Olcoz Herrero, Katzalin

Profile Picture
First Name
Last Name
Olcoz Herrero
Universidad Complutense de Madrid
Faculty / Institute
Ciencias Físicas
Arquitectura de Computadores y Automática
Arquitectura y Tecnología de Computadores
UCM identifierORCIDScopus Author IDDialnet IDGoogle Scholar ID

Search Results

Now showing 1 - 7 of 7
  • Publication
    Resource management for power-constrained HEVC transcoding using reinforcement learning
    (IEEE Computer Society, 2020-12-01) Costero Valero, Luis María; Iranfar, Arman; Zapater, Marina; Atienza, David; Olcoz Herrero, Katzalin
    The advent of online video streaming applications and services along with the users' demand for high-quality contents require High Efficiency Video Coding (HEVC), which provides higher video quality and more compression at the cost of increased complexity. On one hand, HEVC exposes a set of dynamically tunable parameters to provide trade-offs among Quality-of-Service (QoS), performance, and power consumption of multi-core servers on the video providers' data center. On the other hand, resource management of modern multi-core servers is in charge of adapting system-level parameters, such as operating frequency and multithreading, to deal with concurrent applications and their requirements. Therefore, efficient multi-user HEVC streaming necessitates joint adaptation of application- and system-level parameters. Nonetheless, dealing with such a large and dynamic design space is challenging and difficult to address through conventional resource management strategies. Thus, in this work, we develop a multi-agent Reinforcement Learning framework to jointly adjust application- and system-level parameters at runtime to satisfy the QoS of multi-user HEVC streaming in power-constrained servers. In particular, the design space, composed of all design parameters, is split into smaller independent sub-spaces. Each design sub-space is assigned to a particular agent so that it can explore it faster, yet accurately. The benefits of our approach are revealed in terms of adaptability and quality (with up to to 4x improvements in terms of QoS when compared to a static resource management scheme), and learning time (6 x faster than an equivalent mono-agent implementation). Finally, we show that the power-capping techniques formulated outperform the hardware-based power capping with respect to quality.
  • Publication
    Applying game-learning environments to power capping scenarios via reinforcement learning
    (Springer international Publishing, 2022-08-05) Hernández Aguado, Pablo; Costero Valero, Luis María; Olcoz Herrero, Katzalin; Igual Peña, Francisco Daniel
    Research in deep learning for video game playing has received much attention and provided very relevant results in the last years. Frameworks and libraries have been developed to ease game playing research leveraging Reinforcement Learning techniques. In this paper, we propose to use two of them (RLLIB and GYM) in a very different scenario, such as learning to apply resource management policies in a multi-core server, specifically, we leverage the facilities of both frameworks coupled to derive policies for power-capping. Using RLlib and Gym enables implementing different resource management policies in a simple and fast way and, as they are based on neural networks, guarantees the efficiency in the solution, and the use of hardware accelerators for both training and inference. The results demonstrate that game-learning environments provide an effective support to cast a completely different scenario, and open new research avenues in the field of resource management using reinforcement learning techniques with minimal development effort.
  • Publication
    Revisiting Conventional Task Schedulers to Exploit Asymmetry in ARM big.LITTLE Architectures for Dense Linear Algebra
    (Elsevier, 2017-06-01) Costero Valero, Luis María; Igual Peña, Francisco Daniel; Olcoz Herrero, Katzalin; Catalán Pallarés, Sandra; Rodríguez Sánchez, Rafael; Quintana-Ortí, Enrique S.
    Dealing with asymmetry in the architecture opens a plethora of questions related with the performance- and energy-efficient scheduling of task-parallel applications. While there exist early attempts to tackle this problem, for example via ad-hoc strategies embedded in a runtime framework, in this paper we take a different path, which consists in addressing the asymmetry at the library-level by developing a few asymmetry-aware fundamental kernels. The appealing consequence is that the architecture heterogeneity remains then hidden from the task scheduler. In order to illustrate the advantage of our approach, we employ two well-known matrix factorizations, key to the solution of dense linear systems of equations. From the perspective of the architecture, we consider two low-power processors, one of them equipped with ARM big.LITTLE technology; furthermore, we include in the study a different scenario, in which the asymmetry arises when the cores of an Intel Xeon server operate at two distinct frequencies. For the specific domain of dense linear algebra, we show that dealing with asymmetry at the library-level is not only possible but delivers higher performance than a naive approach based on an asymmetry-oblivious scheduler. Furthermore, this solution is also competitive in terms of performance compared with an ad-hoc asymmetry-aware scheduler furnished with sophisticated scheduling techniques.
  • Publication
    MAMUT: Multi-Agent Reinforcement Learning for Efficient Real-Time Multi-User Video Transcoding
    (2019) Costero Valero, Luis María; Iranfar, Arman; Zapater Sancho, Marina; Igual Peña, Francisco Daniel; Olcoz Herrero, Katzalin; Atienza Alonso, David
    Real-time video transcoding has recently raised as a valid alternative to address the ever-increasing demand for video contents in servers’ infrastructures in current multi-user environments. High Efficiency Video Coding (HEVC) makes efficient online transcoding feasible as it enhances user experience by providing the adequate video configuration, reduces pressure on the network, and minimizes inefficient and costly video storage. However, the computational complexity of HEVC, together with its myriad of configuration parameters, raises challenges for power management, throughput control, and Quality of Service (QoS) satisfaction. This is particularly challenging in multi-user environments where multiple users with different resolution demands and bandwidth constraints need to be served simultaneously. In this work, we present MAMUT, a multiagent machine learning approach to tackle these challenges. Our proposal breaks the design space composed of run-time adaptation of the transcoder and system parameters into smaller sub-spaces that can be explored in a reasonable time by individual agents. While working cooperatively, each agent is in charge of learning and applying the optimal values for internal HEVC and system-wide parameters. In particular, MAMUT dynamically tunes Quantization Parameter, selects number of threads per video, and sets the operating frequency with throughput and video quality objectives under compression and power consumption constraints. We implement MAMUT on an enterprise multicore server and compare equivalent scenarios to state-ofthe-art alternative approaches. The obtained results reveal that MAMUT consistently attains up to 8x improvement in terms of FPS violations (and thus Quality of Service), 24% power reduction, as well as faster and more accurate adaptation both to the video contents and available resources.
  • Publication
    Leveraging knowledge-as-a-service (KaaS) for QoS-aware resource management in multi-user video transcoding
    (Springer, 2020-02-25) Costero Valero, Luis María; Igual Peña, Francisco Daniel; Olcoz Herrero, Katzalin; Tirado Fernández, José Francisco
    The coexistence of parallel applications in shared computing nodes, each one featuring different Quality of Service (QoS) requirements, carries out new challenges to improve resource occupation while keeping acceptable rates in terms of QoS. As more application-specific and system-wide metrics are included as QoS dimensions, or under situations in which resource-usage limits are strict, building and serving the most appropriate set of actions (application control knobs and system resource assignment) to concurrent applications in an automatic and optimal fashion become mandatory. In this paper, we propose strategies to build and serve this type of knowledge to concurrent applications by leveraging Reinforcement Learning techniques. Taking multi-user video transcoding as a driving example, our experimental results reveal an excellent adaptation of resource and knob management to heterogeneous QoS requests, and increases in the amount of concurrently served users up to 1.24× compared with alternative approaches considering homogeneous QoS requests.
  • Publication
    Energy efficiency optimization of task-parallel codes on asymmetric architectures
    (2017-07-17) Francisco D. Igual; Costero Valero, Luis María; Igual Peña, Francisco Daniel; Olcoz Herrero, Katzalin; Tirado Fernández, José Francisco
    We present a family of policies that, integrated within a runtime task scheduler (Nanox), pursue the goal of improving the energy efficiency of task-parallel executions with no intervention from the programmer. The proposed policies tackle the problem by modifying the core operating frequency via DVFS mechanisms, or by enabling/disabling the mapping of tasks to specific cores at selected execution points, depending on the internal status of the scheduler. Experimental results on an asymmetric SoC (Exynos 5422) and for a specific operation (Cholesky factorization) reveal gains up to 29% in terms of energy efficiency and considerable reductions in average power
  • Publication
    Dynamic power budget redistribution under a power cap on multi-application environments
    (Elsevier, 2023-03-20) Costero Valero, Luis María; Igual Peña, Francisco Daniel; Olcoz Herrero, Katzalin
    We present a two-level implementation of an infrastructure that allows performance maximization under a power-cap on multi-application environments with minimal user intervention. At the application level, we integrate bar (Power Budget-Aware Runtime Scheduler) into existing task-based runtimes, e.g. OpenMP; bar implements combined software/hardware techniques (thread malleability and DVFS) to maximize the application performance without violating a granted power budget. At a higher level, we introduce barman (Power Budget-Aware Resource Manager), a system-wide software able to manage resources globally, gathering power needs of registered applications, and redistributing the available overall power budget across them. The combination and co-operative operation of both pieces of software yields performance and energy efficiency improvements on environments in which power capping is established globally, and also granted asymmetrically to different co-existing applications. This behaviour is demonstrated to be stable under different workloads (a selection of task-based scientific applications and PARSEC benchmarks are tested) and different levels of power capping.