arXiv subjects

Franck Cappello

Publications and source records attributed to Franck Cappello.

At least 37 records · Page 2Linked to original sources

arXiv2025

Evaluating the Safety and Skill Reasoning of Large Reasoning Models Under Compute Constraints

Test-time compute scaling has demonstrated the ability to improve the performance of reasoning language models by generating longer chain-of-thought (CoT) sequences. However, this increase in performance comes with a significant increase in computational cost. In this work, we investigate two compute constraint strategies: (1) reasoning length constraint and (2) model quantization, as methods to reduce the compute demand of reasoning models and study their impact on their safety performance. Specifically, we explore two approaches to apply compute constraints to reasoning models: (1) fine-tuning reasoning models using a length controlled policy optimization (LCPO) based reinforcement learning method to satisfy a user-defined CoT reasoning length, and (2) applying quantization to maximize the generation of CoT sequences within a user-defined compute constraint. Furthermore, we study the trade-off between the computational efficiency and the safety of the model.

cs.AI↗

arXiv2025

Automated MCQA Benchmarking at Scale: Evaluating Reasoning Traces as Retrieval Sources for Domain Adaptation of Small Language Models

As scientific knowledge grows at an unprecedented pace, evaluation benchmarks must evolve to reflect new discoveries and ensure language models are tested on current, diverse literature. We propose a scalable, modular framework for generating multiple-choice question-answering (MCQA) benchmarks directly from large corpora of scientific papers. Our pipeline automates every stage of MCQA creation, including PDF parsing, semantic chunking, question generation, and model evaluation. As a case study, we generate more than 16,000 MCQs from 22,000 open-access articles in radiation and cancer biology. We then evaluate a suite of small language models (1.1B-14B parameters) on these questions, comparing baseline accuracy with retrieval-augmented generation (RAG) from paper-derived semantic chunks and from reasoning traces distilled from GPT-4.1. We find that reasoning-trace retrieval consistently improves performance on both synthetic and expert-annotated benchmarks, enabling several small models to surpass GPT-4 on the 2023 Astro Radiation and Cancer Biology exam.

cs.CL↗

arXiv2025

MLP-Offload: Multi-Level, Multi-Path Offloading for LLM Pre-training to Break the GPU Memory Wall

Training LLMs larger than the aggregated memory of multiple GPUs is increasingly necessary due to the faster growth of LLM sizes compared to GPU memory. To this end, multi-tier host memory or disk offloading techniques are proposed by state of art. Despite advanced asynchronous multi-tier read/write strategies, such offloading strategies result in significant I/O overheads in the critical path of training, resulting in slower iterations. To this end, we propose MLP-Offload, a novel multi-level, multi-path offloading engine specifically designed for optimizing LLM training on resource-constrained setups by mitigating I/O bottlenecks. We make several key observations that drive the design of MLP-Offload, such as I/O overheads during the update dominate the iteration time; I/O bandwidth of the third-level remote storage tier remains unutilized; and, contention due to concurrent offloading amplifies I/O bottlenecks. Driven by these insights, we design and implement MLP-Offload to offload the optimizer states across multiple tiers in a cache-efficient and concurrency-controlled fashion to mitigate I/O bottlenecks during the backward and update phases. Evaluations on models up to 280B parameters shows that MLP-Offload achieves 2.5$\times$ faster iterations compared to the state-of-the-art LLM training runtimes.

cs.DC↗

arXiv2025

Boosting Scientific Error-Bounded Lossy Compression through Optimized Synergistic Lossy-Lossless Orchestration

As high-performance computing architectures evolve, more scientific computing workflows are being deployed on advanced computing platforms such as GPUs. These workflows can produce raw data at extremely high throughputs, requiring urgent high-ratio and low-latency error-bounded data compression solutions. In this paper, we propose cuSZ-Hi, an optimized high-ratio GPU-based scientific error-bounded lossy compressor with a flexible, domain-irrelevant, and fully open-source framework design. Our novel contributions are: 1) We maximally optimize the parallelized interpolation-based data prediction scheme on GPUs, enabling the full functionalities of interpolation-based scientific data prediction that are adaptive to diverse data characteristics; 2) We thoroughly explore and investigate lossless data encoding techniques, then craft and incorporate the best-fit lossless encoding pipelines for maximizing the compression ratio of cuSZ-Hi; 3) We systematically evaluate cuSZ-Hi on benchmarking datasets together with representative baselines. Compared to existing state-of-the-art scientific lossy compressors, with comparative or better throughput than existing high-ratio scientific error-bounded lossy compressors on GPUs, cuSZ-Hi can achieve up to 249% compression ratio improvement under the same error bound, and up to 215% compression ratio improvement under the same decompression data PSNR.

cs.DC↗

arXiv2025

GPZ: GPU-Accelerated Lossy Compressor for Particle Data

Particle-based simulations and point-cloud applications generate massive, irregular datasets that challenge storage, I/O, and real-time analytics. Traditional compression techniques struggle with irregular particle distributions and GPU architectural constraints, often resulting in limited throughput and suboptimal compression ratios. In this paper, we present GPZ, a high-performance, error-bounded lossy compressor designed specifically for large-scale particle data on modern GPUs. GPZ employs a novel four-stage parallel pipeline that synergistically balances high compression efficiency with the architectural demands of massively parallel hardware. We introduce a suite of targeted optimizations for computation, memory access, and GPU occupancy that enables GPZ to achieve near-hardware-limit throughput. We conduct an extensive evaluation on three distinct GPU architectures (workstation, data center, and edge) using six large-scale, real-world scientific datasets from five distinct domains. The results demonstrate that GPZ consistently and significantly outperforms five state-of-the-art GPU compressors, delivering up to 8x higher end-to-end throughput while simultaneously achieving superior compression ratios and data quality.

cs.DC↗

arXiv2025

Systematic Evaluation of Optimization Techniques for Long-Context Language Models

Large language models (LLMs) excel across diverse natural language processing tasks but face resource demands and limited context windows. Although techniques like pruning, quantization, and token dropping can mitigate these issues, their efficacy in long-context scenarios and system evaluation remains underexplored. This paper systematically benchmarks these optimizations, characterizing memory usage, latency, and throughput, and studies how these methods impact the quality of text generation. We first analyze individual optimization methods for two LLM architectures supporting long context and then systematically evaluate combinations of these techniques to assess how this deeper analysis impacts performance metrics. We subsequently study the scalability of individual optimization methods on a larger variant with 70 billion-parameter model. Our novel insights reveal that naive combination inference optimization algorithms can adversely affect larger models due to compounded approximation errors, as compared to their smaller counterparts. Experiments show that relying solely on F1 obscures these effects by hiding precision-recall trade-offs in question answering tasks. By integrating system-level profiling with task-specific insights, this study helps LLM practitioners and researchers explore and balance efficiency, accuracy, and scalability across tasks and hardware configurations.

cs.CL↗

arXiv2025

QPET: A Versatile and Portable Quantity-of-Interest-Preservation Framework for Error-Bounded Lossy Compression

Error-bounded lossy compression has been widely adopted in many scientific domains because it can address the challenges in storing, transferring, and analyzing unprecedented amounts of scientific data. Although error-bounded lossy compression offers general data distortion control by enforcing strict error bounds on raw data, it may fail to meet the quality requirements on the results of downstream analysis, a.k.a. Quantities of Interest (QoIs), derived from raw data. This may lead to uncertainties and even misinterpretations in scientific discoveries, significantly limiting the use of lossy compression in practice. In this paper, we propose QPET, a novel, versatile, and portable framework for QoI-preserving error-bounded lossy compression, which overcomes the challenges of modeling diverse QoIs by leveraging numerical strategies. QPET features (1) high portability to multiple existing lossy compressors, (2) versatile preservation to most differentiable univariate and multivariate QoIs, and (3) significant compression improvements in QoI-preservation tasks. Experiments with six real-world datasets demonstrate that integrating QPET into state-of-the-art error-bounded lossy compressors can gain 2x to 10x compression speedups of existing QoI-preserving error-bounded lossy compression solutions, up to 1000% compression ratio improvements to general-purpose compressors, and up to 133% compression ratio improvements to existing QoI-integrated scientific compressors.

cs.DB↗

arXiv2025

To Compress or Not To Compress: Energy Trade-Offs and Benefits of Lossy Compressed I/O

Modern scientific simulations generate massive volumes of data, creating significant challenges for I/O and storage systems. Error-bounded lossy compression (EBLC) offers a solution by reducing data set sizes while preserving data quality within user-specified limits. This study provides the first comprehensive energy characterization of state-of-the-art EBLC algorithms--SZ2, SZ3, ZFP, QoZ, and SZx--across various scientific data sets, CPU generations, and parallel/serial modes. We analyze the energy consumption patterns of compression and decompression operations, as well as the energy trade-offs in data I/O scenarios. Our work demonstrates the relationships between compression ratios, runtime, energy efficiency, and data quality, highlighting the importance of considering compressors and error bounds for specific use cases. We demonstrate that EBLC can significantly reduce I/O energy consumption, with savings of up to two orders of magnitude compared to uncompressed I/O for large data sets. In multi-node HPC environments, we observe energy reductions of approximately 25% when using EBLC. We also show that EBLC can achieve compression ratios of 10-100x, potentially reducing storage device requirements by nearly two orders of magnitude. This work provides a framework for system operators and computational scientists to make informed decisions about implementing EBLC for energy-efficient data management in HPC environments.

cs.DC↗

arXiv2025

A Survey on Error-Bounded Lossy Compression for Scientific Datasets

Error-bounded lossy compression has been effective in significantly reducing the data storage/transfer burden while preserving the reconstructed data fidelity very well. Many error-bounded lossy compressors have been developed for a wide range of parallel and distributed use cases for years. They are designed with distinct compression models and principles, such that each of them features particular pros and cons. In this paper we provide a comprehensive survey of emerging error-bounded lossy compression techniques. The key contribution is fourfold. (1) We summarize a novel taxonomy of lossy compression into 6 classic models. (2) We provide a comprehensive survey of 10 commonly used compression components/modules. (3) We summarized pros and cons of 46 state-of-the-art lossy compressors and present how state-of-the-art compressors are designed based on different compression techniques. (4) We discuss how customized compressors are designed for specific scientific applications and use-cases. We believe this survey is useful to multiple communities including scientific applications, high-performance computing, lossy compression, and big data.

cs.DC↗

arXiv2025

IPComp: Interpolation Based Progressive Lossy Compression for Scientific Applications

Compression is a crucial solution for data reduction in modern scientific applications due to the exponential growth of data from simulations, experiments, and observations. Compression with progressive retrieval capability allows users to access coarse approximations of data quickly and then incrementally refine these approximations to higher fidelity. Existing progressive compression solutions suffer from low reduction ratios or high operation costs, effectively undermining the approach's benefits. In this paper, we propose the first-ever interpolation-based progressive lossy compression solution that has both high reduction ratios and low operation costs. The interpolation-based algorithm has been verified as one of the best for scientific data reduction, but previously no effort exists to make it support progressive retrieval. Our contributions are three-fold: (1) We thoroughly analyze the error characteristics of the interpolation algorithm and propose our solution IPComp with multi-level bitplane and predictive coding. (2) We derive optimized strategies toward minimum data retrieval under different fidelity levels indicated by users through error bounds and bitrates. (3) We evaluate the proposed solution using six real-world datasets from four diverse domains. Experimental results demonstrate our solution archives up to $487\%$ higher compression ratios and $698\%$ faster speed than other state-of-the-art progressive compressors, and reduces the data volume for retrieval by up to $83\%$ compared to baselines under the same error bound, and reduces the error by up to $99\%$ under the same bitrate.

cs.DC↗

arXiv2025

Lossy Compression of Scientific Data: Applications Constrains and Requirements

Increasing data volumes from scientific simulations and instruments (supercomputers, accelerators, telescopes) often exceed network, storage, and analysis capabilities. The scientific community's response to this challenge is scientific data reduction. Reduction can take many forms, such as triggering, sampling, filtering, quantization, and dimensionality reduction. This report focuses on a specific technique: lossy compression. Lossy compression retains all data points, leveraging correlations and controlled reduced accuracy. Quality constraints, especially for quantities of interest, are crucial for preserving scientific discoveries. User requirements also include compression ratio and speed. While many papers have been published on lossy compression techniques and reference datasets are shared by the community, there is a lack of detailed specifications of application needs that can guide lossy compression researchers and developers. This report fills this gap by reporting on the requirements and constraints of nine scientific applications covering a large spectrum of domains (climate, combustion, cosmology, fusion, light sources, molecular dynamics, quantum circuit simulation, seismology, and system logs). The report also details key lossy compression technologies (SZ, ZFP, MGARD, LC, SPERR, DCTZ, TEZip, LibPressio), discussing their history, principles, error control, hardware support, features, and impact. By presenting both application needs and compression technologies, the report aims to inspire new research to fill existing gaps.

astro-ph.IM↗

arXiv2025

EAIRA: Establishing a Methodology for Evaluating AI Models as Scientific Research Assistants

Recent advancements have positioned AI, and particularly Large Language Models (LLMs), as transformative tools for scientific research, capable of addressing complex tasks that require reasoning, problem-solving, and decision-making. Their exceptional capabilities suggest their potential as scientific research assistants but also highlight the need for holistic, rigorous, and domain-specific evaluation to assess effectiveness in real-world scientific applications. This paper describes a multifaceted methodology for Evaluating AI models as scientific Research Assistants (EAIRA) developed at Argonne National Laboratory. This methodology incorporates four primary classes of evaluations. 1) Multiple Choice Questions to assess factual recall; 2) Open Response to evaluate advanced reasoning and problem-solving skills; 3) Lab-Style Experiments involving detailed analysis of capabilities as research assistants in controlled environments; and 4) Field-Style Experiments to capture researcher-LLM interactions at scale in a wide range of scientific domains and applications. These complementary methods enable a comprehensive analysis of LLM strengths and weaknesses with respect to their scientific knowledge, reasoning abilities, and adaptability. Recognizing the rapid pace of LLM advancements, we designed the methodology to evolve and adapt so as to ensure its continued relevance and applicability. This paper describes the methodology state at the end of February 2025. Although developed within a subset of scientific domains, the methodology is designed to be generalizable to a wide range of scientific domains.

cs.AI↗

arXiv2025

ZCCL: Significantly Improving Collective Communication With Error-Bounded Lossy Compression

With the ever-increasing computing power of supercomputers and the growing scale of scientific applications, the efficiency of MPI collective communication turns out to be a critical bottleneck in large-scale distributed and parallel processing. The large message size in MPI collectives is particularly concerning because it can significantly degrade overall parallel performance. To address this issue, prior research simply applies off-the-shelf fixed-rate lossy compressors in the MPI collectives, leading to suboptimal performance, limited generalizability, and unbounded errors. In this paper, we propose a novel solution, called ZCCL, which leverages error-bounded lossy compression to significantly reduce the message size, resulting in a substantial reduction in communication costs. The key contributions are three-fold. (1) We develop two general, optimized lossy-compression-based frameworks for both types of MPI collectives (collective data movement as well as collective computation), based on their particular characteristics. Our framework not only reduces communication costs but also preserves data accuracy. (2) We customize fZ-light, an ultra-fast error-bounded lossy compressor, to meet the specific needs of collective communication. (3) We integrate ZCCL into multiple collectives, such as Allgather, Allreduce, Scatter, and Broadcast, and perform a comprehensive evaluation based on real-world scientific application datasets. Experiments show that our solution outperforms the original MPI collectives as well as multiple baselines by 1.9--8.9X.

cs.DC↗

arXiv2024

TurboFFT: Co-Designed High-Performance and Fault-Tolerant Fast Fourier Transform on GPUs

GPU-based fast Fourier transform (FFT) is extremely important for scientific computing and signal processing. However, we find the inefficiency of existing FFT libraries and the absence of fault tolerance against soft error. To address these issues, we introduce TurboFFT, a new FFT prototype co-designed for high performance and online fault tolerance. For FFT, we propose an architecture-aware, padding-free, and template-based prototype to maximize hardware resource utilization, achieving a competitive or superior performance compared to the state-of-the-art closed-source library, cuFFT. For fault tolerance, we 1) explore algorithm-based fault tolerance (ABFT) at the thread and threadblock levels to reduce additional memory footprint, 2) address the error propagation by introducing a two-side ABFT with location encoding, and 3) further modify the threadblock-level FFT from 1-transaction to multi-transaction in order to bring more parallelism for ABFT. Our two-side strategy enables online correction without additional global memory while our multi-transaction design averages the expensive threadblock-level reduction in ABFT with zero additional operations. Experimental results on an NVIDIA A100 server GPU and a Tesla Turing T4 GPU demonstrate that TurboFFT without fault tolerance is comparable to or up to 300\% faster than cuFFT and outperforms VkFFT. TurboFFT with fault tolerance maintains an overhead of 7\% to 15\%, even under tens of error injections per minute for both FP32 and FP64.

cs.DC↗

arXiv2024

DGRO: Diameter-Guided Ring Optimization for Integrated Research Infrastructure Membership

Logical ring is a core component in membership protocol. However, the logic ring fails to consider the underlying physical latency, resulting in a high diameter. To address this issue, we introduce Diameter-Guided Ring Optimization (DGRO), which focuses on constructing rings with the smallest possible diameter, selecting the most effective ring configurations, and implementing these configurations in parallel. We first explore an integration of deep Q-learning and graph embedding to optimize the ring topology. We next propose a ring selection strategy that assesses the current topology's average latency against a global benchmark, facilitating integration into modern peer-to-peer protocols and substantially reducing network diameter. To further enhance scalability, we propose a parallel strategy that distributes the topology construction process into separate partitions simultaneously. Our experiment shows that: 1) DGRO efficiently constructs a network topology that achieves up to a 60% reduction in diameter compared to the best results from an extensive search over $10^5$ topologies, all within a significantly shorter computation time, 2) the ring selection of DGRO reduces the diameter of state-of-the-art methods Chord, RAPID, and Perigee by 10%-40%, 44%, and 60%. 3) the parallel construction can scale up to $32$ partitions while maintaining the same diameter compared to the centralized version.

cs.DC↗

arXiv2024

cuSZ-$i$: High-Ratio Scientific Lossy Compression on GPUs with Optimized Multi-Level Interpolation

Error-bounded lossy compression is a critical technique for significantly reducing scientific data volumes. Compared to CPU-based compressors, GPU-based compressors exhibit substantially higher throughputs, fitting better for today's HPC applications. However, the critical limitations of existing GPU-based compressors are their low compression ratios and qualities, severely restricting their applicability. To overcome these, we introduce a new GPU-based error-bounded scientific lossy compressor named cuSZ-$i$, with the following contributions: (1) A novel GPU-optimized interpolation-based prediction method significantly improves the compression ratio and decompression data quality. (2) The Huffman encoding module in cuSZ-$i$ is optimized for better efficiency. (3) cuSZ-$i$ is the first to integrate the NVIDIA Bitcomp-lossless as an additional compression-ratio-enhancing module. Evaluations show that cuSZ-$i$ significantly outperforms other latest GPU-based lossy compressors in compression ratio under the same error bound (hence, the desired quality), showcasing a 476% advantage over the second-best. This leads to cuSZ-$i$'s optimized performance in several real-world use cases.

cs.DC↗

arXiv2024

High-performance Effective Scientific Error-bounded Lossy Compression with Auto-tuned Multi-component Interpolation

Error-bounded lossy compression has been identified as a promising solution for significantly reducing scientific data volumes upon users' requirements on data distortion. For the existing scientific error-bounded lossy compressors, some of them (such as SPERR and FAZ) can reach fairly high compression ratios and some others (such as SZx, SZ, and ZFP) feature high compression speeds, but they rarely exhibit both high ratio and high speed meanwhile. In this paper, we propose HPEZ with newly-designed interpolations and quality-metric-driven auto-tuning, which features significantly improved compression quality upon the existing high-performance compressors, meanwhile being exceedingly faster than high-ratio compressors. The key contributions lie in the following points: (1) We develop a series of advanced techniques such as interpolation re-ordering, multi-dimensional interpolation, and natural cubic splines to significantly improve compression qualities with interpolation-based data prediction. (2) The auto-tuning module in HPEZ has been carefully designed with novel strategies, including but not limited to block-wise interpolation tuning, dynamic dimension freezing, and Lorenzo tuning. (3) We thoroughly evaluate HPEZ compared with many other compressors on six real-world scientific datasets. Experiments show that HPEZ outperforms other high-performance error-bounded lossy compressors in compression ratio by up to 140% under the same error bound, and by up to 360% under the same PSNR. In parallel data transfer experiments on the distributed database, HPEZ achieves a significant performance gain with up to 40% time cost reduction over the second-best compressor.

cs.DC↗

arXiv2024

FRSZ2 for In-Register Block Compression Inside GMRES on GPUs

The performance of the GMRES iterative solver on GPUs is limited by the GPU main memory bandwidth. Compressed Basis GMRES outperforms GMRES by storing the Krylov basis in low precision, thereby reducing the memory access. An open question is whether compression techniques that are more sophisticated than casting to low precision can enable large runtime savings while preserving the accuracy of the final results. This paper presents the lightweight in-register compressor FRSZ2 that can decompress at the bandwidth speed of a modern NVIDIA H100 GPU. In an experimental evaluation, we demonstrate using FRSZ2 instead of low precision for compression of the Krylov basis can bring larger runtime benefits without impacting final accuracy.

cs.PF↗