Accelerating Particle-in-Cell Monte Carlo Simulations with MPI, OpenMP/OpenACC and Asynchronous Multi-GPU Programming

Jeremy J. Williams; Felix Liu; Jordy Trilaksono; David Tskhakaya; Stefan Costea; Leon Kos; Ales Podolnik; Jakub Hromadka; Pratibha Hegde; Marta Garcia-Gasulla; Valentin Seitz; Frank Jenko; Erwin Laure; Stefano Markidis

doi:10.1016/j.jocs.2025.102590

arXiv · 2404.10270

Accelerating Particle-in-Cell Monte Carlo Simulations with MPI, OpenMP/OpenACC and Asynchronous Multi-GPU Programming

Jeremy J. Williams·Felix Liu·Jordy Trilaksono·David Tskhakaya·Stefan Costea·Leon Kos·Ales Podolnik·Jakub Hromadka·Pratibha Hegde·Marta Garcia-Gasulla·Valentin Seitz·Frank Jenko·Erwin Laure·Stefano Markidis

View original source ↗Read source document ↗Export metadata ↓

Abstract

As fusion energy devices advance, plasma simulations are crucial for reactor design. Our work extends BIT1 hybrid parallelization by integrating MPI with OpenMP and OpenACC, focusing on asynchronous multi-GPU programming. Results show significant performance gains: 16 MPI ranks plus OpenMP threads reduced runtime by 53% on a petascale EuroHPC supercomputer, while OpenACC multicore achieved a 58% reduction. At 64 MPI ranks, OpenACC outperformed OpenMP, improving the particle mover function by 24%. On MareNostrum 5, OpenACC async(n) delivered strong performance, but OpenMP asynchronous multi-GPU approach proved more effective at extreme scaling, maintaining efficiency up to 400 GPUs. Speedup and parallel efficiency (PE) studies revealed OpenMP asynchronous multi-GPU achieving 8.77x speedup (54.81% PE), surpassing OpenACC (8.14x speedup, 50.87% PE). While PE declined at high node counts due to communication overhead, asynchronous execution mitigated scalability bottlenecks. OpenMP nowait and depend clauses improved GPU performance via efficient data transfer and task management. Using NVIDIA Nsight tools, we confirmed BIT1 efficiency for large-scale plasma simulations. OpenMP asynchronous multi-GPU implementation delivered exceptional performance in portability, high throughput, and GPU utilization, positioning BIT1 for exascale supercomputing and advancing fusion energy research. MareNostrum 5 brings us closer to achieving exascale performance.

Explore related subjects

cs.DC cs.PF physics.comp-ph

Keep this discovery

BibTeX RIS

Jeremy J. Williams, Felix Liu, Jordy Trilaksono, David Tskhakaya, Stefan Costea, Leon Kos, Ales Podolnik, Jakub Hromadka, Pratibha Hegde, Marta Garcia-Gasulla, Valentin Seitz, Frank Jenko, Erwin Laure, Stefano Markidis. 2024-04-16. Accelerating Particle-in-Cell Monte Carlo Simulations with MPI, OpenMP/OpenACC and Asynchronous Multi-GPU Programming. https://doi.org/10.1016/j.jocs.2025.102590

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Online Treasure Hunt in Vertex-Permuted Dynamic Rings

We study the problem of treasure hunt by a group of $k \geq 1$ agents in vertex-permuted dynamic rings (VP). In this model, the $n$ vertices remain on a ring but are permuted at each time step. We first show that treasure hunt is impossible for any $k \leq n-3$ agents, if there are no restrictions on the sequence of permutations used in the dynamic ring. We then study the $VP(\delta)$ setting, in which for every pair $i, j$ of vertices, the edge $(i, j)$ is guaranteed to appear within $\delta$ steps. We show that the class $VP(\delta)$ is feasible only for $\delta \geq \left\lceil \frac{n-1}{2}\right\rceil$. For the one-agent case, we show a tight bound of $\Theta(\delta n)$ on the worst-case search time as well as competitive ratio of any online algorithm for treasure hunt, provided $\delta \geq 2n$. We then give an optimal algorithm for $k$ agents, thereby showing that $k$ agents can obtain a speedup of $k$ on the worst-case search time. Finally, in the R-VP setting, in which in every step, the vertices are arranged as a ring according to a random permutation, we show that treasure hunt takes expected $\Theta(n)$ steps against an oblivious adversary and $\Theta(n \log n)$ steps against an adaptive adversary.

cs.DC↗

arXiv2026

The Computing Channel: How Modulation Programs the Airwaves

Distributed computing and distributed artificial intelligence require frequent exchanges of intermediate results, although many applications need only an aggregate rather than messages from individual devices. Conventional systems recover each message before computing the aggregate, whereas over-the-air computation (OAC) exploits simultaneous transmission to obtain it directly. However, dominant OAC implementations rely on analog signaling, creating a mismatch with finite-precision data and digital communication procedures. This article presents digital function-oriented communication, in which finite-alphabet symbol representations and receiver decisions are jointly designed so that multiple-access superposition encodes the desired function without recovering individual inputs. We introduce its computational-constellation principle, main design approaches, extensions, and implementation challenges. Federated edge learning illustrates how the framework can reduce user-dependent data-bearing resources while operating directly on quantized model updates.

cs.DC↗

arXiv2026

Can AI Remediate Backend Failures Safely? GuardedAct with Blast-Radius-Aware Sandboxing

Large Language Models (LLMs) have shown promising capabilities in generating remediation actions for microservice failures. However, directly executing AI-generated repair actions in production risks cascading collateral damage. We propose GuardedAct, a sandbox-first remediation framework that interposes a blast-radius-aware verification layer between the LLM action generator and the production environment. GuardedAct operates in four phases: (1) ingesting a diagnosis report together with the live system topology and recent telemetry, (2) prompting an LLM to produce a ranked list of candidate remediation actions, (3) simulating each action in a lightweight digital-twin sandbox that estimates the blast radius and assigns a risk label, and (4) enforcing a rollback-confidence gate that auto-executes only low-risk actions while escalating high-risk ones for human review. We evaluate GuardedAct on five fault scenarios injected into the DeathStarBench social-network application. Experimental results show that GuardedAct achieves an overall recovery rate of 87.4% while reducing collateral damage by 79.7% relative to direct LLM execution (from 25.6% to 5.2%), at the cost of a modest sandbox-induced increase in mean time to recovery (approximately 8 s). Ablation studies confirm that each component contributes meaningfully to the safety-speed trade-off.

cs.DC↗