|
|
Installation and configuration of Octave computation cluster
Mikulka, Zdeněk ; Hasmanda, Martin (referee) ; Sysel, Petr (advisor)
This diploma thesis contains detailed design of high-performance cluster, primarely focused for parallel computing in Octave application. Each of component of this cluster is described along with instructions for installation and configuration. Cluster is based on GNU/Linux operating system and Message Parsing Interface. Design alllows implementation of this cluster in computers of schoolroom with active lessons.
|
|
|
Dynamic Load-Balancing in Parallel Applications
Dvořáček, Vojtěch ; Nikl, Vojtěch (referee) ; Jaroš, Jiří (advisor)
This thesis aims to implement dynamic load balancing mechanism into the parallel simulation model of the heat distribution in a CPU cooler. The first part introduces theoretical foundations for dynamic load balancing, describing current solution approaches. The second part refers to the heat distribution model and related topics such as MPI communications library or HDF library for data storage. Then it proceeds to the implementation of simulation model with dynamic 2D decomposition of square model domain. Custom geometry based dynamic load balancing algorithm was introduced, which works with this decomposition. Important part of the implementation is Zoltan library, used especially for data migration. At the end, a set of experiments was presented, which demonstrates load balancing abilities of designed model together with conclusions and motivation for future research.
|
|
|
Overclocking of Modern Processors with an Emphasis on Performance, Power Consumption and Temperature
Kelečéni, Jakub ; Vaverka, Filip (referee) ; Nikl, Vojtěch (advisor)
This thesis analyzes the dependency of performance, power consumption and temperature on processor frequency. Theoretical part discusses the processor architecture, benchmarks and algorithm types. Experimental part is focused on benchmarks - matrix multiplication, Quicksort, PI number calculation, Ackermann function, LAMMPS, PMBW, Linpack. This set of benchmarks includes both single-threaded and multi-threaded algorithms. Testing consist of three different settings of processor frequency. Multi-threaded benchmarks using different number of threads. Informations regarding the power consumption of CPU and RAM were recorded during these tests. Every test logs his running time. The impact of parallelization on power consumption and runtime is also reflected. Results from the tests are shown in charts and tables. The proper configuration of CPU for each given algorithm is analyzed in conclusion.
|
|
|
Efficient Communication in Multi-GPU Systems
Špeťko, Matej ; Jaroš, Jiří (referee) ; Vaverka, Filip (advisor)
After the introduction of CUDA by Nvidia, the GPUs became devices capable of accelerating any general purpose computation. GPUs are designed as parallel processors which posses huge computation power. Modern supercomputers are often equipped with GPU accelerators. Sometimes single GPU performance is not enough for a scientific application and it needs to scale over multiple GPUs. During the computation, there is a need for the GPUs to exchange partial results. This communication represents computation overhead and it is important to research methods of the effective communication between GPUs. This means less CPU involvement, lower latency and shared system buffers. This thesis is focused on inter-node and intra-node GPU-to-GPU communication using GPUDirect technologies from Nvidia and CUDA-Aware MPI. Subsequently, k-Wave toolbox for simulating the propagation of acoustic waves is introduced. This application is accelerated by using CUDA-Aware MPI. Peer-to-peer transfer support is also integrated to k-Wave using CUDA Inter-process Communication.
|
|
| |
|
|
Possibilities of implementation parallel task in programming languages
Zejda, Václav ; Ondroušek, Vít (referee) ; Houška, Pavel (advisor)
This work deals with the problem of implementation of parallel tasks in various programming environments. First part of the work follows the basics of parallelisation. It features especially when the parallelisation is suitable, what kinds of parallelisation are used and differences between architecture of various systems. Following parts describes various techniques in developing applications with the parallelisation support. From parallelisation support for common programming languages to graphics development tools and to environments that uses graphic processing unit (GPU) in combination with central processing unit (CPU).
|
|
|
Parallel genetic algorithm
Trupl, Jan ; Kobliha, Miloš (referee) ; Jaroš, Jiří (advisor)
The thesis describes design and implementation of various evolutionary algorithms, which were enhanced to use the advantages of parallelism on the multiprocessor systems along with ability to run the computation on different machines in a computer network. The purpose of these algorithms is to find the global extreme of function of $n$ variables. In the thesis, there are demonstrated various optimization problems, and their effective solution with the help of evolutionary algorithms. There are also described interface libraries MPI(Message Passing Interface) and OpenMP, in the extent needed to understand the problematic of parallel evolutionary algorithms.
|
|
|
Efficient Communication in Multi-GPU Systems
Špeťko, Matej ; Jaroš, Jiří (referee) ; Vaverka, Filip (advisor)
After the introduction of CUDA by Nvidia, the GPUs became devices capable of accelerating any general purpose computation. GPUs are designed as parallel processors which posses huge computation power. Modern supercomputers are often equipped with GPU accelerators. Sometimes the performance or the memory capacity of a single GPU is not enough for a scientific application. The application needs to be scaled into multiple GPUs. During the computation there is need for the GPUs to exchange partial results. This communication represents computation overhead. For this reason it is important to research the methods of the effective communication between GPUs. This means less CPU involvement, lower latency, shared system buffers. Inter-node and intra-node communication is examined. The main focus is on GPUDirect technologies from Nvidia and CUDA-Aware MPI. Subsequently k-Wave toolbox for simulating the propagation of acoustic waves is introduced. This application is accelerated by using CUDA-Aware MPI.
|
|
|
Influence of Network Infrastructure on Distributed Password Cracking
Eisner, Michal ; Zobal, Lukáš (referee) ; Hranický, Radek (advisor)
Password cracking is a process used to obtain the cracking key through which we get access to encrypted data. This process normally works on the principle of the repeated try of attempts and their verification by making calculations of cryptographic algorithms. The difficulty of algorithms affects the time spent on solving of the calculations. In spite of various acceleration methods, it is often necessary to distribute the given problem among several nodes which are interconnected via the local network or the internet. The aim of this thesis is to analyze the influence of network infrastructure on the speed, the scalability, and the utilization during different attacks on cryptographical hashes. For these purposes, there was created an automatized experimental environment, which consists of distinctive topologies, scripts, and sets of testing tasks. Based on the results of the analysis, which were obtained by the usage of tools Fitcrack and Hashtopolis it was possible to observe this influence.
|
|
|
Parallelisation of Ultrasound Simulations Using Local Fourier Decomposition
Dohnal, Matěj ; Hrbáček, Radek (referee) ; Jaroš, Jiří (advisor)
This document introduces a brand new method of the 1D, 2D and 3D decomposition with the use of local Fourier basis, its implementation and comparison with the currently used global 1D domain decomposition. The new method was designed, implemented and tested primarily for future use in the simulation software called The k-Wave toolbox, but it can be applied in many other spectral methods. Compared to the global 1D domain decomposition, the Local Fourier decomposition is up to 3 times faster and more efficient thanks to lower inter-process communication, however it is a little inaccurate. The final part of the thesis discusses the limitations of the new method and also introduces best practices to use 3D Local Fourier decomposition to achieve both more speed and accuracy.
|
guest ::
