Front-end Processors Research Articles

CU Wirebond Technology in 16FFC High Performance Automotive Radar Processor with IR Drop Reduction Methodology

High performance automotive radar front-end processors normally use advanced semiconductor packaging such as Flip Chip, fan-in and fan-out wafer level Ball Grid Array (BGA) packages in order to provide superior performance in their applications. However, for a cost effective package in the competitive business market, wirebonded package can also provide similar uncompromised performance. Advanced Cu wirebond technology was evaluated on the 16FFc radar processor. However, the challenge of high voltage drop, also known as IR drop, in 16FFc technology has to be overcome. As wafer technology moves to smaller advanced nodes, the back-end-of-line (BEOL) metal thickness gets thinner, thus increasing the resistance per unit length. Another trend in semiconductor design has been a reduction in operating voltage, meaning that small changes in supply voltage may represent an increasing percentage of the digital swing and potentially lead to incorrect logic values. As current flows through a resistor, the high voltage drop could slow down the circuitry, impact the circuit timing and lead to functional failure. In order to reduce the IR drop in the radar processor when using Cu wirebond, a large quantity of intra-die wires was applied on the 16FFc silicon. IR drop simulation was performed on 3 layout conditions, 0%, 50% and 100% of the intra-die wires. Empirical data was also collected with actual Cu wirebonded samples. The fully populated intra-die wires reduced IR drop by 30%. To enable the reliable performance of the intra-die wires, Stand-off Stitch Bond (SSB) wirebond technology was introduced within the package. The development and reliability evaluation was conducted for Cu wirebonding on 16FFc automotive radar processor of 30mm2 die size in a 14mm x14mm 2-layer substrate BGA package. Wirebond recipe was developed based on detailed Design of Experiment to establish bonding parameters for both standard wires and the intra-die wires. Packages were tested per AEC Grade 1 and AEC Q006 qualifications requirements. All results from package reliability assessments passed with no abnormality. Board level temperature cycling reliability for the BGA package with the four depopulated corner solder balls was evaluated with results surpassing NXP requirement for AEC Grade 1. This paper will present the results and analysis from (1) IR drop simulation and empirical data collection, (2) wirebond development and reliability of standard and intra-die wires on 16FFc Silicon, and (3) board level reliability of a BGA package with four depopulated corner solder balls.

Control and Information Systems for the National Ignition Facility

Orchestration of every National Ignition Facility (NIF) shot cycle is managed by the Integrated Computer Control System (ICCS), which uses a scalable software architecture running code on more than 1950 front-end processors, embedded controllers, and supervisory servers. The ICCS operates laser and industrial control hardware containing 66 000 control and monitor points to ensure that all of NIF’s laser beams arrive at the target within 30 ps of each other and are aligned to a pointing accuracy of less than 50 μm root-mean-square, while ensuring that a host of diagnostic instruments record data in a few billionths of a second. NIF’s automated control subsystems are built from a common object-oriented software framework that distributes the software across the computer network and achieves interoperation between different software languages and target architectures. A large suite of business and scientific software tools supports experimental planning, experimental setup, facility configuration, and post-shot analysis. Standard business services using open-source software, commercial workflow tools, and database and messaging technologies have been developed. An information technology infrastructure consisting of servers, network devices, and storage provides the foundation for these systems. This paper is an overview of the control and information systems used to support a wide variety of experiments during the National Ignition Campaign.

Status of the National Ignition Facility Integrated Computer Control System (ICCS) on the path to ignition

The National Ignition Facility (NIF) at the Lawrence Livermore National Laboratory is a stadium-sized facility under construction that will contain a 192-beam, 1.8-MJ, 500-TW, ultraviolet laser system together with a 10-m diameter target chamber with room for multiple experimental diagnostics. NIF is the world's largest and most energetic laser experimental system, providing a scientific center to study inertial confinement fusion (ICF) and matter at extreme energy densities and pressures. NIF's laser beams are designed to compress fusion targets to conditions required for thermonuclear burn, liberating more energy than required to initiate the fusion reactions. NIF is comprised of 24 independent bundles of eight beams each using laser hardware that is modularized into more than 6000 line replaceable units such as optical assemblies, laser amplifiers, and multi-function sensor packages containing 60,000 control and diagnostic points. NIF is operated by the large-scale Integrated Computer Control System (ICCS) in an architecture partitioned by bundle and distributed among over 800 front-end processors and 50 supervisory servers. NIF's automated control subsystems are built from a common object-oriented software framework based on CORBA distribution that deploys the software across the computer network and achieves interoperation between different languages and target architectures. A shot automation framework has been deployed during the past year to orchestrate and automate shots performed at the NIF using the ICCS. In December 2006, a full cluster of 48 beams of NIF was fired simultaneously, demonstrating that the independent bundle control system will scale to full scale of 192 beams. At present, 72 beams have been commissioned and have demonstrated 1.4-MJ capability of infrared light. During the next 2 years, the control system will be expanded in preparation for project completion in 2009 to include automation of target area systems including final optics, target positioners and diagnostics. Additional capabilities to support fusion ignition shots in a National Ignition Campaign (NIC) beginning in 2010 will include a cryogenic target system, target diagnostics, and integrated experimental shot data analysis with tools for data visualization and archiving. This talk discusses the current status of the control system implementation and discusses the plan to complete the control system on the path to ignition.

The Process Manager in the ATLAS DAQ System

This paper describes the process manager in the ATLAS DAQ system. The purpose of the process manager is to perform basic process control on behalf of the software components of the DAQ system. It is able to create, destroy and monitor the basic status (e.g., running, exited, killed) of software components on the DAQ workstations and front-end processors. Section I gives a brief overview of the process manager functionalities. Section II focuses on the requirements the process manager system has to fulfil to be fully integrated in the DAQ system. Section III shows how the requirements are met by the current implementation. The communication schema between the different parts of the process manager system, the procedure to launch a process and the possible states in which a process can be are described in Sections IV ,V and VI. Section VII deals with some consideration of the process manager performance while some conclusions are given in Section VIII.

Orchestrating shots for the national ignition racility

The National Ignition Facility (NIF), currently under construction at the Lawrence Livermore National Laboratory, is a stadium-sized facility containing a 192-beam, 1.8 Megajoule, 500-Terawatt, ultra-violet laser system together with a 10-meter diameter target chamber with room for nearly 100 experimental diagnostics. When completed, NIF will be the world's largest and most energetic laser experimental system, providing an international center to study inertial confinement fusion and physics of matter at extreme densities and pressures. The NIF is operated by the Integrated Computer Control System (ICCS), which is a layered architecture of over 700 lower-level front-end processors attached to nearly 60,000 control points and coordinated by higher-level supervisory subsystems in the main control room. An Ada95 based shot automation framework has been developed and deployed during the past year to orchestrate and automate shots performed at the NIF using the ICCS. Ada95 provides language features that have enabled the construction of a complex, highly distributed control application. The Ada tasking model and protected types provide multi-threaded behaviors and synchronization mechanisms required by the ICCS. The Shot Automation framework is designed to automate 4-hour shot sequences including derivation of shot goals from an experiment definition, set up of the laser and diagnostics, automatic alignment of laser beams, and a countdown to charge and fire the lasers. A typical sequence consists of one or more preparatory verification shots leading to an amplified system shot that is followed by post-shot analysis and archiving. The framework provides for flexible model-based workflow execution with work divided into data driven packets called "macro steps". The Shot Director program is the top-level orchestrating component of the shot automation framework, which manages the state machine that defines the structure of the shot sequence. Collaboration Supervisors translate shot lifecycle state commands from the Shot Director into sequences of "macro steps" that are distributed to subsystem shot supervisors while maintaining the order of macro steps for each subsystem and supporting collaboration between different macro steps. Each macro step has phases for database-driven verification and a scripted execution. This provides a highly flexible framework for performing a variety of NIF shot types. Database tables define the order of work and dependencies (workflow) of macro steps to be performed for a shot. A graphical model editor facilitates the definition and viewing of an execution model. A Change Manager tool enables "de-participation" of individual devices, of entire laser segments (beams, quads, or bundles of beams) or of individual diagnostics. This software has been successfully deployed to the NIF facility and is currently being used to support NIF laser commissioning shots and the build-out of additional laser bundles. The software will also be used to automate future target and experimental shot campaigns.

Analysis for Load-Balancing Techniques in Intelligent Network Systems

The main idea behind the intelligent networks (IN) concept is to place the intelligence/service logic in the dedicated server and to meet various service requirements of subscribers and the development of new services in time. The ( N + 1)-type architecture of the IN service system consists of front-end processors (FEP) and back-end processors (BEP), which are currently in service by a telecommunications company in Korea. The main characteristic of this architecture is a separation between the data plane and control plane of the network, such that the load-balancing table is used to distribute the traffic load among the processors, FEP and BEP, taking into account services. Since the values used in the load-balancing table affect the IN system performance significantly, this paper simulates the IN system and evaluates system performance depending on the values used in the table.

The national ignition facility

The National Ignition Facility (NIF) currently under construction at the University of California Lawrence Livermore National Laboratory (LLNL) is a 192-beam, 1.8-Megajoule, 500-Terawatt laser being built by the Department of Energy and the National Nuclear Security Agency (NNSA) for inertial confinement fusion and high-energy-density experimental studies. The stadium-sized facility is currently activating systems for first light and will be completed in 2008.The facility is controlled by the Integrated Computer Control System (ICCS), a layered architecture of 300 front-end processors (FEP) coordinated by supervisor subsystems. The FEP's are distributed computers that interface to physical devices through VME-bus and PCI-bus crates. The functional subsystems - beam control including automatic beam alignment and wavefront correction, laser pulse generation and pre-amplification, diagnostics, pulse power, and shot timing - implement the actions of operators at eight graphic consoles, coordinate control, and display and archive data in a database. The software architecture mimics the hardware design levels: software devices in FEP's model hardware control points and supervisory objects model the line replaceable units that modularize the laser system. Graphic user interfaces are provided to make status and control of each level accessible to operators.The ICCS software is based on an object-oriented framework that incorporates services for archiving, machine configuration, graphical user interface, status monitoring, event logging, scripting, alert management, and access control. Software code development uses a mixed language environment of Ada (for functional controls) and Java (for user interface and database backend). CORBA is used to communicate between languages and processors. Substantial benefits credited to using Ada include the formality of controlled interfaces that rely on Ada's type model, easy-to-construct exception processing and the robustness of Ada's tasks.A strategy of incremental cycles of construction and formal test has been used since project inception. The project has completed more than 30 planned cycles of deployment into testbeds and is now integrating with the first 4 operational beamlines in the facility. Fifty of the planned 300 FEP's have been installed and tested with facility equipment. These implement nearly 200 classes that model physical control hardware - some 2500 software objects. Nearly all of the top-level functional subsystems, embodying some 110 application classes, have been commissioned in the facility. The integrated control system has successfully executed shots into test diagnostics in support of laser integration. The first coordinated facility shots to the 10-m diameter target chamber are expected in early 2003.Issues of robustness and scaling arise as the system integrates larger ensembles of control points and serves an increasing number of operators. The system comprises some 60 intercommunicating processes, and since none of these are known to be defect free, techniques for replacement and restart of individual processes are required. The most common communication pattern - publish and subscribe - is supported by a connection management framework that adds exception handlers to the ORB in order to restore broken connections and restart failed processes without explicit action by application client codes.This work was performed under the auspices of the U.S. Department of Energy by the University of California, Lawrence Livermore National Laboratory under Contract No. W-7405-Eng-48.

The data-acquisition system for the CMS tracker beam tests

This paper describes the conception and the development of a real-time data-acquisition system for prototype detectors of the being designed for the compact muon solenoid (CMS) experiment at the Large Hadron Collider of CERN, European Laboratory for Particle Physics, Geneva, Switzerland. The rationale for the development of a dedicated data-acquisition system was the need to perform two fundamental beam tests (the Barrel 1 and Forward 1), with large-scale prototypes of the detectors planned as the baseline design. The number of readout channels, the complexity of the readout electronics, and the stringent requirements of the milestone tests mandated that a thorough understanding of the issues related to the physics of the detectors themselves be coupled with the application of leading-edge electronic and software engineering technologies. The implementation described in this paper is based on a distributed architecture. An event builder CPU handles the two main tasks of synchronizing a variable number of front-end processors and formatting the data in preparation for the transfer to a dedicated high-performance storage system, while the front-end processors handle the hardware and the real-time readout. Additional workstations are used to decouple the actual task of transferring the data files and monitoring the detector performance on-line from the readout farm. The system has been successfully operated during the two aforementioned Milestone tests, allowing the CMS collaboration to pass them, with the simultaneous readout of up to 40000 detector channels. The results of the two Milestones have led to the compilation of the Tracker Technical Design Report. Subsequently, the same readout system has been used for a number of other beam tests, and it has formed the basis for the development of further, more advanced data-acquisition systems for the new readout electronic of the CMS Tracker.

CDFVME-software framework for testing VME boards

New VME based boards are being produced for the Run II of the Collider Detector at Fermilab (CDF). These boards are being developed and tested at both Fermilab and offsite institutions. A software framework called CDFVME has been developed in which DAQ code can be easily written to control such boards in a test stand. The framework has been used to perform diagnostics at single board, multi-board, and multi-crate levels. This software framework runs on Unix, Linux and Windows NT platforms with a Java GUI communicating via LAN to multiple intelligent front end VME crates. All distributed processes are managed by a custom CORBA based software. The system has been ported to Motorola 68 K and PPC front end processors running the VxWorks real-time kernel.

A neural network based estimator for electricity spot-pricing with particular reference to weekend and public holidays

There is a growing interest in the prediction of system marginal price (SMP) in the power pool since electricity industry vesting in England and Wales in 1990. The prediction of SMP improves the financial performance of an independent power producer bidding in the day-ahead market. This paper proposes a neural-network-based approach to predict system marginal price (SMP) with particular reference to weekend and public holidays. The SMP estimator proposed in this paper consists of two phases. One is a front-end processor. The second is a neural-network-based predictor. The raw data are processed by front-end processors which represent the features of SMP on Saturday, Sunday and public holidays, respectively. The output from the front-end-processor forms a neural-network input file. The network is trained by the error backpropagated algorithm. The results show that mean absolute percentage errors of the estimation are 9.40% on Saturday, 8.93% on Sunday and 12.19% on a public holiday, respectively. The program is coded in visual C++ and runs on a PC with friendly windows interface.

Modularity versus efficiency in OSI system implementations

Modularity and efficiency, in protocol implementations, do not go hand in hand. A tradeoff is inevitable between good structure and good performance. Although a layered architecture has been proposed for open communications, it is generally unknown as to how the layer entities ought to be structured and what the appropriate strategies are that could be adopted in implementations. This paper examines a number of possible implementation strategies based on the various prevailing system facilities: processes, threads, kernel and front-end processors. The pros and cons of the strategies are analysed showing their relative merits in implementations. The analysis leads to the recommendation of an enhanced subsystem architecture that holds the best compromise of the conflicting requirements of modularity and efficiency.

Optimal divisible job load sharing for bus networks

Optimal load allocation for load sharing a divisible job over N processors interconnected in bus-oriented network is considered. The processors are equipped with front-end processors. It is analytically proved, for the first time, that a minimal solution time is achieved when the computation by each processor finishes at the same time. Closed form solutions for the minimum finish time and the optimal data allocation for each processor are also obtained.

MLSE for an unknown channel. II. Tracking performance

For pt. I see ibid., vol.44, no.7, p.836, 1996. The channel-tracking mode performance of the front-end (FE) processors is determined through approximate analysis and computer simulation. A general model for the FE processing, which encompasses these representative FEs, is assumed. Previous analysis techniques are extended to include the effects of FE processing. The specific system analyzed consists of a Rayleigh-fading, diffuse multipath channel, with several data pulse shapes considered. An adaptive maximum likelihood sequence estimation (MLSE) algorithm based on the per-survivor processing (PSP) technique is analyzed and compared to an algorithm based on correct symbol feedback. The results show that significant performance degradation is suffered when suboptimal FE processing is used. Some of the issues which may complicate matters in practice are examined with an emphasis on application in packet mobile radio systems. The limitations of the results and the models used are discussed.

A rule-based phrase parser for real-time text-to-speech synthesis

AbstractText-to-speech systems are currently designed to work on complete sentences and paragraphs, thereby allowing front end processors access to large amounts of linguistic context. Problems with this design arise when applications require text to be synthesized in near real time, as it is being typed. How does the system decide which incoming words should be collected and synthesized as a group when prior and subsequent word groups are unknown? We describe a rule-based parser that uses a three cell buffer and phrasing rules to identify break points for incoming text. Words up to the break point are synthesized as new text is moved into the buffer; no hierarchical structure is built beyond the lexical level. The parser was developed for use in a system that synthesizes written telecommunications by Deaf and hard of hearing people. These are texts written entirely in upper case, with little or no punctuation, and using a nonstandard variety of English (e.g.WHEN DO I WILL CALL BACK YOU). The parser performed well in a three month field trial utilizing tens of thousands of texts. Laboratory tests indicate that the parser exhibited a low error rate when compared with a human reader.

Distributed computation in linear networks: closed-form solutions

A linear network of communicating processors is analyzed. The processors in the network may or may not be equipped with front-end processors. The processing load originates either at the boundary or at the interior of the network. Closed-form solutions and computational techniques are presented for the above situations, to obtain time optimal distribution of processing loads on the processors. Some important results are proved analytically using the closed-form expressions. >

Closed form solutions for bus and tree networks of processors load sharing a divisible job

Optimal load allocation for load sharing a divisible job over processors interconnected in either a bus or a tree network is considered. The processors are either equipped with front-end processors or not so equipped. Closed form solutions for the minimum finish time and the optimal data allocation for each processor are obtained. The performance of large symmetric tree networks is examined by aggregating the component links and processors into a single equivalent processor. This allows an easy examination of large tree networks. In addition, it becomes possible to find a closed form solution for the optimal amount of data that is to be assigned to each processor in the tree network in order to achieve the minimum finish time.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Real-time data processes in a balloon-borne apparatus

In a balloon-borne experiment signals from detectors are processed in a balloon-borne apparatus to fit the data bandwidth of storage devices installed in the apparatus. The process begins at front-end processors receiving signals from detectors and proceeds under successive stages by dedicated processors. Event data are finally fed to a computer bank where the events are further selected before being written into the storage devices. The computer bank is made of transputer modules. The transputers are widely employed and some processors are implemented with the transputers. The integration of the processors is described and the function and implementation of the individual processors are outlined. The authors are planning a series of balloon-borne experiments searching for antiparticles of cosmic origins. Firstly the authors will measure the antiproton flux. Antiprotons need to be identified under the overwhelming background of protons. >

HECODES: a framework for HEterogeneous COoperative distributed expert systems

Cooperation between human experts is important when solving many complex problems and communication between the experts is essential for such cooperation. Cooperation between computerised expert systems is equally essential when dealing with complex problems. The software system described in this paper, HECODES (HEterogeneos COoperative Distributed Expert Systems), is a framework used to realize cooperation between distributed heterogeneous expert systems (ESs). HECODES consists of a centralized control system, a blackboard system, and front-end processors. All parts of HECODES are specified (and implemented) using knowledge-based programming techniques and so they are domain-independent. HECODES can integrate a variety of pre-existing ESs using a variety of kinds of cooperation. The ESs can be heterogeneous with respect to control strategy, knowledge representation, and programming language. A prototype has been implemented using Common Lisp on a network of DEC VAX machines some of which are 60 miles apart. This paper presents the architecture of HECODES and describes the approach it takes to the provision of cooperation, knowledge-based scheduling, domain-independent communication, deadlock treatment, homomorphic transformation of heterogeneous inexact reasoning models, and the synthesis of solutions.

Distributed computation for a tree network with communication delays

Tree networks of communicating processors are examined with the objective of solving a computational problem in a minimal amount of time. The processors in the networks may be equipped either with or without front-end processors for communicating of loading. The determination of the optimal division of processing load is discussed for the network with and the network without front-end processors. The inclusion of solution time, the time taken for sensors to report the solution back to originator, is discussed.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Performance of MAP in the remote operation of a CNC

A major issue for the users of the Manufacturing Automation Protocol (MAP) is its performance in servicing time-critical applications. The author describes an experiment in which a computerized numerical controller (CNC) is remotely operated using MAP. A set of critical timing parameters is defined, a performance analysis of the system from the user's point of view is carried out, and a comparison with other similar systems is made. The transfer time between two front-end processors, implementing the seven MAP layers, for a 100-byte-long message is found to be about 18 ms, and the transfer time for the same message between two user processes on the hosts is about 40 ms. >