Figure - uploaded by Przemysław Rodwald
Content may be subject to copyright.
Source publication
Due to the current challenges in computer forensics and password cracking, a single GPU is no longer sufficient. Thus, distributed password cracking platforms with dozens of GPUs become a necessity in the race against criminals. In this paper, we show a multi-GPU cracking platform build on Hashcat-based open-source distributed tool Hashtopolis for...
Context in source publication
Context 1
... this work, attention is focused on the search for the optimal strategy for breaking a single password with the power of mask attack, assuming that we do not have any knowledge about the structure of the password itself (length, complexity). In Table 2 an estimated time for cracking sample masks with our Hashtopolis-based computer cluster is presented. ...Citations
... A number of scientific fields rely on high-performance computing (HPC) for running simulations and data analysis applications. For example, [1] reviews the deployment of quantum chemistry methods on HPC platforms, [2] describes flood simulation applications in the civil engineering field running on HPC architectures, and [3] considers the scheduling problem for cybersecurity applications. Scheduling plays a critical role in HPC systems, as HPC schedulers are a primary interface between users and computational resources. ...
Applications in high-performance computing (HPC) may not use all available computational resources, leaving some of them underutilized. By co-scheduling, i.e., running more than one application on the same computational node, it is possible to improve resource utilization and overall throughput. Some applications may have conflicting requirements on resources and co-scheduling may cause performance degradation, so it is important to take it into account in scheduling decisions. In this paper, we formalize the co-scheduling problem and propose multiple scheduling strategies to solve it: an optimal strategy, an online strategy and heuristic strategies. These strategies vary in terms of the optimality of the solution they produce and a priori information about the system they require. We show theoretically that the online strategy provides schedules with a competitive ratio that has a constant upper limit. This allows us to solve the co-scheduling problem using heuristic strategies that approximate this online strategy. Numerical simulations show how heuristic strategies compare to the optimal strategy for different input systems. We propose a method for measuring input parameters of the model in practice and evaluate this method on HPC benchmark applications. We show the high accuracy of the measurement method, which allows us to apply the proposed scheduling strategies in the scheduler implementation.
In this paper, a manufacturing process for a single-server permutation Flow Shop Scheduling Problem with sequence dependant, disjoint setups and makespan minimization is considered. The full problem is divided into two levels, and the lower level, aimed at finding an optimal order of setups for a given fixed order of jobs, is tackled. The mathematical model of the problem is presented along with a solution representation. Several problem properties pertaining to the problem solution space are formulated. The connection between the number of feasible solutions and the Catalan numbers is demonstrated and a Dynamic Programming-based algorithm for counting feasible solution is proposed. An elimination property is proved, which allows one to disregard up 99.99% of the solution space for instances with 10 jobs and 4 machines. A refinement procedure allowing us to improve the solution in the time required to evaluate it is shown. To illustrate how the properties can be used, two solving methods were proposed: a Mixed-Integer Linear Programming formulation and Tabu Search metaheuristic. The proposed methods were then tested in a computer experiment using a set instance based on Taillard’s benchmark; the results demonstrated their effectiveness even under a short time limit, proving that they could be used to build algorithms for the full problem.
