Distributed GPU applications increasingly rely on kernel-level, cross-node coordination to reduce launch overheads and improve compute-communication overlap, but such support is lacking. On OFI-based interconnects such as HPE Slingshot, which powers six of the top ten systems in the November 2025 Top500, including the top three, GPU kernels cannot autonomously drive distributed coordination: existing runtimes rely on host-driven progress and lack a bounded mechanism for recycling pre-staged NIC work across repeated GPU-triggered operations. On InfiniBand, GPU-initiated communication is possible, but current implementations incur unnecessary synchronization and locking overheads. This paper presents GICC, a framework that enables GPU kernels to directly trigger NIC-level operations without host involvement on the fast path. In stencils, GPU threads initiate halo exchanges as soon as boundary regions are computed, enabling fine-grained overlap between interior computation and boundary transfer. GICC decouples coordination semantics from data movement and introduces asynchronous resource reclamation: the NIC signals completion to both GPU and host memory, letting a lightweight host thread recycle NIC resources concurrently with GPU execution without injecting latency into the coordination path. This sustains GPU-driven coordination under finite NIC state, absent from existing OFI-based runtimes. We implement GICC on NVIDIA and AMD GPUs over InfiniBand and Slingshot. On Slingshot, GICC reduces per-coordination latency by up to 229x and improves weak scaling efficiency by up to 25%. On InfiniBand, it achieves up to 1.95x lower put latency than NVSHMEM by eliminating unnecessary locking and synchronization. On an industrial stencil proxy on 64 AMD MI250X GCDs, GPU-aware MPI incurs over 52% higher communication time than GICC, which achieves 42% parallel efficiency versus MPI's 35.4%.

Fast and accurate numerical simulations are crucial for designing large-scale geological carbon storage projects ensuring safe long-term CO2containment as a climate change mitigation strategy. These simulations involve solving numerous large and complex linear systems arising from the implicit Finite Volume (FV) discretization of PDEs governing subsurface fluid flow. Compounded with highly detailed geomodels, solving linear systems is computationally and memory expensive, and accounts for the majority of the simulation time. Modern memory hierarchies are insufficient to meet the latency and bandwidth needs of large-scale numerical simulations. Therefore, exploring algorithms that can leverage alternative and balanced paradigms such as dataflow and in-memory computing is crucial. This work introduces a matrix-free algorithm to solve FV-based linear systems using a dataflow architecture to significantly minimize memory latency and bandwidth bottlenecks. Our implementation achieves two orders of magnitude speedup compared to a GPGPU-based reference implementation, and up to 1.2 PFlops on a single dataflow device.

Finite-difference methods based on high-order stencils are widely used in seismic simulations, weather forecasting, and computational fluid dynamics. Recently, multiple research groups have begun exploring the use of dataflow architectures, such as Cerebras’ wafer-scale engine, to accelerate stencil computations. However, implementations of stencil computations for dataflow architectures must address unique challenges, such as managing the routing of data communications and accommodating a significantly constrained memory footprint. These make hand-crafting code for a dataflow architecture difficult and time-consuming. This paper describes a framework for developing portable, high-performance implementations of stencil computations for modern node architectures. The paper focuses on code generation strategies for the Cerebras wafer-scale engine, including code generation of router configurations and sequencing of communication for high-order stencils. A 25-point starshaped stencil written using our tool is $7 \times$ shorter than handcrafted code written in Cerebras Software Language (CSL), and it delivers comparable performance to manually written code.

We introduce two algorithms that invert simulated gravity data to 3D subsurface rock/flow properties. The first algorithm is a data-driven, deep learning-based approach, and the second is also data-driven but considers the temporal evolution of surface gravity events. The target application of these proposed algorithms is the prediction of subsurface CO2plumes as a complementary tool for monitoring CO2sequestration deployments. Each proposed algorithm outperforms traditional inversion methods and produces high-resolution, 3D subsurface reconstructions in near real-time. In addition, our proposed methods achieve Dice scores of up to 0.8 for predicted plume geometry and near-perfect data misfit in terms of μGals. These results indicate that combining 4D surface gravity monitoring (low-cost acquisition) with deep learning techniques represents an effective and non-intrusive method for monitoring CO2storage sites.

From AlexNet to Inception, autoencoders to diffusion models, the development of novel and powerful deep learning models and learning algorithms has proceeded at breakneck speeds. In part, we believe that rapid iteration of model architecture and learning techniques by a large community of researchers over a common representation of the underlying entities has resulted in transferable deep learning knowledge. As a result, model scale, accuracy, fidelity, and compute performance have dramatically increased in computer vision and natural language processing. On the other hand, the lack of a common representation for chemical structure has hampered similar progress. To enable transferable deep learning, we identify the need for a robust 3-dimensional representation of materials such as molecules and crystals. The goal is to enable both materials property prediction and materials generation with 3D structures. While computationally costly, such representations can model a large set of chemical structures. We propose ParticleGrid, a SIMD-optimized library for 3D structures, that is designed for deep learning applications and to seamlessly integrate with deep learning frameworks. Our highly optimized grid generation allows for generating grids on the fly on the CPU, reducing storage and GPU compute and memory requirements. We show the efficacy of 3D grids generated via ParticleGrid and accurately predict molecular energy properties using a 3D convolutional neural network. Our model is able to get 0.006 mean square error and nearly match the values calculated using computationally costly density functional theory at a fraction of the time.

Stencil computations lie at the heart of many scientific and industrial applications. Stencil algorithms pose several challenges on machines with cache based memory hierarchy, due to low re-use of memory accesses if special care is not taken to optimize them. This work shows that for stencil computation a novel algorithm that leverages a localized communication strategy effectively exploits the second generation Cerebras Wafer-Scale Engine (WSE-2), which has no cache hierarchy. This study focuses on a 25-point stencil finite-difference method for the 3D wave equation, a kernel frequently used in earth modeling as numerical simulation. In essence, the algorithm trades memory accesses for data communication and takes advantage of the fast communication fabric provided by the architecture. The algorithm -historically memory-bound- becomes compute-bound. This allows the implementation to achieve near perfect weak scaling, reaching up to 503 TFLOPs on WSE-2, a figure that only full clusters can eventually yield.

Summary Finite‐difference methods based on high‐order stencils are commonly used for modeling of seismic wave propagation, weather forecasting, computational fluid dynamics, convolutional neural networks, and others. Nowadays, the community commonly employs graphics processing units (GPUs) to accelerate such stencil computations. As a result, knowing how to write efficient stencil computations for GPUs is of significant interest. While high‐performance, low‐order stencils on GPUs have been studied extensively in the literature, not all proposed approaches work well for high‐order stencils. Furthermore, coping with boundary conditions used with stencils for seismic modeling makes it challenging to efficiently exploit thread‐level parallelism on GPUs. In this article, we describe several implementations of a 25‐point stencil. We evaluate our stencil code shapes, memory hierarchy usage, data access patterns, and other performance attributes on several modern GPUs and compare them with machine rooflines. On average, our top‐performing kernels achieve six times the performance of a 25‐point stencil code developed in C and mapped to GPUs using OpenACC. Several of our implementations have excellent performance portability across multiple generations of both NVIDIA and AMD GPUs.

Learning to predict multi-label outputs is challenging, but in many problems there is a natural metric on the outputs that can be used to improve predictions. In this paper we develop a loss function for multi-label learning, based on the Wasserstein distance. The Wasserstein distance provides a natural notion of dissimilarity for probability measures. Although optimizing with respect to the exact Wasserstein distance is costly, recent work has described a regularized approximation that is efficiently computed. We describe an efficient learning algorithm based on this regularization, as well as a novel extension of the Wasserstein distance from probability measures to unnormalized measures. We also describe a statistical learning bound for the loss. The Wasserstein loss can encourage smoothness of the predictions with respect to a chosen metric on the output space. We demonstrate this property on a real-data tag prediction problem, using the Yahoo Flickr Creative Commons dataset, outperforming a baseline that doesn't use the metric.

Finite Difference (FD) is a widely used method to solve Partial Differential Equations (PDE). PDEs are the core of many simulations in different scientific fields, such as geophysics, astrophysics, etc. The typical FD solver performs stencil computations for the entire computational domain, thus solving the differential operators. In general terms, the stencil computation consists of a weighted accumulation of the contribution of neighbor points along the cartesian axis. Therefore, optimizing stencil computations is crucial in reducing the application execution time. Stencil computation performance is bounded by two main factors: the memory access pattern and the inefficient reuse of the accessed data. We propose a novel algorithm, named Semi-stencil , that tackles these two problems. The main idea behind this algorithm is to change the way in which the stencil computation progresses within the computational domain. Instead of accessing all required neighbors and adding all their contributions at once, the Semi-stencil algorithm divides the computation into several updates. Then, each update gathers half of the axis neighbors, partially computing at the same time the stencil in a set of closely located points. As Semi-stencil progresses through the domain, the stencil computations are completed on precomputed points. This computation strategy improves the memory access pattern and efficiently reuses the accessed data. Our initial target architecture was the Cell/B.E., where the Semi-stencil in a SPE was 44% faster than the naive stencil implementation. Since then, we have continued our research on emerging multicore architectures in order to assess and extend this work on homogeneous architectures. The experiments presented combine the Semi-stencil strategy with space- and time-blocking algorithms used in hierarchical memory architectures. Two x86 (Intel Nehalem and AMD Opteron) and two POWER (IBM POWER6 and IBM BG/P) platforms are used as testbeds, where the best improvements for a 25-point stencil range from 1.27 to 1.76× faster. The results show that this novel strategy is a feasible optimization method which may be integrated into auto-tuning frameworks. Also, since all current architectures are multicore based, we have introduced a brief section where scalability results on IBM POWER7-, Intel Xeon-, and MIC-based systems are presented. In a nutshell, the algorithm scales as well as or better than other stencil techniques. For instance, the scalability of Semi-stencil on MIC for a certain testcase reached 93.8 × over 244 threads.

Oil and gas companies trust Reverse Time Migration (RTM), the most advanced seismic imaging technique, with crucial decisions on drilling investments. The economic value of the oil reserves that require RTM to be localized is in the order of 10^{13} dollars. But RTM requires vast computational power, which somewhat hindered its practical success. Although, accelerator-based architectures deliver enormous computational power, little attention has been devoted to assess the RTM implementations effort. The aim of this paper is to identify the major limitations imposed by different accelerators during RTM implementations, and potential bottlenecks regarding architecture features. Moreover, we suggest a wish list, that from our experience, should be included as features in the next generation of accelerators, to cope with the requirements of applications like RTM. We present an RTM algorithm mapping to the IBM Cell/B.E., NVIDIA Tesla and an FPGA platform modeled after the Convey HC-1. All three implementations outperform a traditional processor (Intel Harpertown) in terms of performance (10x), but at the cost of huge development effort, mainly due to immature development frameworks and lack of well-suited programming models. These results show that accelerators are well positioned platforms for this kind of workload. Due to the fact that our RTM implementation is based on an explicit high order finite difference scheme, some of the conclusions of this work can be extrapolated to applications with similar numerical scheme, for instance, magneto-hydrodynamics or atmospheric flow simulations.

3D stencil computations are compute-intensive kernels often appearing in high-performance scientific and engineering applications. The key to efficiency in these memory-bound kernels is full exploitation of data reuse. This paper explores the design aspects for 3D-Stencil implementations that maximize the reuse of all input data on a FPGA architecture. The work focuses on the architectural design of 3D stencils with the form n × (n + 1) × n, where n = {2, 4, 6, 8, ...}. The performance of the architecture is evaluated using two design approaches, ¿Multi-Volume¿ and ¿Single-Volume¿. When n = 8, the designs achieve a sustained throughput of 55.5 GFLOPS in the ¿Single-Volume¿ approach and 103 GFLOPS in the ¿Multi-Volume¿ design approach in a 100-200 MHz multi-rate implementation on a Virtex-4 LX200 FPGA. This corresponds to a stencil data delivery of 1500 bytes/cycle and 2800 bytes/cycle respectively. The implementation is analyzed and compared to two CPU cache approaches and to the statically scheduled local stores on the IBM PowerXCell 8i. The FPGA approaches designed here achieve much higher bandwidth despite the FPGA device being the least recent of the chips considered. These numbers show how a custom memory organization can provide large data throughput when implementing 3D stencil kernels.