Explore chapters and articles related to this topic
Task Management and Timekeeping, POSIX API
Published in Gedare Bloom, Joel Sherrill, Tingting Hu, Ivan Cibrario Bertolotti, Real-Time Systems Development with RTEMS and Multicore Processors, 2020
Gedare Bloom, Joel Sherrill, Tingting Hu, Ivan Cibrario Bertolotti
The function returns zero when successful. When it fails, it shall return one of these non-zero error codes: is returned when one of the arguments to is invalid, namely: The given is unknown to the system.The contents of the data structure pointed by are invalid or, for an absolute wait, are out of range for the .The specifies the CPU time clock of the calling thread. The clock does not support . Among the clocks explicitly defined by the standard, only and do, whereas CPU time clocks do not. The function returned prematurely because it was interrupted by a signal. The return takes place after the execution of the corresponding signal handler.
Heuristic and metaheuristic algorithms
Published in Dušan Teodorović, Miloš Nikolić, Quantitative Methods in Transportation, 2020
Dušan Teodorović, Miloš Nikolić
In many cases, optimal solution of the problem cannot be achieved within an acceptable central processing unit (CPU) time. Often, there is a combinatorial explosion of the promising combinations of the decision variables that could be optimal; for instance, in a problem, if we have considered 1,000 binary variables that can take values 0 or 1, the total number of all possible solutions is equal to 21000. In some other situations, it is hard to estimate defined objective function values. In order to conquer these difficulties, various heuristic (`ευρισκω, Greek for “I find”) algorithms were proposed over the past five decades. Many of the heuristic algorithms developed have been able to generate sufficiently good solution(s) in acceptable amounts of CPU time. On the other hand, heuristic algorithms that depend primarily on the experience and/or judgment of the analyst frequently do not produce the optimal solution.
Synthetic Aperture Radar Techniques for Through-the-Wall Imaging
Published in Moeness G. Amin, Through-the-Wall Radar Imaging, 2017
In order to create ultra-wideband (UWB) SAR images of targets, we must obtain the radar signature of that target over a large number of frequencies and aspect angles. Most of the examples in this chapter are based on radar signature data generated through computer simulations. Several computational electromagnetic (CEM) codes have been used specifically for modeling scenarios of interest in TWRI. AFDTD [7], developed by ARL, and XFDTD [8], developed by Remcom, are both based on the finite-difference time-domain (FDTD) method [9]. Large models of a building were reported by the Lawrence Livermore National Laboratory via a finite-element time-domain (FETD) algorithm [10]. Other codes, based on ray-tracing and other high-frequency methods, include Xpatch [11], developed by Science Applications International Corporation (SAIC) under a grant from the U.S. Air Force, and NEC-BSC [12] developed at the Ohio State University. In general, the exact CEM techniques (such as FDTD or FETD) are very computationally intensive both in terms of central processing unit (CPU) time and memory. This issue makes approximate, high-frequency methods attractive approaches for rapid modeling of very complex scenarios. However, before one applies these methods to the specific problem of TWRI, one must validate them against exact numerical solutions for that class of problems. The numerical examples described in this chapter are based on the AFDTD and Xpatch codes. A direct validation of these simulation methods for the specific TWRI environment was performed in [13–16].
Comparison of Erlang/OTP and JADE implementations for standby redundancy in a holonic controller
Published in International Journal of Computer Integrated Manufacturing, 2019
G. T. Hawkridge, A. H. Basson, K. Kruger
For this paper, the computational resource requirements of each implementation are quantified using three metrics: CPU time, peak RAM usage and thread count. CPU time is a measure of the cumulative CPU usage by the application’s threads across all the device’s logical cores. It gives an indication of processing requirements of each implementation and provides an estimate of the tier of device required to execute them. Peak RAM usage is an indicator of each implementation’s memory requirements. Peak RAM usage is a consideration particularly for embedded devices, since RAM is a scarce resource. If usage exceeds the available RAM, then paging may be used, but this is usually undesirable due to the low speed of secondary storage for these devices. Ideally, the minimum and average RAM usage would also be profiled, but due to the complication of testing on multiple heterogeneous distributed devices, this could not be achieved.
Evaluation criteria for holonic control implementations in manufacturing systems
Published in International Journal of Computer Integrated Manufacturing, 2019
To assess the computational resource requirements, two measures are used: CPU time – as an indicator of CPU usage, CPU time is the measurement of the combined time, over all available cores, that the CPU executes instructions for the holonic control implementation (Microsoft TechNet, 2018a). CPU time can be measured by the operating system and in Windows is available in the processes window of the Task Manager application. The CPU time should be recorded at the start of production (thus the CPU time involved with system start-up is excluded) and again when production of the last product is completed.Memory usage – the random access memory (RAM) consumed by each implementation is monitored during simulated production. Windows includes the Performance Monitor application, which allows the user to record many of the counters exposed by the operating system. There exist counters for every active process on the PC. The Private Working Set counter measures the RAM (in bytes) that is consumed by a single process (Microsoft TechNet, 2018b) – this counter should be recorded for the duration of a production run.
Assignment of duplicate storage locations in distribution centres to minimise walking distance in order picking
Published in International Journal of Production Research, 2021
Wei Jiang, Jiyin Liu, Yun Dong, Li Wang
Table 3 gives the CPU computation time results of different algorithms for the medium and large-scale problems. Here, the CPU time is measured in seconds. From the results, we can see that though we set the parameters to expect the same computation time for all algorithms to solve the same problem group, their actual computation times for a group are slightly different. Nevertheless, the differences are not significant and the computing times of all the algorithms increase with problem scale in a similar pattern. The computation time for the problem group with 500 products and 1000 locations is about 10 minutes, indicating that these algorithms can solve larger scale problems.