Analytical Performance Modeling of Hierarchical Interconnect Fabrics

Nikitin, Nikita; Pedro, Javier de_San; Carmona, Josep; Cortadella, Jordi

doi:10.1109/nocs.2012.20

Cited by 9 publications

(3 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…During the first stage, analytical models are used to rapidly prune the design space and generate a set of promising configurations in the area/ power/performance space. The analytical model from [16] is used to evaluate CMP configurations and discriminate those with poor performance. Static and dynamic power are also evaluated using analytical approximations based on the area and activity of the CMP components.…”

Section: Architectural Explorationmentioning

confidence: 99%

Physical-aware system-level design for tiled hierarchical chip multiprocessors

Cortadella

Pedro

Nikitin

et al. 2013

Proceedings of the 2013 ACM International Symposium on Physical Design

Self Cite

View full text Add to dashboard Cite

Tiled hierarchical architectures for Chip Multiprocessors (CMPs) represent a rapid way of building scalable and power-efficient many-core computing systems. At the early stages of the design of a CMP, physical parameters are often ignored and postponed for later design stages. In this work, the importance of physical-aware system-level exploration is investigated, and a strategy for deriving chip floorplans is described. Additionally, wire planning of the on-chip interconnect is performed, as its topology and organization affect the physical layout of the system. Traditional algorithms for floorplanning and wire planning are customized to include physical constraints specific for tiled hierarchical architectures. Over-the-cell routing is used as one of the major area savings strategy. The combination of architectural exploration and physical planning is studied with an example and the impact of the physical aspects on the selection of architectural parameters is evaluated.

show abstract

Section: Architectural Explorationmentioning

confidence: 99%

Physical-aware system-level design for tiled hierarchical chip multiprocessors

Cortadella

Pedro

Nikitin

et al. 2013

Proceedings of the 2013 ACM International Symposium on Physical Design

Self Cite

View full text Add to dashboard Cite

show abstract

“…During the first stage, analytical models are used to rapidly prune the design space and generate a set of promising configurations in the area/power/ performance space. The analytical model from [114] is used to evaluate CMP configurations and discriminate those with poor performance. Static and dynamic power are also evaluated using analytical approximations based on the area and activity of the CMP components [115].…”

Section: Architectural Explorationmentioning

confidence: 99%

Structure discovery techniques for circuit design and process model visualization

San Pedro Martín

View full text Add to dashboard Cite

Graphs are one of the most used abstractions in many knowledge fields because of the easy and flexibility by which graphs can represent relationships between objects. The pervasiveness of graphs in many disciplines means that huge amounts of data are available in graph form, allowing many opportunities for the extraction of useful structure from these graphs in order to produce insight into the data. In this thesis we introduce a series of techniques to resolve well-known challenges in the areas of digital circuit design and process mining. The underlying idea that ties all the approaches together is discovering structures in graphs. We show how many problems of practical importance in these areas can be solved utilizing both common and novel structure mining approaches. In the area of digital circuit design, this thesis proposes automatically discovering frequent, repetitive structures in a circuit netlist in order to improve the quality of physical planning. These structures can be used during floorplanning to produce regular designs, which are known to be highly efficient and economical. At the same time, detecting these repeating structures can exponentially reduce the total design time. The second focus of this thesis is in the area of the visualization of process models. Process mining is a recent area of research which centers on studying the behavior of real-life systems and their interactions with the environment. Complicated process models, however, hamper this goal. By discovering the important structures in these models, we propose a series of methods that can derive visualization-friendly process models with minimal loss in accuracy. In addition, and combining the areas of circuit design and process mining, this thesis opens the area of specification mining in asynchronous circuits. Instead of the usual design flow, which involves synthesizing circuits from specifications, our proposal discovers specifications from implemented circuits. This area allows for many opportunities for verification and re-synthesis of asynchronous circuits. The proposed methods have been tested using real-life benchmarks, and the quality of the results compared to the state-of-the-art. Els grafs són una de les representacions abstractes més comuns en molts camps de recerca, gràcies a la facilitat i flexibilitat amb la que poden representar relacions entre objectes. Aquesta popularitat fa que una gran quantitat de dades es puguin trobar en forma de graf, i obre moltes oportunitats per a extreure estructures d'aquest grafs, útils per tal de donar una intuïció millor de les dades subjacents. En aquesta tesi introduïm una sèrie de tècniques per resoldre reptes habitualment trobats en les àrees de disseny de circuits digitals i mineria de processos industrials. La idea comú sota tots els mètodes proposats es descobrir automàticament estructures en grafs. En la tesi es mostra que molts problemes trobats a la pràctica en aquestes àrees poden ser resolts utilitzant nous mètodes de descobriment d'estructures. En l'àrea de disseny de circuits, proposem descobrir, automàticament, estructures freqüents i repetitives en les definicions del circuit per tal de millorar la qualitat de les etapes posteriors de planificació física. Les estructures descobertes poden fer-se servir durant la planificació per produir dissenys regulars, que son molt més econòmics d'implementar. Al mateix temps, la descoberta i ús d'aquestes estructures pot reduir exponencialment el temps total de disseny. El segon punt focal d'aquesta tesi és en l'àrea de la visualització de models de processos industrials. La mineria de processos industrials es un tema jove de recerca que es centra en estudiar el comportament de sistemes reals i les interaccions d'aquests sistemes amb l'entorn. No obstant, quan d'aquest anàlisi s'obtenen models massa complexos visualment, l'estudi n'és problemàtic. Proposem una sèrie de mètodes que, gràcies al descobriment automàtic de les estructures més importants, poden generar models molt més fàcils de visualitzar que encara descriuen el comportament del sistema amb gran precisió. Combinant les àrees de disseny de circuits i mineria de processos, aquesta tesi també obre un nou tema de recerca: la mineria d'especificacions per circuits asíncrons. En l'estil de disseny asíncron habitual, sintetitzadors automàtics generen circuits a partir de les especificacions. En aquesta tesi proposem el pas invers: descobrir automàticament les especificacions de circuits ja implementats. Així, creem noves oportunitats per a la verificació i la re-síntesi de circuits asíncrons. Els mètodes proposats en aquesta tesi s'han validat fent servir dades obtingudes d'aplicacions pràctiques, i en comparem els resultats amb els mètodes existents.

show abstract

“…This chapter addresses the problem of analytical modeling of CMP architectures. The contribution in this field is the analytical method for estimating performance of large-scale hierarchical CMPs with hundreds and thousands of cores, published in [102]. Specifically, it was shown that contention of the interconnect network has a major significance when analyzing CMP performance.…”

Section: Chapter 4 Analytical Modeling Of Cmp Architecturesmentioning

confidence: 99%

Automatic synthesis and optimization of chip multiprocessors

Nikitin

View full text Add to dashboard Cite

The microprocessor technology has experienced an enormous growth during the last decades. Rapid downscale of the CMOS technology has led to higher operating frequencies and performance densities, facing the fundamental issue of power dissipation. Chip Multiprocessors (CMPs) have become the latest paradigm to improve the power-performance efficiency of computing systems by exploiting the parallelism inherent in applications. Industrial and prototype implementations have already demonstrated the benefits achieved by CMPs with hundreds of cores.CMP architects are challenged to take many complex design decisions. Only a few of them are:- What should be the ratio between the core and cache areas on a chip?- Which core architectures to select?- How many cache levels should the memory subsystem have?- Which interconnect topologies provide efficient on-chip communication?These and many other aspects create a complex multidimensional space for architectural exploration. Design Automation tools become essential to make the architectural exploration feasible under the hard time-to-market constraints. The exploration methods have to be efficient and scalable to handle future generation on-chip architectures with hundreds or thousands of cores.Furthermore, once a CMP has been fabricated, the need for efficient deployment of the many-core processor arises. Intelligent techniques for task mapping and scheduling onto CMPs are necessary to guarantee the full usage of the benefits brought by the many-core technology. These techniques have to consider the peculiarities of the modern architectures, such as availability of enhanced power saving techniques and presence of complex memory hierarchies.This thesis has several objectives. The first objective is to elaborate the methods for efficient analytical modeling and architectural design space exploration of CMPs. The efficiency is achieved by using analytical models instead of simulation, and replacing the exhaustive exploration with an intelligent search strategy. Additionally, these methods incorporate high-level models for physical planning. The related contributions are described in Chapters 3, 4 and 5 of the document.The second objective of this work is to propose a scalable task mapping algorithm onto general-purpose CMPs with power management techniques, for efficient deployment of many-core systems. This contribution is explained in Chapter 6 of this document.Finally, the third objective of this thesis is to address the issues of the on-chip interconnect design and exploration, by developing a model for simultaneous topology customization and deadlock-free routing in Networks-on-Chip. The developed methodology can be applied to various classes of the on-chip systems, ranging from general-purpose chip multiprocessors to application-specific solutions. Chapter 7 describes the proposed model.The presented methods have been thoroughly tested experimentally and the results are described in this dissertation. At the end of the document several possible directions for the future research are proposed.

show abstract

Analytical Performance Modeling of Hierarchical Interconnect Fabrics

Cited by 9 publications

References 24 publications

Physical-aware system-level design for tiled hierarchical chip multiprocessors

Physical-aware system-level design for tiled hierarchical chip multiprocessors

Structure discovery techniques for circuit design and process model visualization

Automatic synthesis and optimization of chip multiprocessors

Contact Info

Product

Resources

About