Pasar al contenido principal

TECNOLOGIAS INNOVADORAS DE PROCESADORES, ACELERADORES Y REDES, PARA CENTROS DE DATOS Y COMPUTACION DE ALTAS PRESTACIONES

RTI2018-098156-B-C51

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal de I+D+i Orientado a los Retos de la Sociedad
Convocatoria Proyectos I+D+i Retos de Investigación
Año convocatoria 2018
Plan Estatal 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSITAT POLITÈCNICA DE VALÈNCIA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 23
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Cache-Poll: Containing Pollution in Non-Inclusive Caches Through Cache Partitioning

  • Pons-Escat, Lucía|||0000-0002-4582-7744
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Pons Terol, Julio|||0000-0002-5654-6753
[EN] Current server processors have redistributed the cache hierarchy space over previous generations. The private L2 cache has been made larger and the shared last level caches (LLC) smaller but designed as non-inclusive to reduce the number of replicated blocks. As a result, the new organization shrinks the per-core cache area. Cache management in this organization becomes more critical than in inclusive caches due to two main reasons: there is less storage capacity per core both in the L3 and when considering the sum of L2 and L3 cache sizes, and there is higher L2-L3 traffic especially when running high cache-demanding applications. This paper focuses on minimizing L3 cache pollution to make a more efficient use of the limited space. Three main types of pollution are identified and measured: useless prefetches, bad speculated loads, and poor locality. This paper proposes Cache-Poll, a pollution-aware management policy that concentrates on limiting the cache space to polluting and L3 insensitive applications, allowing critical applications occupy more space. Unlike state-of-the-art work on non-inclusive caches, Cache-Poll is able to improve performance in an Intel Xeon Scalable processor even when running heavy cache-demanding workloads, consisting of 12-application workloads, as many applications as cores in the processor. Results show that Cache-Poll improves fairness and turnaround time by 44% and 24%, respectively, over the Linux OS, while even improving performance up to 3.5%., This work has been supported by the Spanish Ministerio de Universidades under the grant FPU18/01948 and by Spanish Ministerio de Universidades and European ERDF under grant RTI2018-098156-BC51.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

An efficient cache flat storage organization for multithreaded workloads for low power processors

  • Puche, José
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[EN] The cache hierarchy of current multicores typically consists of three levels, ranging from the faster and smaller L1 level to the slower and larger L3 level. This approach has been demonstrated to be effective in high performance processors, since it reduces the average memory access time. However, when implemented in devices where energy efficiency becomes critical, like low power or embedded processors, conventional cache hierarchies may present some concerns. These concerns, which incur a waste of area and energy, are multiple cache lookups, block replication, block migration and private cache space overprovisioning. To deal with these issues, in this work we propose FOS-Mt, a new cache organization aimed at addressing energy savings in current multicores for multithreaded applications. FOS-Mt's cache hierarchy consists of only two levels: the L1 cache level located in the core pipeline, and a single and flattened second level which conforms an aggregated cache space which is accessible by all the execution cores. This level is sliced into multiple small buffers, which are dynamically assigned to any of the running thread when they are expected to improve the system performance. Those buffers that are not allocated to any core are powered off to save energy. Experimental results show that FOS-Mt significantly reduces both static and dynamic energy consumption over other conventional cache organizations like NUCA or shared caches with the same storage capacity. Compared to the widely known cache decay approach, FOS-Mt achieves an improvement in the energy delay product by 19.3% on average. Moreover, despite the fact that FOS-Mt is an energy-aware architecture, performance is scarcely affected, since it is kept similar to that one achieved by conventional and cache decay approaches., This work has been supported by the Spanish Ministerio de Economia y Competitividad under grant RTI2018-098156-B-C51, and by the Generalitat Valenciana, Spain under grant AICO/2019/317.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

VMT: Virtualized Multi-Threading for Accelerating Graph Workloads on Commodity Processors

  • Feliu-Pérez, Josué|||0000-0003-3017-4266
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Naithani, Ajeya
  • Qureshi, Moinuddin
  • Eeckhout, Lieven
[EN] Modern-day graph workloads operate on huge graphs through pointer chasing which leads to high last-level cache (LLC) miss rates and limited memory-level parallelism (MLP). Simultaneous Multi-Threading (SMT) effectively hides the memory access latencies for multi-threaded graph workloads provided that sufficient threads are supported in hardware. Unfortunately, providing a sufficiently large number of physical threads incurs an unjustifiably high hardware cost for commodity SMT processors which typically implement only two physical hardware threads. Ideally, we would like to achieve aggressive-SMT performance when running graph workloads on modest commodity processors. In this paper, we propose Virtualized Multi-Threading (VMT), a low-overhead multi-threading paradigm for accelerating graph workloads on commodity processors. Unlike prior multi-threading paradigms, VMT virtualizes both the physical hardware threads and the architecture state: VMT maps a large number of logical software threads to a small number of physical hardware threads, while maintaining the architecture state of the logical threads in the processor's cache hierarchy. Implemented on top of a quad-core 2-way SMT processor, VMT achieves an average speedup of 1.74x for a set of representative graph workloads, while incurring minimal hardware cost (195 bytes per core to support up to 32 logical threads). VMT's low hardware cost paves the way for implementation in commodity processors., This work was supported in part by the Spanish MCIU and AEI, Spain, as well as European Commission FEDER funds, under grants RTI2018-098156-B-C53 and RTI2018-098156-BC51. The work of Josue Feliu was supported by a Juan de la Cierva Formacion Contract under Grant FJC2018-036021-I. The work of Lieven Eeckhout was supported in part by the European Research Council Advanced Grant agreement no. 741097 and in part by the Flanders Research Council under Grant G.0144.17N.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Diseño y uso de redes neuronales para estimar las prestaciones en solitario de las aplicaciones

  • Lurbe, Manel
  • Feliu-Pérez, Josué|||0000-0003-3017-4266
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[ES] Cuando varias aplicaciones se ejecutan concurrentemente en un computador con recursos compartidos, como las caches de último nivel o la memoria principal, se producen pérdidas de prestaciones por la interferencia entre aplicaciones en el uso de estos recursos respecto a su ejecución en solitario. Para reducir estas pérdidas, se han diseñado políticas para gestionar estos recursos (por ej., políticas de reparticionando del espacio de cache o de configuración de la prebúsqueda hardware). Un importante desafío en diferentes ámbitos, como la nube o los sistemas de tiempo real, es predecir cómo la acción de una determinada política afectaría a las prestaciones de cada aplicaciones antes de realizarla. Este trabajo presenta una propuesta basada en redes neuronales que, a partir de datos obtenidos por una aplicación en ejecución concurrente con otras aplicaciones, estima las prestaciones que tendría una aplicación en solitario. Los resultados experimentales muestran que la propuesta se adapta dinámicamente a los cambios en el comportamiento de las aplicaciones de la carga. Las prestaciones estimadas presentan, de media, un error del 11,7% y del 2,3% para las métricas MAPE y MSE, respectivamente., Trabajo financiado por el Ministerio de Ciencia, Innovaciçon y Universidades y ERDF bajo la ayuda RTI2018-098156-B-C51, por la GVA AICO/2021/266, por la UPV PAID-01-20. Josué Feliu con apoyo de un contrato de formación Juan de la Cierva (FJC2018-036021-I)
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Segment Switching: A New Switching Strategy for Optical HPC Networks

  • Duro, José
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[EN] Photonics are becoming realistic technologies for implementing interconnection networks in near future Exascale supercomputer systems. Photonics present key features to design high-performance and scalable supercomputer networks, such as higher bandwidth and lower latencies than their electronic supercomputer networks counterparts. Some research work is focused on conventional network topologies built with photonic technologies, with the aim of taking advantage of photonic characteristics. Nevertheless, these approaches fail in that they keep low the network utilization. We looked into this downside and we found that circuit switching was the main performance limitation. In this article we propose a new switching mechanism, called Segment Switching, to address this constraint and improve the network utilization. Segment Switching splits the circuit in segments of the whole path, and uses buffering on selected nodes on the network. Experimental results show that the devised approach signicantly outperforms photonic circuit switching in conventional torus and fat tree networks by 70% and 90%, respectively., This work was supported in part by the Ministerio de Ciencia, Innovacion y Universidades and in part by the European ERDF under Grant RTI2018-098156-B-C51.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Bandwidth-Aware Dynamic Prefetch Configuration for IBM POWER8

  • Navarro, Carlos
  • Feliu-Pérez, Josué|||0000-0003-3017-4266
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[EN] Advanced hardware prefetch engines are being integrated in current high-performance processors. Prefetching can boost the performance of most applications, however, the induced bandwidth consumption can lead the system to a high contention for main memory bandwidth, which is a scarce resource in current multicores. In such a case, the system performance can be severely damaged. This article characterizes the applications' behavior in an IBM POWER8 machine, which presents many prefetch settings, varying the bandwidth contention. The study reveals that the best prefetch setting for each application depends on the main memory bandwidth availability, that is, it depends on the co-running applications. Based on this study, we propose Bandwidth-Aware Prefetch Configuration (BAPC) a scalable adaptive prefetching algorithm that improves the performance of multi-program workloads. BAPC increases the performance of the applications in a 12, 15, and 16 percent of 6-, 8-, and 10-application workloads over the IBM POWER8 default configuration. In addition, BAPC reduces bandwidth consumption in 39, 42, and 45 percent, respectively., This work was supported in part by Ministerio de Ciencia, Innovacion y Universidades and the European ERDF under Grant RTI2018-098156-B-C51, Generalitat Valenciana under Grant AICO/2019/317, and Universitat Politenica de Valencia (PAID-06-18) under Grant SP20180140.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Fast-Track Cache: A Huge Racetrack Memory L1 Data Cache

  • Tárrega, Hugo
  • Valero, Alejandro
  • Lorente Garcés, Vicente Jesús
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[EN] First-level (L1) caches have been traditionally implemented with Static Random-Access Memory (SRAM) technology, since it is the fastest memory technology, and L1 caches call for tight timing constraints in the processor pipeline. However, one of the main downsides of SRAM is its low density, which prevents L1 caches to improve their storage capacity beyond a few tens of KB. On the other hand, the recent Domain Wall Memory (DWM) technology overcomes such a constraint by arranging multiple bits in a mag- netic racetrack, and sharing a header to access those bits. Accessing a bit requires a shift operation to align the target bit under the header. Such shifts increase the final access latency, which is the main reason why DWM has been mostly used to implement slow last-level caches. This paper proposes a novel DWM-based L1 cache data array design, namely Fast-Track Cache (FTC), that allows L1 caches with bigger storage capacities while reducing the shift overhead thanks to an enhanced exploitation of spatial and temporal localities. Experimental results show that most FTC accesses do not re- quire shifts. As a consequence, and due to its larger capacity, FTC improves the processor performance on average by 15% over a conventional SRAM memory subsystem and the state-of-the-art TapeCache architecture based on DWM. At the same time, energy savings are improved on average by 34% over the conventional design., This work has been partially supported by Ministerio de Ciencia, Innovación y Universidades and the European ERDF under Grant RTI2018-098156-B-C51, by Agencia Estatal de Investigación (AEI) under Grant PID2019-105660RB-C21/AEI/10.13039/501100011033, and by Dept. of Science, University, and Knowledge Society, Government of Aragon, under Grant gaZ: T5820R research group.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Thread Isolation to Improve Symbiotic Scheduling on SMT Multicore Processors

  • Feliu-Pérez, Josué|||0000-0003-3017-4266
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Eeckhout, Lieven
[EN] Resource sharing is a critical issue in simultaneous multithreading (SMT) processors as threads running simultaneously on an SMT core compete for shared resources. Symbiotic job scheduling, which co-schedules applications with complementary resource demands, is an effective solution to maximize hardware utilization and improve overall system performance. However, symbiotic job scheduling typically distributes threads evenly among cores, i.e., all cores get assigned the same number of threads, which we find to lead to sub-optimal performance. In this paper, we show that asymmetric schedules (i.e., schedules that assign a different number of threads to each SMT core) can significantly improve performance compared to symmetric schedules. To leverage this finding, we propose thread isolation, a technique that turns symmetric schedules into asymmetric ones yielding higher overall system performance. Thread isolation identifies SMT-adverse applications and schedules them in isolation on a dedicated core to mitigate their sharp performance degradation under SMT. Our experimental results on an IBM POWER8 processor show that thread isolation improves system throughput by up to 5.5 percent compared to a state-of-the-art symmetric symbiotic job scheduler., Josue Feliu has been partially supported through a postdoctoral fellowship by the Generalitat Valenciana (APOSTD/2017/052). Additional support has been provided by the Ministerio de Ciencia, Innovacion y Universidades and the European ERDF under Grant RTI2018-098156-B-C51, as well as, by the Universitat Politenica de Valencia through the "Ayudas a Primeros Proyectos de Investigacion" (PAID-06-18) under grant SP20180140. Lieven Eeckhout's research program is supported through FWO grants no. G.0434.16N and G.0144.17N, and the European Research Council (ERC) Advanced Grant agreement no. 741097.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Caracterización de las prestaciones y la polución en las caches no inclusivas

  • Pons-Escat, Lucía|||0000-0002-4582-7744
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Pons Terol, Julio|||0000-0002-5654-6753
[ES] La jerarquía de cache ha estado en continua evolución en los procesadores generación tras generación. La última tendencia ha sido la redistribución del espacio de las estructuras de la jerarquía. Se ha reducido el área de la cache disponible por núcleo pero adoptando un esquema no inclusivo en la cache de último nivel. Como resultado, esta nueva organización disminuye el área total de cache disponible por núcleo. Por ello, la gestión del espacio de cache en esta organización se vuelve más crítica que en las caches inclusivas. Este artículo presenta un análisis detallado del impacto que tiene una cache de último nivel (L3) no inclusiva. Se presentan tres estudios de caracterización de las aplicaciones SPEC CPU2006 y SPEC CPU2017. Primero, se compara el reciente esquema no inclusivo con el esquema inclusivo. En segundo lugar, se analiza la relación entre el espacio de cache no inclusiva y las prestaciones -tiempo de ejecución- de las aplicaciones. Finalmente, se analiza la polución de la cache L3, donde se identifican y cuantifican tres principales fuentes de polución: useless prefetches, bad speculation, y poor locality. Los resultados experimentales muestran que las aplicaciones cuyas prestaciones son más sensitivas al espacio de cache de L3 experimentan un número de stalls_L2 y HPKI_L3 alto. Además, es posible identificar aquellas aplicaciones que introducen polución usando los contadores de prestaciones para así poder limitarles el espacio de cache., Este trabajo ha sido financiado por el Ministerio de Ciencia, Innovacion y Universidades mediante la beca FPU18/01948 y junto al FEDER europeo por el proyecto RTI2018-098156-B-C51.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Enforcing Predictability of Many-cores with DCFNoC

  • Picornell-Sanjuan, Tomás
  • Duato Marín, José Francisco
  • Flich Cardo, José
  • Hernández Luz, Carles|||0000-0003-1192-6223
[EN] The ever need for higher performance forces industry to include technology based on multi-processors system on chip (MPSoCs) in their safety-critical embedded systems. MPSoCs include a network-on-chip (NoC) to interconnect the cores between them and with memory and the rest of shared resources. Unfortunately, the inclusion of NoCs compromises guaranteeing time predictability as network-level conflicts may occur. To overcome this problem, in this paper we propose DCFNoC, a new time-predictable NoC design paradigm where conflicts within the network are eliminated by design. This new paradigm builds on top of the Channel Dependency Graph (CDG) in order to deterministically avoid network conflicts. The network guarantees predictability to applications and is able to naturally inject messages using a TDM period equal to the optimal theoretical bound without the need of using a computationally demanding offline process. DCFNoC is integrated in a tile-based many-core system and adapted to its memory hierarchy. Our results show that DCFNoC guarantees time predictability avoiding network interference among multiple running applications. DCFNoC always guarantees performance and also improves wormhole performance in a 4 × 4 setting by a factor of 3.7× when interference traffic is injected. For a 8 × 8 network differences are even larger. In addition, DCFNoC obtains a total area saving of 10.79% over a standard wormhole implementation., This work has been supported by MINECO under Grant BES-2016-076885, by MINECO and funds from the European ERDF under Grant TIN2015-66972-C05-1-R and Grant RTI2018-098156-B-C51, and by the EC H2020 RECIPE project under Grant 801137.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Tesis de maestría (MasterThesis)

Gestión eficiente del espacio de la LLC en procesadores de altas prestaciones

  • Pons-Escat, Lucía|||0000-0002-4582-7744
[ES] La memoria caché de último nivel (LLC) desempeña un papel clave en el rendimiento final del sistema en los procesadores multinúcleo actuales al reducir significativamente la cantidad de accesos a memoria principal. Sin embargo, la interferencia entre aplicaciones producida en este recurso compartido puede llevar al sistema a situaciones no deseadas, tanto de rendimiento como de equidad (fairness). Recientemente se han presentado propuestas para procesadores comerciales que han abordado con éxito problemas como la inequidad (unfairness) o han logrado mejorar el tiempo de ejecución (turnaround time o TT). Sin embargo, estas propuestas deben enfrentarse con la problemática de mantener el rendimiento del sistema en términos de instrucciones por ciclo (IPC), lo que constituye todo un reto. Este Trabajo Fin de Máster aborda el desafío de reducir el TT manteniendo (o incluso mejorando) el IPC, haciendo una gestión eficiente del espacio de la LLC. Las dos contribuciones principales de este trabajo son las siguientes. En primer lugar, se caracterizan las aplicaciones considerando su comportamiento en la LLC (rendimiento, reuso y ocupación) y el impacto en el rendimiento final del sistema (IPC). Este estudio i) identifica nuevos comportamientos de aplicaciones respecto a la LLC que no se han identificado en trabajos previos y que pueden tener un impacto adverso en las prestaciones finales, y ii) concluye que las aplicaciones pueden experimentar diferentes comportamientos respecto a la LLC a lo largo de su ejecución, que coinciden con las diferentes fases en las que cambia el IPC. En segundo lugar, en base a este estudio, se presenta una propuesta de particionado de caché, Critical-Phase Aware Partitioning Approach (CPA), que extiende una propuesta nuestra previa, Critical-Aware Partitioning Approach (CA). CPA considera los nuevos comportamientos de caché identificados con el fin de maximizar la ganancia de prestaciones. La ejecución del algoritmo CPA está ligada a los cambios de fase, con lo que el espacio de LLC se redistribuye solo cuando se detecta un cambio de fase en alguna de las aplicaciones en ejecución que resulta en un cambio de comportamiento. Los resultados experimentales muestran que CPA logra obtener mejoras significativas en el TT (más de un 40 % en algunas mezclas de aplicaciones) y el promedio normalizado del tiempo de ejecución (ANTT), mientras se mantiene el IPC o incluso se mejora más de un 3 % en algunas mezclas. Además, los resultados muestran que comparado con Dunn, una propuesta de particionado de caché del estado del arte, CPA nunca pierde más de un 1.5 % en prestaciones (IPC) en las mezclas estudiadas. Dunn, en cambio, puede sufrir pérdidas de rendimiento de más de un 5 % en algunas mezclas., [EN] The Last Level Cache (LLC) plays a key role in the final system performance of current multicore processors by significantly reducing the amount of main memory accesses. The inter-application interference at this shared resource, however, can lead the system to undesired situations regarding both performance and fairness. Recent approaches using commercial processors have successfully addressed either fairness or turnaround time (TT). Nevertheless, these approaches must face sustaining system performance in terms of instructions per cycle (IPC), which is challenging. This Master Thesis addresses the challenge of reducing the TT while sustaining (and even improving) the IPC by making an effective use of the LLC space. This work makes two main contributions. Firstly, applications are characterized considering the LLC behavior (cache performance, reuse and occupancy) and its impact on the system performance (IPC). This characterization i) identifies novel application behaviors regarding the LLC, not found in previous related works and that can adversely impact on the final performance, and ii) finds that applications may experience different LLC behaviors along their execution, and each behavior change matches with IPC phase changes. Secondly, based on this study, we propose the Critical-Phase Aware Partitioning Approach (CPA), which extends our previous work, the Critical-Aware Partitioning Approach (CA), to also consider the newly identified LLC behaviors in order to maximize the system performance. CPA is phase change driven, so cache space redistribution only happens when a phase change that results in a behavior change is detected in at least one of the applications concurrently running on the system. Experimental results show that CPA achieves significant improvements in TT (more than 40% in some workloads) and in average normalized turnaround time (ANTT), while sustaining the IPC or even improving it by more than 3% in several mixes. In contrast, results show that compared to Dunn, a state-of-the-art cache partitioning approach, CPA never decreases performance (i.e. IPC) by more than 1.5% in the studied mixes. Dunn, however, decreases it by more than 5% in some mixes., This work has been partially supported by the Ministerio de Ciencia, Innovación y Universidades and the European ERDF under Grant RTI2018-098156-B-C51.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

The Picture Fuzzy Distance Measure in Controlling Network Power Consumption

  • Roan, Ngan Thi
  • Andújar, Fran
  • Le, Son Hoang
  • Vu, Manh Van
  • Smarandache, Felorentin
  • Coll, Salvador|||0000-0002-7795-7252
  • Alonso Díaz, Marina|||0000-0001-7079-5898
  • Martínez-Rubio, Juan-Miguel|||0000-0002-9748-0332
  • López Rodríguez, Pedro Juan|||0000-0003-4544-955X
[EN] In order to solve the complex decision-making problems, there are many approaches and systems based on the fuzzy theory were proposed. In 1998, Smarandache introduced the concept of single-valued neutrosophic set as a complete development of fuzzy theory. In this paper, we research on the distance measure between single-valued neutrosophic sets based on the H-max measure of Ngan et al. [8]. The proposed measure is also a distance measure between picture fuzzy sets which was introduced by Cuong in 2013 [15]. Based on the proposed measure, an Adaptive Neuro Picture Fuzzy Inference System (ANPFIS) is built and applied to the decision making for the link states in interconnection networks. In experimental evaluation on the real datasets taken from the UPV (Universitat Politècnica de València) university, the performance of the proposed model is better than that of the related fuzzy methods., This work has been supported by MINECO and the European ERDF under grants PID2019-105903RB-I00 and RTI2018-098156-B-C51. The autor (R.T. Ngan) also would like to thank the Erasmus+ Mobility Program (2016-1-ES01-KA107-023453) for supporting her work.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Phase-Aware Cache Partitioning to Target Both Turnaround Time and System Performance

  • Pons-Escat, Lucía|||0000-0002-4582-7744
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Pons Terol, Julio|||0000-0002-5654-6753
  • Selfa, Vicent
[EN] The Last Level Cache (LLC) plays a key role in the system performance of current multi-cores by reducing the number of long latency main memory accesses. The inter-application interference at this shared resource, however, can lead the system to undesired situations regarding performance and fairness. Recent approaches have successfully addressed fairness and turnaround time (TT) in commercial processors. Nevertheless, these approaches must face sustaining system performance, which is challenging. This work makes two main contributions. LLC behaviors regarding cache performance, data reuse and cache occupancy, that adversely impact on the final performance are identified. Second, based on these behaviors, we propose the Critical-Phase Aware Partitioning Approach (CPA), which reduces TT while sustaining (and even improving) IPC by making an effective use of the LLC space. Experimental results show that CPA outperforms CA, Dunn and KPart state-of-the-art approaches, and improves TT (over 40 percent in some workloads) over Linux default behavior while sustaining or even improving IPC by more than 3 percent in several mixes., This work has been supported in part by Ministerio de Ciencia, Innovaci~on y Universidades and the European ERDF under Grant RTI2018-098156-B-C51, and Generalitat Valenciana under Grant AICO/2019/317.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Impact of the Array Shape and Memory Bandwidth on the Execution time of CNN Systolic Arrays

  • Yago, Eduardo
  • Castelló, Pau
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[EN] The use of Convolutional Neural Networks (CNN) has experienced a huge rise over the last recent years and its popularity has increased exponentially, mainly due to its application both for image recognition and certain applications related to artificial intelligence. The new applications of CNN request computing demands that are difficult to address by conventional processors. As a consequence, accelerators ¿both prototypes and commercial products¿ focusing on CNN computation have been proposed. Among these accelerators, those based on systolic arrays have acquired a special relevance; some examples are the Google¿s TPU and Eyeriss. Current research has focused on regular squared systolic arrays and most existing work assumes that there is enough memory bandwidth to feed the systolic array with input data. In this paper we explore the design of non-squared systolic arrays and address the impact of the memory bandwidth from a performance perspective. This work makes two main contributions. First, we found that some workloads with non-squared arrays achieve similar performance to systolic arrays twice as large, which can translate in area and/or energy benefits. Second, we present a performance comparison varying the main memory bandwidth for current DRAM devices. The analysis reveals that main memory bandwidth has a great impact on performance and that the decision of which technology use is key for the system performance. For the 64x64 array size it is necessary to use HBM2 memory to avoid the slowdown that would introduce cheaper technologies (e.g. DDR5 and DDR4)., This work has been supported by Ministerio de Ciencia, Innovacion y Universidades and the European ERDF under Grant RTI2018-098156-B-C51.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

HP-DCFNoC: High Performance Distributed Dynamic TDM Scheduler Based on DCFNoC Theory

  • Picornell-Sanjuan, Tomás
  • Duato Marín, José Francisco
  • Flich Cardo, José
  • Hernández Luz, Carles|||0000-0003-1192-6223
[EN] The need for increasing the performance of critical real-time embedded systems pushes the industry to adopt complex multi-core processor designs with embedded networks-on-chip. In this paper we present hp-DCFNoC, a distributed dynamic scheduler design that by relying on the key properties of a delayed confict-free NoC (DCFNoC) is able to achieve peak performance numbers very close to a wormhole-based NoC design without compromising its real-time guarantees. In particular, our results show that the proposed scheduler achieves an overall throughput improvement of 6.9x and 14.4x over a baseline DCFNoC for 16 and 64-node meshes, respectively. When compared against a standard wormhole router 95% of its network throughput is preserved while strict timing predictability as property is kept. This achievement opens the door to new high performance time predictable NoC designs., This work was supported in part by the Secretara de Estado de Investigacin Desarrollo e Innovacin (MINECO) under Grant BES-2016-076885, in part by the European Regional Development Fund (ERDF) under Grant TIN2015-66972-C05-1-R and Grant RTI2018-098156-B-C51, and in part by the EC H2020 European Institute of Innovation and Technology (SELENE) Project under Grant 871467.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Efficient Management of Cache Accesses to Boost GPGPU Memory Subsystem Performance

  • Candel-Margaix, Francisco
  • Valero Bresó, Alejandro
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[EN] To support the massive amount of memory accesses that GPGPU applications generate, GPU memory hierarchies are becoming more and more complex, and the Last Level Cache (LLC) size considerably increases each GPU generation. This paper shows that counter-intuitively, enlarging the LLC brings marginal performance gains in most applications. In other words, increasing the LLC size does not scale neither in performance nor energy consumption. We examine how LLC misses are managed in typical GPUs, and we find that in most cases the way LLC misses are managed are precisely the main performance limiter. This paper proposes a novel approach that addresses this shortcoming by leveraging a tiny additional Fetch and Replacement Cache-like structure (FRC) that stores control and coherence information of the incoming blocks until they are fetched from main memory. Then, the fetched blocks are swapped with the victim blocks (i.e., selected to be replaced) in the LLC, and the eviction of such victim blocks is performed from the FRC. This approach improves performance due to three main reasons: i) the lifetime of blocks being replaced is enlarged, ii) the main memory path is unclogged on long bursts of LLC misses, and iii) the average LLC miss latency is reduced. The proposal improves the LLC hit ratio, memory-level parallelism, and reduces the miss latency compared to much larger conventional caches. Moreover, this is achieved with reduced energy consumption and with much less area requirements. Experimental results show that the proposed FRC cache scales in performance with the number of GPU compute units and the LLC size, since, depending on the FRC size, performance improves ranging from 30 to 67 percent for a modern baseline GPU card, and from 32 to 118 percent for a larger GPU. In addition, energy consumption is reduced on average from 49 to 57 percent for the larger GPU. These benefits come with a small area increase (by 7.3 percent) over the LLC baseline., This work has been supported by the Spanish Ministerio de Ciencia, Innovacion y Universidades and the European ERDF under Grants T-PARCCA (RTI2018-098156-B-C51), and TIN2016-76635-C2-1-R (AEI/ERDF, EU), by the Universitat Politecnica de Valencia under Grant SP20190169, and by the gaZ: T58_17R research group (Aragon Gov. and European ESF).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Cloud White: Detecting and Estimating QoS Degradation of Latency-Critical Workloads in the Public Cloud

  • Pons-Escat, Lucía|||0000-0002-4582-7744
  • Feliu-Pérez, Josué|||0000-0003-3017-4266
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Pons Terol, Julio|||0000-0002-5654-6753
  • Huang, Chaoyi
[EN] The increasing popularity of cloud computing has forced cloud providers to build economies of scale to meet the growing demand. Nowadays, data-centers include thousands of physical machines, each hosting many virtual machines (VMs), which share the main system resources, causing interference that can significantly impact on performance. Frequently, these data-centers run latency-critical workloads, whose performance is determined by tail latency, which is very sensitive to the interference of co-running workloads. To prevent QoS violations, cloud providers adopt overprovisioning strategies but they reduce the server utilization and increase the costs. A mechanism that accurately estimates performance degradation dynamically in a production system would allow cloud providers to improve the servers' utilization. In this work we propose Cloud White, an approach that is able to detect the inter-VM interference in scenarios with multiple co-located latency-critical VMs and estimate the performance degradation using multi-variable regression models. Unlike previous proposals, Cloud White is built taking into account the limitations of a public cloud production system. Experimental results show that Cloud White is able to estimate performance degradation with a small overall prediction error of 5%., This work has been supported by Huawei Cloud, and in part by Spanish Ministerio de Universidades under grant FPU18/01948, and by Spanish Ministerio de Universidades and European ERDF under grants RTI2018-098156-B-C51 and PID2021-123627OB-C51. Funding for open access charge: CRUE-Universitat Politec-nica de Valencia.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Análisis del impacto de una memoria cache en un controlador de memoria principal NVRAM

  • Avargues, Miguel A.
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[ES] La descentralización de la computación permite a los dispositivos móviles y ultramóviles delegar la ejecución de aplicaciones con mucho cómputo en servidores a través de Internet. Muchas de estas aplicaciones utilizan grandes cantidades de datos que exceden en gran medida la capacidad de almacenamiento de la memoria principal (basada en tecnología DRAM) de los servidores actuales. En consecuencia, las memorias DRAM convencionales solo son capaces de almacenar una pequeña fracción de los datos utilizados en un momento dado de la ejecución de estas aplicaciones, convirtiéndose en el principal cuello de botella en las prestaciones. Las memorias RAM no volátiles (NVRAM), gracias a su mayor densidad y capacidad de almacenamiento, podrían considerarse como una posible solución alternativa a este problema. Sin embargo, no son una solución directa sino que se requiere rediseñar el subsistema de memoria para ocultar sus altas latencias de acceso. En este artículo se presenta un controlador de memoria principal NVRAM que incorpora una memoria cache para ocultar las altas latencias de acceso. Los resultados demuestran que se no solo se igualan sino que es posible mejorar las prestaciones sobre memoria principal DRAM., Este trabajo ha sido parcialmente financiado por el Ministerio de Ciencia, Innovacion y Universidades y el FEDER europeo a traves del proyecto RTI2018- 098156-B-C51.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Hy-Sched: A Simple Hyperthreading-Aware Thread to Core Allocation Strategy

  • Navarro-Edo, Marta|||0000-0002-4334-460X
  • Pons-Escat, Lucía|||0000-0002-4582-7744
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[EN] Simultaneous multithreading processors are dominating the High Computing Performance market. Among these processors, those supporting only two threads are being the most widely deployed in current systems, thus, only two threads compete at run-time for intra-core resources. The performance of these processors can be boosted by selecting symbiotic applications to be executed on the same core, which reduces the inter-application interference considerably. In this letter we propose Hy-Sched, an scheduling algorithm that exploits symbiosis to make pairs of applications to be launched on the same physical core. The proposed approach lies on the categories of the Top-Down Method for Performance Analysis. Different variants of the algorithm are explored. Experimental results show that Hy-Sched outperforms Linux on average by 15 percent in the studied workloads., This work was supported in part by Ministerio de Ciencia, Innovacion y Universidades and the European ERDF under Grant RTI2018-098156-B-C51, and Generalitat Valenciana under Grant AICO/2019/317.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Caracterización de prestaciones de aplicaciones paralelas de dispositivos móviles

  • Navarro-Edo, Marta|||0000-0002-4334-460X
  • Pons-Escat, Lucía|||0000-0002-4582-7744
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • Pons Terol, Julio|||0000-0002-5654-6753
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Lorente, Vicent
[ES] Los dispositivos m¿oviles, como smartphones o tabletas, est¿an presentes en los distintos ¿ambitos de nuestra vida cotidiana. Recientemente, estos dispositivos m¿oviles han comenzado a incorporar procesadores SMT, donde las interferencias entre aplicaciones son m¿as intensas porque tambi¿en hay compartici¿on de recursos dentro del n¿ucleo. Este trabajo analiza la interferencia que hay en las aplicaciones paralelas entre los hilos o procesos que ha creado.El hecho de utilizar cargas paralelas supondr¿¿a nuevos desaf¿¿os al planificador. En este trabajo se realiza una caracterizaci¿on de las aplicaciones paralelas, como paso preliminar al dise¿no de un planificador para este tipo de aplicaciones, analizando el comportamiento din¿amico de los hilos de las aplicaciones paralelas analizadas. Se ha analizado c¿omo afecta el n¿umero de hilos de las aplicaciones a las prestaciones y cu¿anto afecta a las prestaciones que los hilos de la aplicaci¿on compartan los recursos de todo el sistema, incluidos los del n¿ucleo f¿¿sico. De esta manera, un planificador podr¿¿a basar su gesti¿on en este estudio para mejorar las prestaciones. Se ha comprobado que las aplicaciones multihilo/multiproceso presentan una amplia variedad de categor¿¿as y que algunas aplicaciones apenas presentan p¿erdidas de prestaciones cuando se ejecutan en modo SMT, a pesar de ejecutarse en la mitad de los n¿ucleos f¿¿sicos, es decir, son SMT amigables., Este trabajo ha sido financiado por los fondos FEDER europeos bajo el proyecto RTI2018-098156-BC51 y por la Generalitat Valenciana bajo la ayuda AICO/2021/266.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Software de investigación Software (Software)

Critical Phase-Aware Partitioning Approach (CPA), Phase-Aware Cache Partitioning to Target both Turnaround Time and System Performance

  • Pons-Escat, Lucía|||0000-0002-4582-7744
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Pons Terol, Julio|||0000-0002-5654-6753
  • Selfa Oliver, Vicent
CPA is LLC (Last Level Cache) partitioning approach that performs an efficient cache space distribution among executing applications. To assign partitions (ways) of the LLC, Intel CAT is used. This policy is included in a framework named manager which is able to launch the experiments., This work has been partially supported by Ministerio de Ciencia, Innovación y Universidades and the European ERDF under Grant RTI2018-098156-B-C51, and Generalitat Valenciana under Grant AICO/2019/317.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

DeepP: Deep Learning Multi-Program Prefetch Configuration for the IBM POWER 8

  • Lurbe-Sempere, Manel
  • Feliu-Pérez, Josué|||0000-0003-3017-4266
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
[EN] Current multi-core processors implement sophisticated hardware prefetchers, that can be configured by application (PID),to improve the system performance. When running multiple applications, each application can present different prefetch requirements,hence different configurations can be used. Setting the optimal prefetch configuration for each application is a complex task since itdoes not only depend on the application characteristics but also on the interference at the shared memory resources (e.g. memorybandwidth). In his paper, we proposeDeepP, a deep learning approach for the IBM POWER8 that identifies at run-time the bestprefetch configuration for each application in a workload. To this end, the neural network predicts the performance of each applicationunder the studied prefetch configurations by using a set of performance events. The prediction accuracy of the network is improvedthanks to a dynamic training methodology that allows learning the impact of dynamic changes of the prefetch configuration onperformance. At run-time, the devised network infers the best prefetch configuration for each application and adjusts it dynamically.Experimental results show that the proposed approach improves performance, on average, by 5,8%, 6,7%, and 15,8% compared tothe default prefetch configuration across different 6-, 8-, and 10-application workloads, respectively., This work was supported in part by Ministerio de Ciencia, Innovacion y Universidades and the European ERDF under Grant RTI2018-098156-B-C51, in part by Generalitat Valenciana under Grant AICO/2021/266, and in part by Ayudas Contratos predoctorales UPV -subprograma 1 (PAID-01-20). The work of Josue Feliu was supported by a Juan de la Cierva Formacion Contract under Grant FJC2018-036021-I.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Effect of Hyper-Threading in Latency-Critical Multithreaded Cloud Applications and Utilization Analysis of the Major System Resources

  • Pons-Escat, Lucía|||0000-0002-4582-7744
  • Feliu-Pérez, Josué|||0000-0003-3017-4266
  • Petit Martí, Salvador Vicente|||0000-0003-2426-4134
  • Pons Terol, Julio|||0000-0002-5654-6753
  • Gómez Requena, María Engracia|||0000-0003-1466-4118
  • Sahuquillo Borrás, Julio|||0000-0001-8630-4846
  • Puche-Lara, José
  • Huang, Chaoyi
[EN] Multithreaded latency-critical applications represent an important subset of workloads running on public cloud systems. Most of these systems deploy powerful computing servers including Intel Hyper-Threading processors. Understanding how performance is affected by the consumption of the main system resources is a major concern for cloud providers in order to devise virtualization strategies that improve the system efficiency. With this aim, this paper first characterizes the impact of QPS on tail latency, analyzing different scenarios varying the number of threads and the thread-to-core allocation (single-task and multi-task execution) policy. The characterization study reveals that the performance of some applications does not scale with the number of threads, and the performance of some others is insensitive to the Hyper-Threading technology, so they can be allocated in less physical cores and improve system utilization. Identifying these applications, however, at run-time is challenging. Despite identifying these applications at run-time is challenging, this paper shows that they can be successfully detected at run-time by analyzing the utilization trend of the major system resources. In addition to CPU, we have also studied how assigning the share of each application of other major shared system resources impacts on performance. We outline considerations cloud providers should take into account to improve performance and resource utilization., Acknowledgments This work has been supported by Huawei Cloud, and in part by Spanish Ministerio de Universidades under grant FPU18/01948, and by Spanish Ministerio de Universidades and European ERDF under grant RTI2018-098156-B-C51.