local-sequence alignment algorithm. We then use this SWat algorithm as a baseline to compare our GPU implementation, i.e., CUDA-SWat, to our implementation on the cell broadband engine, i.e., Cell-SWat.

The Smith-Waterman algorithm is a dynamic programming method for determining optimal local alignments between nucleotide or protein sequences. However, it suffers from quadratic time and space complexity. As a result, many algorithmic and architectural enhancements have been proposed to solve this problem, but at the cost of reduced sensitivity in the algorithms or significant expense in hardware, respectively.