Search

The ATLAS Trigger and Data Acquisition System has been designed to use more than 2000 CPUs. During the current development stage it is crucial to test the system on a number of CPUs of similar scale. A dedicated farm of this size is difficult to find, and can only be made available for short periods. On the other hand many large farms have become available recently as part of computing grids, leading to the idea of using them to test the TDAQ system of ATLAS. However the task of adapting the TDAQ system to run on the Grid is not trivial, as it requires full access to the computing resources it runs on and real-time interaction. Moreover the Grid virtualises the resources to present a common interface to the user. We will describe the implementation and first tests of a scheme that resolves these issues using a pilot job mechanism. The Tier2 cluster in Manchester was successfully used to run a full TDAQ system on 400 nodes using this implementation. Apart from the tests described above, this scheme also has great potential for other applications, like running Grid remote farms to perform detector calibration and monitoring in real-time, and automatic nightly testing of the TDAQ system. Index Terms--ATLAS, data acquisition, grid computing, nuclear physics, real time systems, scaling, software testing, trigger.

A system is presented for use in Nuclear Physics Data Acquisition. The system is hosted on a Personal Computer's ISA bus, and is based on the TMS320C31 digital signal processor (DSP), a fast floating point DSP that simultaneously handles several different tasks in this system, by means of an interrupt-driven architecture. The system includes three different data acquisition modules: a multi-channel analyzer, a multi-channel scaler, and a signal analyzer. Using the interrupt-driven architecture, the DSP pre-processes the data coming from the three different data acquisition modules, and accepts data requests from the host PC. On the PC side, the system runs under Microsoft Windows 95 or Windows NT, and it implements the client-server model. While the data acquisition board performs its chores independently, the host PC sends it several commands on behalf of an arbitrary number of clients. Those clients each represent a window over the data being acquired, allowing different users to take different views of the data. Each client can run on the same host computer or on any other computer, provided there is a computer network (the Internet is a possibility, as long as there is enough bandwidth) connecting it to the host. This allows the experimentalist to remotely examine the experimental results from his office without having to go to the lab. On the other hand, different experimentalists can have different views of the data being acquired. All these are benefits stemming from the client-server model.

A multichannel TDC subsystem was designed and built at the Institute for Nuclear Physics at Mainz University. The system is based on a TDC ASIC chip called TDC 2001 and consists of several TDC modules housed in a VME-type crate. The TDC modules are controlled and read out by a transputer. The transputer can deliver the data through the transputer links to VME-systems, workstations, or PCs as data acquisition master. The subsystem, as it was developed for the A1 experiment at the MAMI accelerator in Mainz, is described, and first results from the operation at the experiment are given.

A modular system to determine correlations among many detectors in radiation measurements is described. It is based on PC computer and supports several types of CAMAC crate controllers. Besides standard techniques and procedures such as data taking, event building, event sorting, also sophisticated data visualization, processing, and a new efficient method of multidimensional data analysis with on-line compression have been implemented in the system.

The new general purpose data acquisition system developed at GSI is currently installed at about 30 experiments at GSI and other sites. It is based on the LynxOS operating system [1]. Several CPUs, i.e. GSI developed CAMAC computer boards, VME processor boards, and Aleph Event Builders, are connected by a memory mapped bus, i.e. a VSB or VME bus. It can be configured easily for various hardware setups and is highly optimized for fast trigger rates and data throughput. The first and successful experiment taking data with the system was the SHIP production run for elements 110 and 111.

A field portable data acquisition system has been developed around a VMEbus based micro-processor and standard TCP/IP network protocols. The hardware consists of a compact VME crate and a single CAMAC crate containing the signal processing electronics. The component processes of the data acquisition system transfer control and event data over a set of TCP/IP socket connections. The use of network sockets for the interprocess communications allows the data acquisition system to be operated transparently on one workstation or on a number of workstations distributed around a local network.

The capacity of data acquisition systems will improve by an order of magnitude with the advent of the new high-bandwidth network technology. Data acquisition systems will really be scalable. To be in the front line of the application of this new technology, we participate in the Dutch national asynchronous transfer mode (ATM) pilot project SURFnet4. In this framework we have initiated a project: 'Remote data acquisition and analysis'. The aim is to have an ATM network between remote VME front-ends and controlling and monitoring workstations.

A new data acquisition system has been developed at the Hahn-Meitner-Institut to handle almost 550 parameters of nuclear physics experiments. The system combines a UNIX host running a portable data buffer router [1] and a VME frontend based on the OS9 real time operating system. Different kinds of pulse analyzers are located in several CAMAC crates which are controlled by the VME system via a VICbus connection. Data readout is performed by means of an ECL daisy chain. Besides controlling CAMAC the main purpose of the VME front-end is event data formatting and histogramming. Using TCP/IP services, the UNIX host receives formatted data packages for data storage and display. During a beam time at the antiproton accelerator LEAR/CERN, the PS208 experiment has accumulated about 100 Gbyte of event data [2].

Asynchronous Transfer Mode (ATM) is a candidate technology to implement the high performance network in the data collection system for the ATLAS experiment. This work presents the results of modelling and simulation studies which aim at integrating the detailed organization of the detector read-out, the trigger requirements and the capabilities of ATM switching networks. The status of hardware development of small scale demonstrators is outlined.

Event data flow to switch-type event builder has a coherent traffic shape. Random data traffic has been intensively studied for telecommunication data. However, coherent traffic has not been studied yet. We have analyzed the stability of coherent data traffic. The study is indispensable to use a switching network for an event builder.

Software parallelisation allows an efficient use of available computing power to increase the performance of applications. In a case study we have investigated the parallelisation of high energy physics event reconstruction software in terms of costs (effort, computing resource requirements), benefits (performance increase) and the feasibility of a systematic parallelisation approach. Guidelines facilitating a parallel implementation are proposed for future software development.

DART is the high speed, Unix based data acquisition system being developed by Fermilab in collaboration with seven High Energy Physics Experiments [1 - 6!. This paper describes DART run control, which has been developed over the past year and is a flexible, distributed, extensible system for the control and monitoring of the data acquisition systems. We discuss the unique and interesting concepts of the run control and some of our experiences in developing it. We also give a brief update and status of the whole DART system.

This paper describes a real-time response of UNIDAQ[1,2]. Interrupt task response time and task switching (context switching) time are measured for real-time UNIX (LynxOS) systems and standard UNIX systems. Here we show the correlation between the context switching time and the values of a benchmark program, SPECint92. We also show distributions of the response time for various operating systems (OS). A performance evaluation tool, DAQBENCH[3], was used for the measurements.

A new data acquisition system has been designed for physics experiments scheduled to run at CEBAF starting in the summer of 1994. This system runs on Unix workstations connected via ethernet, FDDI, or other network hardware to multiple intelligent front end crates - VME, CAMAC or FASTBUS. CAMAC crates may either contain intelligent processors, or may be interfaced to VME. The system is modular and scalable, from a single front end crate and one workstation linked by ethernet, to as many as 32 clusters of front end crates ultimately connected via a high speed network to a set of analysis workstations. The system includes an extensible, device independent slow controls package with drivers for CAMAC, VME, and high voltage crates, as well as a link to CEBAF accelerator controls. All distributed processes are managed by standard remote procedure calls propagating change-of-state requests, or reading and writing program variables. Custom components may be easily integrated. The system is portable to any front end processor running the VxWorks real-time kernel, and to most workstations supplying a few standard facilities such as rsh and X-windows, and Motif and socket libraries. Sample implementations exist for 2 Unix workstation families connected via ethernet or FDDI to VME (with interfaces to FASTBUS or CAMAC), and via ethernet to FASTBUS or CAMAC.

Experiments at Fermilab require an ongoing program of development for high speed, distributed data acquisition systems. The physics program at the lab has recently started the operation of a Fixed Target run in which experiments are running the DART[1] data acquisition system. The CDF and D?? experiments are preparing for the start of the next Collider run in mid 2000. Each will read out on the order of 1 million detector channels. In parallel, future experiments such as BTeV R&D and Minos have already started prototype and test beam work. BTeV in particular has challenging data acquisition system requirements with an input rate of 1500 Gbytes/sec into Level 1 buffers and a logging rate of 200 Mbytes/sec. This paper will present a general overview of these data acquisition systems on three fronts -- those currently in use, those to be deployed for the Collider Run in 2000, and those proposed for future experiments. It will primarily focus on the CDF and D?? architectures and tools.