arXiv subjects

Josep Ll. Berral

Publications and source records attributed to Josep Ll. Berral.

12 recordsLinked to original sources

arXiv2026

SLIM: Saturation-Aware Lightweight Performance Modeling for LLM Serving

Large language model (LLM) serving commonly increases batch size to improve throughput, but performance eventually reaches a deployment-dependent plateau beyond which larger batches provide marginal gains while increasing latency and GPU memory consumption. Previous studies have attributed this behavior to HBM/DRAM bandwidth limitations, but the underlying causes have primarily been supported by conceptual arguments or high-level performance observations. As our first contribution, we present a detailed GPU characterization using hardware profiling techniques, demonstrating that throughput saturation originates in the attention kernels during the decode phase. Specifically, we show that their nearly constant arithmetic intensity as active-context lengths increases -not merely larger batch sizes- drives DRAM-bandwidth saturation, while the achieved compute throughput remains far below the hardware limit. Building on this analysis, we present the Batching Configuration Advisor (BCA), which selects the highest-throughput batching configuration satisfying a target latency constraint and identifies up to 55 GB of GPU memory allocation that can be avoided for the evaluated OPT models with minimal throughput loss. To enable these recommendations, we introduce SLIM (Saturation-Aware Lightweight Performance Model), a semi-analytical model that predicts LLM inference throughput and latency from analytical formulations of Transformer computation and memory traffic. Across the evaluated scenarios, SLIM outperforms representative performance-modeling baselines while successfully generalizing to previously unseen operating conditions.

cs.DC↗

arXiv2026

Data Driven Optimization of GPU efficiency for Distributed LLM-Adapter Serving

Large Language Model (LLM) adapters enable low-cost model specialization, but introduce complex caching and scheduling challenges in distributed serving systems where hundreds of adapters must be hosted concurrently. While prior work has largely focused on latency and throughput optimization, minimizing GPU resource requirements through near-peak utilization remains largely underexplored. This paper presents a data-driven pipeline that, for a given workload, computes an adapter placement that serves the workload with the minimum number of GPUs while avoiding request starvation and GPU memory errors. To that end, the approach identifies the maximum feasible throughput attainable on each GPU by leveraging accurate performance predictions learned from real serving behavior. The proposed pipeline integrates three components: (i) a Digital Twin (DT) tailored to LLM-adapter serving, (ii) a distilled machine learning (ML) model trained on DT-generated data, and (iii) a greedy placement algorithm that exploits ML-based performance estimates to maximize GPU efficiency. The DT emulates real system dynamics with high fidelity, achieving below 5% throughput estimation error while executing up to 90x faster than full LLM benchmarking across both predictable and unpredictable workloads. The learned ML models further accelerate performance estimation with marginal accuracy degradation, enabling scalable optimization. Experimental results demonstrate that the pipeline substantially improves GPU efficiency, reducing the number of GPUs required to sustain target workloads by 60\% on average across the evaluated scenarios. Beyond GPU efficiency, the pipeline can be adapted to alternative objectives, such as latency minimization, highlighting its versatility for future large-scale LLM serving infrastructures.

cs.DC↗

arXiv2025

A Data-driven ML Approach for Maximizing Performance in LLM-Adapter Serving

With the rapid adoption of Large Language Models (LLMs), LLM-adapters have become increasingly common, providing lightweight specialization of large-scale models. Serving hundreds or thousands of these adapters on a single GPU allows request aggregation, increasing throughput, but may also cause request starvation if GPU memory limits are exceeded. To address this issue, this study focuses on determining the joint configuration of concurrent and parallel adapters that maximizes GPU throughput without inducing starvation, given heterogeneous adapter and traffic properties. We propose a data-driven ML approach leveraging interpretable models to tackle this caching problem and introduce the first Digital Twin capable of reproducing an LLM-adapter serving system, enabling efficient training data generation. Experiments with the vLLM framework and LoRA adapters show that the Digital Twin reproduces throughput within 5.1% of real results, while the ML approach predicts optimal numbers of concurrent and parallel adapters with an error of at most 7.2% under heterogeneous, real-world workloads. The code is publicly available at https://github.com/FerranAgulloLopez/GPULLMAdapterOptimization.

cs.PF↗

arXiv2025

FRIDA: Free-Rider Detection using Privacy Attacks

Federated learning is increasingly popular as it enables multiple parties with limited datasets and resources to train a machine learning model collaboratively. However, similar to other collaborative systems, federated learning is vulnerable to free-riders - participants who benefit from the global model without contributing. Free-riders compromise the integrity of the learning process and slow down the convergence of the global model, resulting in increased costs for honest participants. To address this challenge, we propose FRIDA: free-rider detection using privacy attacks. Instead of focusing on implicit effects of free-riding, FRIDA utilizes membership and property inference attacks to directly infer evidence of genuine client training. Our extensive evaluation demonstrates that FRIDA is effective across a wide range of scenarios.

cs.LG↗

arXiv2025

Towards Pareto Optimal Throughput in Small Language Model Serving

Large language models (LLMs) have revolutionized the state-of-the-art of many different natural language processing tasks. Although serving LLMs is computationally and memory demanding, the rise of Small Language Models (SLMs) offers new opportunities for resource-constrained users, who now are able to serve small models with cutting-edge performance. In this paper, we present a set of experiments designed to benchmark SLM inference at performance and energy levels. Our analysis provides a new perspective in serving, highlighting that the small memory footprint of SLMs allows for reaching the Pareto-optimal throughput within the resource capacity of a single accelerator. In this regard, we present an initial set of findings demonstrating how model replication can effectively improve resource utilization for serving SLMs.

cs.CL↗

arXiv2025

Mind the Memory Gap: Unveiling GPU Bottlenecks in Large-Batch LLM Inference

Large language models have been widely adopted across different tasks, but their auto-regressive generation nature often leads to inefficient resource utilization during inference. While batching is commonly used to increase throughput, performance gains plateau beyond a certain batch size, especially with smaller models, a phenomenon that existing literature typically explains as a shift to the compute-bound regime. In this paper, through an in-depth GPU-level analysis, we reveal that large-batch inference remains memory-bound, with most GPU compute capabilities underutilized due to DRAM bandwidth saturation as the primary bottleneck. To address this, we propose a Batching Configuration Advisor (BCA) that optimizes memory allocation, reducing GPU memory requirements with minimal impact on throughput. The freed memory and underutilized GPU compute capabilities can then be leveraged by concurrent workloads. Specifically, we use model replication to improve serving throughput and GPU utilization. Our findings challenge conventional assumptions about LLM inference, offering new insights and practical strategies for improving resource utilization, particularly for smaller language models. The code is publicly available at https://github.com/FerranAgulloLopez/vLLMBatchingMemoryGap.

cs.DC↗

arXiv2024

Enabling an OpenStack-based cloud on top of RISC-V hardware

The European Union's technological sovereignty strategy centers around the RISC-V Instruction Set Architecture, with the European Processor Initiative leading efforts to build production-ready processors. Focusing on realizing a functional RISC-V cloud ecosystem, the Vitamin-V European project developed an OpenStack cluster utilizing genuine hardware. In this poster, we detail the efforts done in porting and setting up the cluster and the many software services required by OpenStack to properly run on real hardware. In this poster, we detail our efforts on building an minimal viable prototype OpenStack cluster using real hardware. The cluster is almost functional, and we expect it to be complete in the next few months.

cs.DC↗

arXiv2023

Challenges and Opportunities for RISC-V Architectures towards Genomics-based Workloads

The use of large-scale supercomputing architectures is a hard requirement for scientific computing Big-Data applications. An example is genomics analytics, where millions of data transformations and tests per patient need to be done to find relevant clinical indicators. Therefore, to ensure open and broad access to high-performance technologies, governments, and academia are pushing toward the introduction of novel computing architectures in large-scale scientific environments. This is the case of RISC-V, an open-source and royalty-free instruction-set architecture. To evaluate such technologies, here we present the Variant-Interaction Analytics use case benchmarking suite and datasets. Through this use case, we search for possible genetic interactions using computational and statistical methods, providing a representative case for heavy ETL (Extract, Transform, Load) data processing. Current implementations are implemented in x86-based supercomputers (e.g. MareNostrum-IV at the Barcelona Supercomputing Center (BSC)), and future steps propose RISC-V as part of the next MareNostrum generations. Here we describe the Variant Interaction Use Case, highlighting the characteristics leveraging high-performance computing, indicating the caveats and challenges towards the next RISC-V developments and designs to come from a first comparison between x86 and RISC-V architectures on real Variant Interaction executions over real hardware implementations.

cs.DC↗

arXiv2021

Towards Automatic Model Specialization for Edge Video Analytics

Judging by popular and generic computer vision challenges, such as the ImageNet or PASCAL VOC, neural networks have proven to be exceptionally accurate in recognition tasks. However, state-of-the-art accuracy often comes at a high computational price, requiring hardware acceleration to achieve real-time performance, while use cases, such as smart cities, require images from fixed cameras to be analyzed in real-time. Due to the amount of network bandwidth these streams would generate, we cannot rely on offloading compute to a centralized cloud. Thus, a distributed edge cloud is expected to process images locally. However, the edge is, by nature, resource-constrained, which puts a limit on the computational complexity that can execute. Yet, there is a need for a meeting point between the edge and accurate real-time video analytics. Specializing lightweight models on a per-camera basis may help but it quickly becomes unfeasible as the number of cameras grows unless the process is automated. In this paper, we present and evaluate COVA (Contextually Optimized Video Analytics), a framework to assist in the automatic specialization of models for video analytics in edge cameras. COVA automatically improves the accuracy of lightweight models through their specialization. Moreover, we discuss and review each step involved in the process to understand the different trade-offs that each one entails. Additionally, we show how the sole assumption of static cameras allows us to make a series of considerations that greatly simplify the scope of the problem. Finally, experiments show that state-of-the-art models, i.e., able to generalize to unseen environments, can be effectively used as teachers to tailor smaller networks to a specific context, boosting accuracy at a constant computational cost. Results show that our COVA can automatically improve accuracy of pre-trained models by an average of 21%.

cs.CV↗

arXiv2020

Improving Maritime Traffic Emission Estimations on Missing Data with CRBMs

Maritime traffic emissions are a major concern to governments as they heavily impact the Air Quality in coastal cities. Ships use the Automatic Identification System (AIS) to continuously report position and speed among other features, and therefore this data is suitable to be used to estimate emissions, if it is combined with engine data. However, important ship features are often inaccurate or missing. State-of-the-art complex systems, like CALIOPE at the Barcelona Supercomputing Center, are used to model Air Quality. These systems can benefit from AIS based emission models as they are very precise in positioning the pollution. Unfortunately, these models are sensitive to missing or corrupted data, and therefore they need data curation techniques to significantly improve the estimation accuracy. In this work, we propose a methodology for treating ship data using Conditional Restricted Boltzmann Machines (CRBMs) plus machine learning methods to improve the quality of data passed to emission models. Results show that we can improve the default methods proposed to cover missing data. In our results, we observed that using our method the models boosted their accuracy to detect otherwise undetectable emissions. In particular, we used a real data-set of AIS data, provided by the Spanish Port Authority, to estimate that thanks to our method, the model was able to detect 45% of additional emissions, of additional emissions, representing 152 tonnes of pollutants per week in Barcelona and propose new features that may enhance emission modeling.

cs.CY↗

arXiv2015

ALOJA: A Framework for Benchmarking and Predictive Analytics in Big Data Deployments

This article presents the ALOJA project and its analytics tools, which leverages machine learning to interpret Big Data benchmark performance data and tuning. ALOJA is part of a long-term collaboration between BSC and Microsoft to automate the characterization of cost-effectiveness on Big Data deployments, currently focusing on Hadoop. Hadoop presents a complex run-time environment, where costs and performance depend on a large number of configuration choices. The ALOJA project has created an open, vendor-neutral repository, featuring over 40,000 Hadoop job executions and their performance details. The repository is accompanied by a test-bed and tools to deploy and evaluate the cost-effectiveness of different hardware configurations, parameters and Cloud services. Despite early success within ALOJA, a comprehensive study requires automation of modeling procedures to allow an analysis of large and resource-constrained search spaces. The predictive analytics extension, ALOJA-ML, provides an automated system allowing knowledge discovery by modeling environments from observed executions. The resulting models can forecast execution behaviors, predicting execution times for new configurations and hardware choices. That also enables model-based anomaly detection or efficient benchmark guidance by prioritizing executions. In addition, the community can benefit from ALOJA data-sets and framework to improve the design and deployment of Big Data applications.

cs.LG↗

arXiv2015

ALOJA-ML: A Framework for Automating Characterization and Knowledge Discovery in Hadoop Deployments

This article presents ALOJA-Machine Learning (ALOJA-ML) an extension to the ALOJA project that uses machine learning techniques to interpret Hadoop benchmark performance data and performance tuning; here we detail the approach, efficacy of the model and initial results. Hadoop presents a complex execution environment, where costs and performance depends on a large number of software (SW) configurations and on multiple hardware (HW) deployment choices. These results are accompanied by a test bed and tools to deploy and evaluate the cost-effectiveness of the different hardware configurations, parameter tunings, and Cloud services. Despite early success within ALOJA from expert-guided benchmarking, it became clear that a genuinely comprehensive study requires automation of modeling procedures to allow a systematic analysis of large and resource-constrained search spaces. ALOJA-ML provides such an automated system allowing knowledge discovery by modeling Hadoop executions from observed benchmarks across a broad set of configuration parameters. The resulting performance models can be used to forecast execution behavior of various workloads; they allow 'a-priori' prediction of the execution times for new configurations and HW choices and they offer a route to model-based anomaly detection. In addition, these models can guide the benchmarking exploration efficiently, by automatically prioritizing candidate future benchmark tests. Insights from ALOJA-ML's models can be used to reduce the operational time on clusters, speed-up the data acquisition and knowledge discovery process, and importantly, reduce running costs. In addition to learning from the methodology presented in this work, the community can benefit in general from ALOJA data-sets, framework, and derived insights to improve the design and deployment of Big Data applications.

cs.LG↗