Trends in Test part 3: Peer-to-Peer Computing
15 July 2010
In this five-part series, National Instruments draws upon its knowledge and experience of a wide range of industries to identify key technologies and methodologies impacting test and measurement in 2010. This third instalment focuses on peer-to-peer computing and the benefits it offers the test industry.
For some time now, the computing industry has implemented distributed architectures by dividing computing among multiple processing nodes. Google are a rather grand example of this, as their search queries are not run on a single, centralised supercomputer, but on a network of more than 450,000 interconnected servers, with a combined RAM of around two Petabytes. A smaller, perhaps more comprehensible, example of distributed computing is the use of graphics processing units (GPU). Most modern personal computers house some form of GPU, which incorporates specialised processors to offload graphics rendering from the microprocessor.
Peer-to-peer, commonly abbreviated to P2P, is a type of computing that uses a decentralised architecture to distribute a portion of its resources, such as processing power or disk storage, among multiple processing nodes. In contrast to traditional systems, P2P does not require a central coordination hub to transfer data and managing processing, as each member of the network has the same capabilities and can establish a communication sessions with other network members.
How can the automated test industry benefit from this computing architecture? The ever-increasing intricacy of modern electronic devices results in increasingly convoluted requirements to test them. Furthermore, with dizzyingly fast sampling rates and high channel counts, the volume of data acquired by modern test systems is increasing at exponential rates. We are entering the petabyte-age, where high-end instruments will be processing millions of gigabytes of data per day. If automated test systems are to counteract this added complexity, they need to evolve to work smarter and not just harder.
Software-defined instrumentation is giving engineers unprecedented control over their automated test systems and facilitating all manner of novel applications. This is partially due to the easy access to raw measurement data, which can be analysed and processed according to exact application needs. Software-defined ATE can benefit from Peer-to-peer, which, for example, may take the form of acquiring raw data directly from an instrument like an oscilloscope or digitiser and streaming it to a field-programmable gate array (FPGA) for advanced inline signal processing. The pre-processed data may then be transferred to a controller for analysis or storage. Thus, devices in a system are able to share information without burdening other system resources.
New, high-performance, distributed architectures are required to transfer and process all of this data. This provides strict, challenging requirements. The sheer quantity of data requiring transfer means that high throughput is a requisite. Low latency is another key requirement, as data often requires processing within fractions of a second of being acquired. Furthermore, the processing nodes must be user-programmable, so that analysis and processing can meet the user’s exact test requirements.
Very few distributed architectures are able to meet the uncompromising criteria of peer-to-peer. Ethernet, for example, provides an effective point-to-point topology with a diverse set of processing nodes, but it has high latency and only moderate throughput. The communications platform that has shown the most promise for future, high-performance P2P architectures is PCI Express. PCI Express, abbreviated to PCIe, provides transfer rates of up to 16 gigabytes per second, with latencies of less than a microsecond.
PXIe (PCIe eXtensions for Instrumentation) builds on the standard PCIe technology by integrating synchronisation buses and several key software features. The PXIe platform allows triggers and clock sources to be readily shared amongst multiple instruments, making it an ideal solution for test and measurement, data acquisition, and manufacturing applications.
PXI Express is already seeing use as a distributed architecture in military and aerospace applications. While defining its next-generation test systems, the U.S. Department of Defence Synthetic Instrument Working Group identified PCI Express as the only bus capable of providing the data throughput and latency required for user-customised instrumentation. This architecture is now seen in BAE Systems’ synthetic instruments that use PCI Express to stream downconverted RF data directly to separate processing modules for inline signal processing.
Senior Technology Fellow at BAE Systems, Wade Lowdermilk, recently commented that “next-generation synthetic instruments will require high-performance signal processing while maintaining user configurability. Using peer-to-peer streaming over PXI Express, we can send acquired RF signals directly to PXI Express FPGA processing modules to realise 10-times speed improvements over previous solutions.”
National Instruments peer-to-peer streaming technology uses PXI Express to enable direct, point-to-point transfers between multiple instruments. The chassis backplane switches provide direct links to the slots occupied by modules, meaning that you do not need to transfer data through the host controller or into system resources such as the CPU or host memory.
For example, an NI PXIe-5622 digitiser can use peer-to-peer data streaming to send data directly to an NI PXIe-7965R FlexRIO FPGA module for inline processing in real-time. Potentially, for advanced processing applications, the FPGA module can then send data to other FPGA modules for additional analysis.
Looking beyond the physical hardware architecture of this kind of communication model, peer-to-peer computing will change how engineers configure and program their test systems. With so many disparate processing nodes, test developers require new tools to visualise and direct the flow of data. Furthermore, software development must become more intuitive, abstracting the complexity of advanced processing nodes, such as FPGAs and GPUs, away from the developer. Application development environments like NI LabVIEW can meet these needs.
Peer-to-peer computing is early in its life-cycle and there are still many innovations to come. With ever-increasing volumes of data and progressively complicated test requirements, engineers will need to learn how to best leverage these new technologies to develop smarter, faster, better test systems.
By Richard Roberts – Technical Marketing Engineer, National Instruments UK & Ireland
Next instalment: Trends in Test Part 4 – Embedded Design and Test
Contact Details and Archive...
Related Articles...
Most Viewed Articles...