Data-parallel environments

For data-parallel environments (SIMD programming model), performance can be represented using standard profiling techniques that associate fraction of overall execution time with particular pieces of code. As noted earlier, this programming model has limited utility for the applications we address, and we do not dwell on performance evaluation. 

In this paper we propose a novel preexecution model for data parallel applications [11]. We present a purely software-based adaptive preexecution approach. It is lightweight, easy to implement, and requires no special hardware other than a multicore CPU. With run-time parameter tuning the algorithm configures itself for maximum performance and adapts itself to the executirai environment and the memory access characteristics of the application. We apply this method in a real-life data-intensive application of data coding and show that it achieves an average speedup of 10-30%. 

A nanostore is a single-chip computer that includes 3-D stacked layers of dense silicon non-volatile memory with a layer of compute cores and a network interface. A large number of individiral nanostores can cottununicate over a simple interconnect and run a data-parallel execution environment like MapReduce to support large-scale distributed data-centric workloads. The two most impor-... 

Figure 4.15. Atomically resolved TEM images of a Cu/ZnO model catalyst in various gas environments together with the corresponding Wulff construction of the Cu particle (a,b) Cu nanocrystal faceted by (100), (110) and (111) surfaces the TEM image was recorded at 1.5 mbar of H2 at 220 °C with the electron beam parallel to the [Oil] zone-axis of copper. The insert shows EELS data at the Cu L2,3-edge...

The connections within a "network" that consists of just one node can only link it with its environment, but as soon as the number of nodes in a network is increased, choices exist in the way that the nodes are connected. Two nodes could be arranged in parallel, so that both accept input data and both provide output (Figure 2.18). Alternatively, nodes could be placed in a serial configuration, so that the output of one node becomes the input of the second (Figure 2.19). 

In parallel with improvements in chemical sensor performance, analytical science has also seen tremendous advances in the development of compact, portable analytical instruments. For example, lab-on-a-chip (LOAC) devices enable complex bench processes (sampling, reagent addition, temperature control, analysis of reaction products) to be incorporated into a compact, device format that can provide reliable analytical information within a controlled internal environment. LOAC devices typically incorporate pumps, valves, micromachined flow manifolds, reagents, sampling system, electronics and data processing, and communications. Clearly, they are much more complex than the simple chemo-sensor described above. In fact, chemosensors can be incorporated into LOAC devices as a selective sensor, which enables the sensor to be contained within the protective internal environment. Figure 5... 

Mench et al. developed a technique to embed microthermocouples in a multilayered membrane of an operating PEM fuel cell so that the membrane temperature can be measured in situ. These microthermocouples can be embedded inside two thin layers of the membrane without causing delamination or leakage. An array of up to 10 thermocouples can be instrumented into a single membrane for temperature distribution measurements. Figure 32 shows the deviation of the membrane temperature in an operating fuel cell from its open-circuit state as a function of the current density. This new data in conjunction with a parallel modeling effort of Ju et al. helped to probe the thermal environment of PEM fuel cells. 

A spectrum of metal compound reactivities in petroleum could arise for several reasons. Nickel and vanadium exist in a diversity of chemical environments. These can be categorized into porphyrinic and non-porphyrinic species vanadyl and nonvanadyl or associated with large asphaltenic groups and small, isolated metal-containing molecules. Each can be characterized by unique intrinsic reactivity. Reaction inhibition which occurs between the asphaltenes and the nonasphaltenes, as well as between Ni and V species, can also contribute to reactivity distributions. The parallel reaction interpretation of the observed reaction order discrepancy is therefore compatible with the multicomponent nature of petroleum. Data obtained at low conversion could appear as first order and only at higher conversions would higher-order effects become obvious. The... 

Unfortunately, most of these tools are extremely expensive, and are fairly complex to deploy, requiring a database backend for alert storage. Surprisingly, they also have limited correlation capabilities, only providing a dozen or so rules as example for the development of environment-specific correlation rules. They should be viewed as a development framework for writing correlation rules. Since our correlation needs are very diverse, we could not find a platform that would allow us to run multiple correlation processes in parallel, from dynamic statistical analysis to vulnerability assessment. Also, manipulation of contextual data with interfaces to the inventory and configuration databases of the companies, was a strong requirement that no commercial tool satisfied at the time we launched the project. 

In this chapter we will discuss current approaches for analytical characterization of combinatorial libraries in a pharmaceutical industry environment. Recently, several analytical groups have presented very similar strategies for analysis of libraries [7-9]. As will be shown later, the key to successful analytical characterization of a combinatorial library is to perform analytical and chemical work in parallel with the library development. The accumulation of data and analytical experience during this process results in an assessment of library quality with a high level of confidence, even if as little as 5-10% of the library components are analyzed. Utilization of the strategy will be demonstrated using two examples analysis of a library synthesized on a robotic station in spatially addressed format and analysis of a library synthesized in accordance with split-and-mix technology. 

---