Generating Process Network Communication Infrastructure for Custom Multi-Core Platforms

We present an approach for generating implementations of abstraction layers implementing the communication infrastructure of applications modeled as process networks. Our approach is unique in that it does not rely on assumptions about the capabilities and topology of the underlying platform. Instead, a generic implementation is adapted to the particular platform based on information retrieved from analyzing the platform. At the heart of the approach is a novel method for analyzing the capabilities of custom execution platforms composed of components. The versatility and usefulness of the approach and analysis method is demonstrated through a case study.

Joint Uplink and Downlink Performance Profiling of LTE Protocol Processing on a Mobile Platform

This article provides a detailed profiling of the layer 2 (L2) protocol processing for 3G successor Long Term Evolution (LTE). For this purpose, the most processing intensive part of the LTE L2 data plane is executed on top of a virtual ARM based mobile phone platform. The authors measure the execution times as well as the maximum data rates at different system setups. The profiling is done for uplink (UL) and downlink (DL) directions separately as well as in a joint UL and DL scenario. As a result, the authors identify time critical algorithms in the protocol stack and check to what extent state-of-the-art hardware platforms with a single-core processor and traditional hardware acceleration concepts are still applicable for protocol processing in LTE and beyond LTE mobile devices.

A Multidimensional Software Cache for Scratchpad-Based Systems

In many kernels of multimedia applications, the working set is predictable, making it possible to schedule the data transfers before the computation. Many other kernels, however, process data that is known just before it is needed or have working sets that do not fit in the scratchpad memory. Furthermore, multimedia kernels often access two or higher dimensional data structures and conventional software caches have difficulties to exploit the data locality exhibited by these kernels. For such kernels, the authors present a Multidimensional Software Cache (MDSC), which stores 1- 4 dimensional blocks to mimic in cache the organization of the data structure. Furthermore, it indexes the cache using the matrix indices rather than linear memory addresses. MDSC also makes use of the lower overhead of Direct Memory Access (DMA) list transfers and allows exploiting known data access patterns to reduce the number of accesses to the cache. The MDSC is evaluated using GLCM, providing an 8% performance improvement compared to the IBM software cache. For MC, several optimizations are presented that reduce the number of accesses to the MDSC.

Hierarchical Agent Monitored Parallel On-Chip System

In this paper, the authors present a formal specification of a novel design paradigm, hierarchical agent monitored SoCs (HAMSOC). The paradigm motivates dynamic monitoring in a hierarchical and distributed manner, with adaptive agents embedded for local and global operations. Formal methods are of essential importance to the development of such a novel and complex platform. As the initial effort, functional specification is indispensable to the non-ambiguous system modeling before potential property verification. The formal specification defines the manner by which the system can be constructed with hierarchical components and the representation of run-time information in modeling entities and every type of the monitoring operations. The syntax follows the standard set theory with additional glossary and notations introduced to facilitate practical SoC design process. A case study of hierarchical monitoring for power management in NoC (Network-on-chip), written with the formal specification, is demonstrated.

Joint Validation of Application Models and Multi-Abstraction Network-on-Chip Platforms

Application models are often disregarded during the design of multiprocessor Systems-on-Chip (MPSoC). This is due to the difficulties of capturing the application constraints and applying them to the design space exploration of the platform. In this article we propose an application modelling formalism that supports joint validation of application and platform models. To support designers on the trade-off analysis between accuracy, observability, and validation speed, we show that this approach can handle the successive refinement of platform models at multiple abstraction levels. A case study of the joint validation of a single application successively mapped onto three different platform models demonstrates the applicability of the presented approach.

Modeling Communication in Multi–Processor Systems–on–Chip Using Modular Connectors

Formal methods of concurrent programming can be used to develop and verify complex Multi–Processor Systems–On–Chip in order to ensure that these systems satisfy their functional and communication requirements. The authors use the Action Systems formalism and show how asynchronous communication of Multi–Processor Systems–on–Chip can be modeled using generic connectors composed out of simple channel components. The paper proposes a new approach to modeling generic and hierarchical connectors for handling the complexity of on–chip communication and data flow. The authors’ goal is to avoid overloaded bus–based architectures and give a distributed framework. A case study presents the authors’ modeling methodology.

Schedulability Analysis for Real Time On-Chip Communication with Wormhole Switching

In this paper, the authors discuss a real-time on-chip communication service with a priority-based wormhole switching policy. The authors present a novel off-line schedulability analysis approach, worst case network latency analysis. By evaluating diverse inter-relationships and service attributes among the traffic flows, this approach can predict the packet network latency for all practical situations. The simulation results provide evidence that communication latency calculated using the real time analysis approach is safe, closely matching the figures obtained from simulation.

Performance Analysis of On-Chip Communication Structures under Device Variability

On-chip communication is becoming an important bottleneck in the design and operation of high performance systems where it has to face additional challenges due to device variability. Communication structures such as tapered buffer drivers, interconnects, repeaters, and data storage elements are vulnerable to variability, which can limit the performance of the on-chip communication networks. In this regard, it becomes important to have a complete understanding of the impact that variability will have on the performance of these circuit elements in order to design high yield and reliable systems. In this paper, the authors have characterized the performance of the communication structures under the impact of random dopant fluctuation (RDF) for the future technology generations of 25, 18, and 13 nm. For accurate characterization of their performance, a Monte Carlo simulation method has been used along with predictive device models for the given technologies. Analytical models have been developed for the link failure probability of a repeater inserted interconnect which uses characterization data of all communication structures to give an accurate prediction of the link failure probability. The model has also been extended to calculate the link failure probability of a wider communication link.

System Architecture for 3GPP-LTE Modem using a Programmable Baseband Processor

The evolution of third generation mobile communications toward high-speed packet access and long-term evolution is ongoing and will substantially increase the throughput with higher spectral efficiency. This paper presents the system architecture of an LTE modem based on a programmable baseband processor. The architecture includes a baseband processor that handles processing time and frequency synchronization, IFFT/FFT (up to 2048-p), channel estimation and subcarrier de-mapping. The throughput and latency requirements of a Category four User Equipment (CAT4 UE) is met by adding a MIMO symbol detector and a parallel Turbo decoder supporting H-ARQ, which brings both low silicon cost and enough flexibility to support other wireless standards. The complexity demonstrated by the modem shows the practicality and advantage of using programmable baseband processors for a single-chip LTE solution.

Parallel Programming and Its Architectures Based on Data Access Separated Algorithm Kernels

A novel master-multi-SIMD architecture and its kernel (template) based parallel programming flow is introduced as a parallel signal processing platform. The name of the platform is ePUMA (embedded Parallel DSP processor architecture with Unique Memory Access). The essential technology is to separate data accessing kernels from arithmetic computing kernels so that the run-time cost of data access can be minimized by running it in parallel with algorithm computing. The SIMD memory subsystem architecture based on the proposed flow dramatically improves the total computing performance. The hardware system and programming flow introduced in this article will primarily aim at low-power high-performance embedded parallel computing with low silicon cost for communications and similar real-time signal processing.